 Thanks a lot to Wyther for the invitation. This is work that was financed as part of a call for research proposal by UNU Wyther on tax benefit systems and crisis. So the work I'm going to present today is looking at earnings inequality, top earners during the COVID-19 pandemic using administrative Ecuadorian data. This joint work with Olivier Bargan on University of Bordeaux and Paul Carrillo-Mardonado, University of Las Américas. And I must say that when we started this work, we had some prior hypothesis, mainly that given the big inequality in the region, perhaps top earners did not suffer that much during the pandemic. So that's why we said, okay, let's go and look for administrative data to see if we can show this. However, when we started using the data, we came up with some things that we completely did not expected. So the paper, actually the topic, if I could change it nowadays, I would. And you would see why. So the motivation behind the work is that due to data limitations during the pandemic, most of the assessment of the impact of the pandemic on income distribution was made using either rapidly available or rapidly collected service or simulated data, okay, based on household survey data. However, we know that household survey data suffers from a top income under coverage. Therefore, we are not able to assess correctly what happened at the top of the income distribution. In that sense, the aim of this paper is to assess whether at which extent top earners in Ecuador experience changes in earnings during the pandemic compared to the rest of the population. In this sense, we complement other studies that have used survey data to assess the impact of the pandemics on the income distribution. And then because the call for research proposal was about the role of taxes and benefits, we also look at the potential of tax reforms then in order to see how this might reduce inequalities. Perhaps tax reform before the pandemic already might have helped to collect income to have better social protection, but most general in order to reduce inequalities in the country. The data we use is administrative data from social security records in Ecuador covering the entire universe of individuals paying social insurance contributions, monthly data from January 2019 to December 2021, okay. The data is extremely rich. We have panel data for all these periods. Monthly earnings plus the data has been merged with other sources of administrative data. Therefore, we have labor marketing information such as days of work per month, industry occupation, sector of work, and also we have sociodemographic variables like age, gender, education, marital status. So it's extremely rich in that sense. It also allows us to look at the heterogeneity of the impact of the pandemic during this period. However, there are three main limitations with the data. As you might know, the main limitation where you use administrative data for developing countries is that you miss the whole informal sector, okay. So this is a huge disadvantage because you are not able to do distributional analysis. However, if you're looking at the top, this is still the best you can use, right. However, we have two other limitations in the data. The first one is that multiple sources of income are not properly captured in the data. Why? Because it's social security registers. So people usually people who have different sources of income, they usually affiliate only to one type of regime. And therefore, you'll capture only that the income from that type of regime where they are affiliated. Okay, so imagine people who have employment and self-employment income, if they affiliate on the employees, social insurance contribution system, we might capture only their income from employment. Moreover, and this is an important limitation, affiliation to social security is optional for the self-employed, okay. So our data does not capture properly self-employed workers. And this is a big limitation because from previous work we have done using tax records data, self-employed workers, high income self-employed workers are one of the big groups in the top of the distribution, okay. In terms of the methodology, we follow the literature on top incomes based on administrative data, in particular, and when we started the work, what we wanted to do is see what happened to those that were top earners before the pandemic. Okay, so we created in top income groups, 10%, 1%, 0.1% based on annual earnings in 2019. And then we follow these individuals and we see how their employment evolved from January 2019 to December 2021 compared to other, to the rest of the population, okay. So the analysis has four main parts. First, we assess overall changes in employment, so not distinguishing between top earners and the rest of the population. We look at changes in employment and earnings inequality as well. Then we characterize the group of top earners and then we analyze the trajectory in terms of employment and earnings for top earners compared to the rest of the population to finally look at the impact of some hypothetical tax reforms. So this is what we get if we look at the overall change in employment from January 2019 to December 2021. And if we look at this picture, it would look like the precise picture that you would expect from the pandemic, right? So basically registered employment is more or less stable during the pre-pandemic period. It drops sharply during the second quarter of 2020 when in most countries in Ecuador in particular, there were strict lockdown periods installed. And then it recovers, okay. So kind of really what we would expect. If you look at monthly earnings, mean monthly earnings, these are in US dollars, the currency that we use in Ecuador, with something similar against what you would expect. Mean earnings are stable in the pre-pandemic period. They drop sharply in the second quarter and then they recover and slightly higher than in the pre-pandemic period, okay. For your information, these peaks that you see here in March and in December, this is due to some bonuses that are paid that are what is known as the 13th month and the 14th month payments during the year. And basically that's what you see, you know, these peaks of the bonus is paid there, okay. So, so far, nothing really out of the ordinary. However, we wanted all the story we could conclude here is, you know, there was the pandemic, the economy is recovering, right? However, if we disaggregate with the information we have, the picture is completely different. This is when we see changes in employment across industries, okay, where we have normalized employment relative to employment in January 2019. Here, what we see actually is that already in the pre-pandemic period, employment or registered employment was decreasing in all industries across the country, okay. The only industries where employment or registered employment was increasing was in low-paid services and activities such as arts and entertainment, services like hiring equipment. And is these people in these sectors that are driving the recovery during the pandemic, okay. Why is this worrying? Because this means that before the pandemic, the Ecuadorian economy was doing a bit bad, okay. So, people were deciding to exit registered employment to move to non-registered employment. If we look at changes in earnings, then this why I was telling you the previous picture is so worrying, look at mean earnings of the group that is driving the recovery in terms of employment. These are really low-paid, these are really low-paid sector, okay, compared to other sectors, yes. Exactly, yeah, these are deflates. So, then again, reasons to be worried because although the economy recovered from the pandemic, the economy is not recovering from their structural problems, okay. Now, moving to the top earners group here, just a few details and characteristics. As I was saying, the problem here of not capturing well the self employed, some interesting facts, for example, public sector workers are in the top 10% to 1%. So, these are high-paid sector, but of course, then when we move to the very top, then we have few public sector workers, okay. In terms of education, tertiary education, again, they are concentrated in top 10, top 1%, but we have a drop in the top 0.1%. What I was referring in terms of industries, look here, for those who are not in the top, the main group is those in other services and activities, okay. How does this translate into the effects of non-top earners versus top earners? This is what we get and mainly driven by what we saw in terms of industries, okay. So, employment for those who are not at the top before the pandemic was quite stable, it dropped during the pandemic and it recovers. And this pattern of those who are not at the top is what is driving the overall picture that I show at the beginning. At the top, however, as I mentioned, employment, formal employment or registered employment was already decreasing somehow during before the pandemic and it continues decreasing throughout the pandemic, okay. In terms of monthly earnings, a similar picture, so the recovery in terms of mean earnings was mainly for individuals at low-paid activities. We see, if we look at the top 10%, we see a recovery in 2021 with monthly earnings coming to pre-pandemic levels, but if we look at the very top, actually, monthly earnings dropped and they remained at the levels of the pandemic, okay. So, something that we completely did not expect at the beginning, we thought that top earners would be doing better than the rest of the population. Now here, we just tried to exploit some of the information that we have in the data, looking at differences across gender. Something interesting that we see is that at the bottom of the distribution of register earners, the gender gap seems to increase, okay. This raw gender gap whereas at the top, it seems to decrease, okay. We have not got into the details, but this might be driven because of what are the characteristics of the people who are exiting the labor market during this time at the bottom compared to the top. We do something similar for education level and we see the same thing, having the tertiary education of the gap between individuals who have tertiary education, non-tertiary education increases at the bottom and it's sometimes somehow narrows at the top. So, finally, just in the minute that I have left, then we try to see what role fiscal policy plays here and what we did was to first simulate social insurance contribution payments and personal income tax as it was designed in these years and then we look at a hypothetical reform of personal income tax to make it more progressive and more redistributive. In particular, as we saw, there was these big peaks, remembering the periods where bonuses are paid, these bonuses are not subject to personal income tax. So, one of the changes that we do in our reform is to make them subject to personal income tax. Another change that we do in our simulated reform is that in Ecuador, we have extremely generous deductions for personal expenditures as part of the features of personal income tax. This means that an individual falls in the fourth tax bracket paying a marginal tax rate of 15 percent with deductions from personal expenditures. This person can easily fall out of the tax brackets and pay zero tax. So, in our hypothetical reform, what we do is to completely abolish deduction for personal expenditures. And here I'll just show you the impact of this on the Gini coefficient. The blue line depicts the Gini coefficient from pre-tax earnings. The orange line depicts the Gini coefficient from post-tax earnings under the current or the actual system. And then we see our hypothetical reform in the dashed line in red. Okay? And here, just the outtake of this is that although personal income tax and social insurance contributions decrease income inequality, they could do much better if we had this type of reform where we abolish these generous deductions and when we make these bonuses subject to personal income tax. In particular, tax revenue from this reform would increase by around 35 percent in the pre-pandemic period. This means that that budget could have been used to enhance social protection and we might have had better protection for individuals in the or during the pandemic. I will conclude here. So, the main outtake from this work when we started doing it or when we realized that what we didn't expect is that Ecuador was already experiencing a deterioration in terms of a registered employment before the pandemic with a progressive decline in the number of earners affiliated to social insurance contributions compensated by an increase in employment of low paid activities. Okay? In the wake of the pandemics, these patterns are reinforced. Earning drops in the second quarter of 2020, they recovered for those at the bottom but they remained low for those at the top and then finally looking at personal income tax with just some features we could do better in terms of collecting tax revenue and this will have a redistributive effect in Ecuador. Thank you very much.