 Okay, thank you very much for the invitation to talk about this paper, which is part of the project that Konow and Simone have talked about. So that's the case of Brazil, and you see that's pretty much, I'm following pretty much the same structure that Simone had on her presentation, and I believe that's common in all papers presented here on this project. Okay, so one thing that's important to look at earnings in a quality in Brazil is that it declined a lot during the 2000s, right? So here basically we can see from 2003 to 2015 what happened using a household survey, and then there was a change in the household survey that became, it was annual household survey and now it is a quarter, and we can have a panel. So that's why you have this C here, which is basically say that's a continuous household. And it's because of the change in the methodology, they collect data. We see this change here between the orange and the green lines here, but you can see that what happens is that you had a decline in earnings in a quality, and then something that looks like an increase in earnings in a quality. We had a big depression here in 2015, 16, and 17, which might explain this rising inequality. You're going to see that when you show those growth incidence curves that this decline is basically driven by this pro-poor growth, which is related to or can be explained by many other things. So the literature has, about Brazil, has shown that wage or the return to education has been an important factor explaining this decline, minimum wage as well, especially during this period, wage, the way, the experience gaps also got tighter. So that's something that I did with Chico and Julian Messina, and many other gaps that were reduced during this period also explained reduction in inequality. And there's a huge literature on that. So, but there's not much about role of occupations that could explain the decline in inequality in Brazil. So what we do is basically try to understand whether this polarization effect that can be seen in many other countries also occurred in Brazil and see whether the employment structure change had any impact in the wage inequality in Brazil. And then we're going to see, we're going to contrast what happened or the importance of those employment changes when you compare to other factors. And here, as Simon did, we're going to decompose those effects into a composition effect and a structure effect. Basically, if there is a change in the proportion of jobs or occupations that were more routine intense or it was the payment to those jobs that changed it and that could explain trends in inequality in Brazil. So just a summary of our main finds. Here's that. First, I mean, when you look at occupations by using that measure that Simon described, this RTI, routine task intensity, you see that there is a strong correlation between average earnings in each occupation and the task content and routine task intensity. So it's a negative correlation. The more intense and routine tasks, the smaller the average earnings for that occupation. We also try to decompose the total inequality between two terms, one which would be between and within occupations. And you see that between jobs inequality or between occupations inequality explains about half of inequality in a given point in time in Brazil. What you have is not any employment changes or employment structure changes in terms of the middle being the proportion of workers in the middle of the employment distribution decreasing. So it's not that employment there decreased and increased in the tails, but what happened was that the average earnings in the tails increased while the structure remained the same. What you see is a little bit of a change in the composition or the fact, the composition effect coming from RTI, especially in the first sub-period that we analyze, but that's quite small when we compare to other factors. Yeah, that's here, what I had, yeah, that's pretty much what I said, okay, good. So we describe the data, the method, Brazilian context, polarization Brazil and Genie using this brief of the composition. So the data that you have is the household survey data from 2003 to 2019, so I have those two different household surveys. As I said, we're going to get both workers in the formal and informal sectors, and informal sector is quite important in Brazil, it's about 40% of implied people is in the informal sector, age range is between 15, 64 years old, male and female, rural and urban employment. So what we do to make things comparable to other countries, we have this crossover between the Brazilian occupation classification, which is a Brazilian type of classification of occupations to the East Coast 88. And then we apply ONET, and also we use Simone's paper with Qualters to extrapolate and have this country-specific measure. Okay, so what we do first, we see whether we have this polarization or not by running the same regressions that was shown before. And then we look at the Genie and try to decompose the Genie, we know the Genie is not decomposable, but to do kind of the composition to get the importance of employment or the importance of occupations in explaining changes in the Genie or how Genie evolved over time. But then we look at also our Genie, the reef regressions, but also at specific quantiles using this reef, the composition methodology. Okay, so what we have, so this is the decline, so this table shows the decline in earnings inequality for the first sub-period, okay? You can see that it declined here using the variance of log earnings or looking at the Genie, either in log or in levels. We see that there was an decline, that's what we saw in the previous figure, first figure, and here we have this growth instance curve that showed exactly what happened, right? So the yellow, I'm sorry, the red line is the first sub-period, and we can see that basically the poorest part of the distribution, God, the highest increases in wages. For the second sub-period, which is the yellow one, you see pretty much something flatter, but here you can see that those people here in the bottom didn't get raised. It was actually a negative growth, right? So that contributed to this increasing inequality. Overall, which is this blue line, we see that have kind of inverse U-shape for the growth instance curve. Okay, so when we look at how this measure, this routine task intensity measure evolved over time, you see that there was a decrease over time in result, either using country-specific, which is larger in level than using ONAT, but you see a decline using both measures. And then when you look at what happened over time, and here you have, you probably cannot see, but this is 2003, and then this is 2011, and this is the last one. Okay, the triangle is the last one, 2018. And these are not percentiles, right? Do not have 20 percentiles. Those are the so-called demidiciles or van-tiles, and I just learned that from Wikipedia recently. Okay? This is the name of when you divide distribution this way in 20 fractions. Anyway, so you see that the decrease, right? Those curves are pretty much the same, but there was a decrease over time, and this decrease happened here at this part of distribution. So five years and 25th percentile, and you see that until I would say the 30th percentile there was a decrease, especially in the first sub-period, okay? And that's why you see when you do the composition, a composition effect coming from RTI. And those are the results of the same regressions that Simone ran for Ghana, we did the same for Brazil, and the things that are important here to note is that in order to have this polarization effect, right, you need to fit to this quadratic curve. So basically, you'd like to see a negative coefficient here and a positive coefficient there. And that's pretty much what we see here in this second sub-period. We do not see that happening here in the first one, right? Right? Either looking at the employment share, or in particular for the mean earnings, right? That's where you see most of the things going on, and we look at that just at our square that's pretty much small. So this is not something, well, the first thing to note is that, I mean, you're earning occupation regressions. So we have 78 regressions. So for any microeconomician like myself, that's something that we are not very proud of showing, right? But we are replicating this method, and we can see that this part here fits exactly the story that we wanted to tell. That, I mean, there's not much going on in terms of employment structure, but it's pretty much happening in terms of earnings being paid at the tails of distribution that changed. When you look at, instead of using earnings in the previous period, but using those task measures, you got something similar, right? And here, why we're not getting a negative here is that because, I mean, remember the negative relation between task intensity and earnings, right? So you would see, you get exactly this kind of behavior, positive here and positive there. That's what we would expect and get that in many specifications that you have. So if anything happened, it was more in terms of happening in earnings, not employment shares. Then what we did was, okay, let's try to decompose the genie in between and within components using this Shorok, the composition, which is a clever way to provide exactly the composition. And we can see that basically half of, for each year, between, I mean, it's not exactly half, but I would say 46, 47%, 44, and 44 there are the part that's explained by between occupations, okay? And then we do lots of exercise, like let's fix the share of those employment or those groups, right, those job groups, the same as in 2003. And then what we did is let's keep the average earnings being the same in 2003. So we can kind of look at the difference between what happened here, for example, right? This 48, 5, with this 0.49, when you keep the share constants basically, what's going on in terms of average earnings affecting inequality here. So we can do that for all kinds of the compositions here. And there is not much going on from here. So basically occupations do not seem to be, I mean, they explain a lot in a quality level but not in changes. That's the main message from those tables. And here, let me jump directly to here, because here's when you got those differences here, right, 0.537 to 0.485. So it's basically five or six points in the genie. So how do you decompose that? So that's basically, we are decomposed into those effects, a composition effect and a structure effect. And this middle here, it's something that is produced by the methodology. And for Brazil, it's not very good, because we'd like to have this very close to 0. So that's basically why in the other paper that I have with Chico, we didn't do exactly this way. We didn't have a way that we didn't have this part here, which is produced by the way that we're doing those compositions. But basically, when you look at RTI here, you see that it had an impact, but not as large as many other factors to reduce inequality. So that's a composition effect and nothing going on later. And it was especially in the first sub-period. When you look at those, when you plot what happened by quantile in terms of changing the logarithm, so those are the growth incidence curves, and you decompose that by using a re-weighting scheme like in the another Fortale Mie. You see that the blue is the actual one in the first sub-period. So like something that's declining. And then when you see what's the fact, what are the importance of the structure effect? You see the structure effect is much more important, especially for the first half of distribution in the composition effect. Composition effect is basically flat, but increasing in the upper tails. The second sub-period, there's nothing much going on, especially the composition effect here seemed to be inequality enhancing. And then we did exactly the same kind of decomposition by quantiles that Simon did. And to be honest, I don't like this figure here. I think it's very difficult to see what's going on. But you can see that RTI doesn't seem to play an important role there. And I think that's age, well, no age is here. Anyway, I think that the thing that I don't like here is because we are looking at those pure structure effects and when we should combine those two things here. So that's why I don't think this is something that we should be doing, but we are all doing the same things in all papers, but perhaps Brazil should do something different. Okay, so the conclusion here is that no evidence of earnings or employment polarization, more like poor on pro-rich growth, reduction in the qualities driven by structure effects, increasing the quality driven by composition effect, small role of RTI, reduction in RTI, increasing the quality between 2003 and 2019, or the overall effect, right? But to us, inequality reducing the first period, the composition part, right? And enhancing the second. And that's pretty much what I had. Thank you.