 Okay, well, thanks a lot. It's a pleasure to present in this forum. I'm going to talk about COVID and labor markets in Latin America. And most of what I'm going to talk about today is based on a joint project with my co-author Gustavo Leyva from Banco de Mexico, and he's a project sponsored by the Inter-American Development Bank. So, as we all know, the COVID pandemic and all the related confinement policies have acted as an unprecedented shock to labor markets in several countries. And I'm going to address two questions today. One is, how were labor markets in Latin America affected? This is going to be a very descriptive, data-oriented part of the presentation. I'm going to start with evidence from Mexico. Mexico, this is a country that we are most familiar with. And then we are going to stand it to other Latin American countries. And in particular, I'm going to focus on the role of the participation in the labor force and the informality margins. This is because these two margins have proved to be quite important in emerging economies in terms of understanding business cycle fluctuations for the labor market. And I have a previous work with Gustavo Leyva in which we exploit exactly these margins. I'm going to compare also what has happened in this recession in Latin America with past recessions to give you an idea of how different this recession is in terms of labor market outcomes. And in the interest of time, I'm not going to go very deeply inside the last bullet, but I'm going to mention some heterogeneity of these effects inside the labor force, for instance, by gender, by age, by sector. And the second part of the presentation is going to be about the model. We are going to use a structural model of the economy to try to decompose what are the main shocks affecting this economy. And as I said, a more detailed presentation of all these results are in the paper that you can see at the bottom. So let me start by talking about Mexico. In this picture and the one that I'm going to follow, the yellow bars, like this one, these yellow bars, represent the change of key indicators. Sorry, I think I went one too. So these yellow bars represent the change of some labor market indicators during the COVID pandemic, starting in 2020 Q2 and up to the end of 2020. And the red bars that you will see at the side represent an alternative recession. In the case of Mexico, we're talking about the 2008, 2009 great recession just to see how different this recession is. So the first thing that you see is a plunge in employment, not surprising. But what is surprising is the magnitude. The size of the fall in employment in Mexico was completely unprecedented compared to past recessions. Again, the recovery has been much faster as well. In terms of the two margins that I was talking about, unemployment versus labor force participation. In the past, it used to be a combination of people living the labor force of increasing inactivity and an increase in unemployment. In this case, it is mostly all about living the labor force. So these people that lose their jobs at the beginning of the pandemic, most of them ended up living the labor force rather than staying unemployed. Unemployment is a very non-cyclical, non-informative margin in countries like Mexico. Labor force participations tend to be more cyclical and have more explaining powers in terms of the fall in employment. Now, about informality, here you see a huge qualitative difference between this recession and the previous ones. And this is the behavior of the informality rate. It used to be the case in past recessions that informality rate acted counter cyclically. What I mean is that when we had a recession, the informality rate tended to increase. And this is not so much because there is a reallocation of formal workers to informal sector. Actually both formal and informal employment tend to decrease, but informal employment was sort of a cushion that was buffering the fall in total employment and as a result, informality rate increased. But here in this recession, the opposite happened. What we had is a big decline in informal employment that accounts for most of the decline in employment in Mexico. And as a result, formal employment also fell, but to comparable levels from the past. And as a result, we got a big drop in the informality rate, again, recovering quite fastly by the end of 2020. We also have, since we have access to micro data, we can construct gross labor flows. And here I'm just summarizing in terms of job creation and job destruction for Mexico. Here is everything aggregated, formal and informal. And the point that I wanted to make here is that in the past, the drops in employment were resolved both of a lack of job creation and some job destruction, especially at the end. But here it's clear that the recession started or the plunge in employment started basically as a massive job destruction that we haven't observed in the past. What is interesting though is that job creation seems to be now playing a role in terms of the recovery of employment by the end of 2020. In the paper, I'm going to mention this just briefly in the interest of time, but in the paper we have a very detailed discussion of what I'm going to show in this slide. We find that there is a big role for two non-conventional margins at the beginning of the pandemic. These two margins were negligible in past recessions. One is called temporary layoffs. Those are people that in the household service report to be not employed, not receiving wages. However, they expect to come back to their old jobs relatively quickly. So this margin increased a lot again at the beginning of the pandemic and faded away by the end of 2020. The mirror of that so this could indicate that the drop in employment that we observe is maybe not so bad in the sense that most of these layoffs were just temporary. But there is also another margin in the in the surveys that we look at, which is called absent employees. Option employees are people that report to be employed. They receive some compensation, but they are not currently working. They're at home. And this category also increased a lot at the beginning of the pandemic and faded away by the end of 2020. And that might indicate that actually the drop in employment is larger than what the employment rate indicates. Something that is interesting is that this margin of temporary layoff is more important in the informal sector, while in the formal sector we see more of these type of absent employees. So this is one thing that we document in the paper. Another thing that we document in the paper in detail, and I'm not going to be able to talk too much about that now, is about the different differential impact of the pandemic on different groups. So as one might expect, in terms of the drop in employment that was larger for women, for younger workers, and also in some contact intensive sectors as services. However, what I want to stress is that the fall in the informality rate that I reported at the beginning in the aggregate is not driven by a composition effect only. I mean, it is true that women, younger workers, and services are groups in which the informality is higher. However, the drop in the overall informality can be seen in all sectors. It can be seen in manufacturing, in agriculture. It can be seen among men and among older workers as well. Here we extend some of these results to older a set of five Latin American countries, including Brazil, Chile, Colombia, and Peru. Very briefly, we confirm the same changes in labor markets that I just reported for Mexico for this wider, bigger set of Latin American countries, the large drop in employment, large declines in participation rate, declines also in informality rate with one exception. The only exception here is Peru, which is the country with the largest informality rate, and it's the only country in the sample in which the informality rate actually increased during the pandemic. So about the second half of the presentation, I'm going to talk about what are the shocks that have affected the these economies and can explain the behavior of these labor market indicators that I have just discussed. For this, of course, I need a structural model in order to filter those shocks. The model is based on a previous work with Gustavo, this is a paper that is already published in the Journal of International Economics. I'm just going to mention very briefly, I'm not going to show you any equations, there is no time for that, but it's a model of a small open economy subject to technology interest rate shocks. It does have some of the labor market margins that I have mentioned like endogenous labor force participation, formal and informal sectors, matching friction that give rise to unemployment. And this model is calibrated to Mexican data before the pandemics. And then for the pandemic itself, we extend them all with two new shocks, a labor supply shock, and a sector specific productivity shock to the informal sector that are meant to represent in a reduced form the impact of COVID on this economy, COVID and the confinement policies arising from it. So we use the model to recover the shocks that rationalize the behavior of output, employment, informality, also the foreign interest rate. And then what I'm going to show you here, and this is the last slide, I'm going to show you what are the contributions of the shocks to the pandemic recession, in particular to the slump in 2020 Q2, and compare it to the 2008-2009 Great Recession for Mexico. And as we can see in this pandemic recession, the two new shocks, that is the labor supply shocks and the shock to the informal sector, which are key to account for the drop in employment and informality rate, play a huge and massively important role in terms of all the variables including output. While in the Great Recession, the traditional shocks, that is technology and interest rate shocks, were driving most of the action in terms of both output and the labor market. So in that sense, and just to conclude, this pandemic recession is different from previous downturns, not only in terms of size, but in terms also of what is the behavior of the informal sector. It is, we have identified some shocks that might affect the economy, but still we need to do more work in terms of validating and rationalizing these shocks through things that we actually observe happen during the pandemics. And that will give us a safer ground to start talking about policy. It seems that policy that incentivizes recovery should focus on formal job creation, but how to target this particular margin is a very interesting question that we might talk more about. Let me stop sharing and thanks a lot for your time. Hi, I'll try to share my screen now. Great. Can you see the presentation? Yes, we can. Perfect. Thanks. Thank you so much for this opportunity to present our project, The Impacts of Ingressive Solidarium, to confront the COVID-19 crisis in Colombia. This work is joint work from a very impressive team from the IDD, DNP, and La Universidad de los Arios. The twin health and economic crises of the pandemic led to two public policy challenges. First, governments needed to rapidly expand the coverage of transfers to new households. And second, governments needed to make disbursements in a way that would not contribute to the spread of the coronavirus. So we study Ingressive Solidario, the flagship emergency transfer program in Colombia. Ingressive Solidario was designed to address both of these policy challenges. It expanded the coverage of the social safety net. It provides transfers to households that are vulnerable, but previously not poor, and they were not recipients of other social programs. The government also collaborated with financial institutions to bank beneficiaries, delivering transfers digitally to simplified savings accounts could reduce the risk of contagion. I find this context particularly interesting for two primary reasons. First, there's relatively little evidence on the impacts of cash transfers from programs that are designed and implemented as emergency programs. And this is relevant for designing new programs to confront future crises and designing more flexible social safety nets. Second, our study is composed of vulnerable, but previously middle class households. Middle income households are especially vulnerable to labor market shocks. Almost all of their income is earned through the labor market, but they don't have enough wealth to build sufficient resilience. In contrast, almost a quarter of low income households income comes from social programs, which is not vulnerable to labor market shocks. The Ingressive Solidario program covers over 3 million Colombian households. Households must meet several requirements for eligibility to highlight two of them. The Ingressive Solidario program includes households that are not covered by any other social program and are informal workers. Among households that meet the requirements, eligibility is determined by SISPEN score. The SISPEN score approximates per capita household income relative to the departmental poverty line. The program consists of monthly transfers starting in April 2020. Each transfer is 160,000 Colombian pesos, or about 125 US dollars at PPP. Using data from 2019, the monthly transfer represents about 17% of the median income for households in poverty. A key feature of this program is that the government collaborated with mobile phone operators and financial institutions to bank previously unbanked beneficiaries. We conducted a nationwide telephone survey in November and December 2020. We prioritized surveying households closest to the cut-off for program eligibility. Our sample contains a little over 3,500 households, about equally split between eligible and ineligible households. Our response rate to the survey was about 25% and it was equal in both eligible and ineligible households. This table shows descriptive statistics for our sample. I'll highlight just a few key rows. Relative to February 2020 income, 38% of households lost income in June 2020 and the average income drop was 11%. 30% of households were skipping meals and around half of households were eating less meat and vegetables. 26% of households lost a job and 35% of households experienced a business closure. We use a standard regression discontinuity design at the household level and we evaluate the impact of the program by comparing eligible and ineligible households around the cut-off of suspense score that determines eligibility. In our sample, eligible households have suspense scores just below to the left of the cut-off and ineligible households have suspense scores just to the right or just above the cut-off. This means that our sample is a subgroup of households that have higher pre-pandemic income than the average program beneficiary. However, our evaluation is informative in two important dimensions. First, the eligible households in our sample have higher pre-pandemic income than the average beneficiary so we could expect higher average impacts of the program. Second, the evaluation delivers a critical policy parameter. This is the impact of expanding or contracting eligibility for the program at the margin. We estimate an intention to treat effect. This means that we estimate the effect of being eligible for the program and not necessarily receiving the program transfers, although I'll show in the next slide that take-up and compliance is very high. Compliance is very high and we see a discontinuous jump in receiving the program transfers at the eligibility cut-off. 87% of eligible households in our sample receive ingressive solidario and similarly non-compliance is extremely low. This means that in practice being eligible for the program is nearly the same as receiving the transfers. The validity of our RD design requires that households cannot perfectly manipulate their suspense score meaning that they cannot definitively decide whether to fall on the left or the right side of the cut-off. We don't expect manipulation in this context because the data was collected on the data to determine suspense score was collected months before the pandemic began or the program was announced and also the determination of the cut-off scores were centralized and not publicly announced. But this figure shows that the distribution of the density of households around the cut-off of program eligibility is there's no jump at the cut-off. So the red bars represent our survey data and the blue bars represent administrative data or the universe. And we don't see any abrupt changes in the density around the threshold implying that there's no evidence of manipulation in the suspense scores. We look at outcomes in six main categories of household well-being. Income, labor markets, consumption and spending, education, domestic violence and mental health and financial inclusion. This figure shows that precisely at the eligibility cut-off there's a discontinuity in monthly income per capita in September and June 2020. Eligible households report greater income in both months. This effect is driven by an increase in the probability of reporting a positive non-zero income which emphasizes the severity of the income shock that many households faced. There are two competing channels through which transfers could have impacted labor market outcomes. First the transfer may have helped households adapt their businesses to the realities of operating during the pandemic. Conversely, the transfers may have allowed households to withdraw from the labor market and decrease the risk of exposure to COVID-19. We find no impacts on labor market outcomes such as formal or informal employment or hours worked. This suggests that the program did not create disincentives to engage in the formal labor market or informal labor market. Overall, we don't find significant effects on food consumption or expenditure or on a food security index but we also look at heterogeneity by labor market shock. Specifically, we calculate the percentage of workers in the household who lost their livelihood. The panel on the left shows the effect among households in the bottom two-thirds of this distribution and here we don't find any significant effects on food consumption per capita. The panel on the right shows this effect among households in the top one-third of this distribution. In this case, we find a discontinuous increase in food consumption per capita indicating that the program was able to attenuate the adverse effects of job losses. Generally, we don't find significant effects on aggregate monthly non-food expenditure but we also look at categories of non-food spending by looking at both the intense of and extensive margins. Some expenses like rent or basic utilities are fixed so the decision is more whether or not to make a payment than how much to pay. We find an increase in the probability of health and cleaning expenditures and on education expenditure. We also find an increase in per capita education expenditures. The higher expenditures on education could have increased the productivity of study time. Here we show that it actually also increased study time. In column two, you can see that the program increased the probability that children complete at least four hours or about half a day of school work by 11 percentage points. It also increased the amount of time studying by almost half an hour. We do not find significant effects on other educational outcomes. Next, we look at self-reported domestic violence and stress outcomes. But despite the abrupt changes in income and circumstance during the pandemic, we don't find any significant effects in this area. As I briefly mentioned before, the Ingressive Solidario program included a collaborative public-private effort to bank previously unbanked beneficiaries. The program encouraged households to receive transfers digitally through simplified savings accounts that could be opened through mobile phones. The panel on the left shows a 14 percentage point increase in the probability of opening an account in 2020 and this represents 57 percent relative to ineligible households. The panel on the right shows that eligible households were 7.7 percentage points more likely to make a digital financial transaction. This represents an increase around 100 percent relative to ineligible households. We also found an increase in using digital platforms to check account balances and to send and receive transfers to other households. But we find no effects of savings in digital accounts. To summarize our main results, we found that the program had positive effects on several measures of household well-being, such as income, food consumption among households that suffered severe labor market shocks, spending on health and education, and education, and financial inclusion. Although many of the impacts are modest, the impacts in health, education, and especially financial inclusion could have important implications in the longer term. Thank you. Thank you very much, Brigitte. Let's move on directly to the next one. So, Nicholas, and then we can take questions at the end. Thanks. All right. So, let me share. Do you all see the presentation? Yes. Yeah, good. All right. So, I'll start. Thank you very much for having me here. This is a joint paper with Thiago Cavalcanti and Damian D'Amato. Let me start with what I think is a fact, which is that populism has been rising globally in the last 20, 25 years, according at least to the most reasonable measures in economics and political science, the number of countries under left or right-wing populism is now probably the maximum, you know, because you're in the last 100 or 120 years. Now, the goal of this paper is not to define, and it's not even to discuss, what are the characteristics of populist governments, but there's one characteristic that seems to be, I mean, one feature of populist governments that seem to be quite, quite widespread, at least among modern populists, which is their anti-elit, anti-scientific rhetoric in the way that communicates with the population. And again, this is something that both right-wing and left-wing populists tend to share. This is some examples, you know, Trump undervaluing global warming. President Bolsonaro from Brazil literally saying that he's not responsible if you become an alligator when you take the vaccine or even Lopez Obrador, who's the most left-wing populist, saying that if you go to study abroad, then you're going to learn to be a thief and you're going to be an elitist and place classes and races. So, you know, right-wing left-wing populist, this idea of anti-elit rhetoric seems to be very widespread. So the question that we try to answer in this paper is very specific what we try to show or what we try to test if this anti-scientific rhetoric has any meaningful impact on socioeconomic variables that we typically care about. In this case, it's going to be related to public health. And more specifically, what I'm going to show you is what we found in very specific but I think very interesting context, which is the context of Brazil in the beginning, the first week of this pandemic. And the idea of the paper is basically to show that when the president of Brazil Bolsonaro publicly and emphatically challenged the scientific recommendations of, well, of the experts and even of his own ministry of health, as a consequence, that triggered a behavioral change among his followers that started to break isolation rules and started to go out and so on and so forth. So that's basically the idea of the paper. Now there's obviously an empirical challenge, which is that if we want to show that there's a causal impact of Bolsonaro's words and actions on people's behavior, then we need to identify the most salient, the most controversial salient speeches because it's not the case that he one day changed his speech. It's more like a continuum. So it's important to identify those events that are particularly a problem. And to do this, we do it in the most, I think, kind of systematic way that we come, which is basically by analyzing front pages of newspapers. So we took all the front pages of the main newspapers in Brazil that cover probably 90 something percent of the market share. For every day, we identify in the front pages the news that were referring to Bolsonaro speaking against isolation or undervaluing the threats of the disease and so on. Then we measured the proportion of these news out of the total size of the front page of each newspaper and then we created an index averaging them out. So for example, this is 18 percent and this is probably 20 percent. So with this, we can create a measure, which is this one that you observe over here, which is basically an index that goes from 0 to 100 percent. So in this case, the maximum coverage was 20 something percent and the minimum was 0 percent. And as you can see, there are two events that are particularly prominent, March 15 and March 25, which are actually these two events that you observe here in this event. It was in the middle of the first days of the pandemic. He joined a protest, touched people took selfies, even against the recommendation of his sound health minister. And this is March 25, in which he made a public announcement. So all the TV broadcasters were, I mean, it was mandatory to broadcast this in which he basically urged people to go out and restore normal lives and so on. Now these two events just for a business, if instead of taking a look to regular media, you take a look to the popularity of these news in social media, the picture is very similar. This is using Twitter indexes, Twitter indexes and retweets. Again, the two most prominent events are very clear, same when we use likes and same when we use Google search. So it's very clear that there are two events that were particularly prominent in which Bolsonaro publicly and emphatically dismissed the or undervalued the threat of this disease. And the idea is to show that after these two events, there was a significant change in the behavior of Bolsonaro support. Okay. In terms of data, we use quite a lot of data sets. I'm going to describe the three most important ones, social distancing, we measure it, we take it off shell, it's built by a company, a company in Brazil that tracks cell phones. They basically created this index that is the social distancing index, which is the proportion of mobile phone devices that remain at home in a given day. Okay, so they do it at the municipality day level. And they trap more than 60 million mobile phones. Okay, so this is going to be our main outcome. We also analyze economic variables in particular consumer spending. So we got access to daily data, again, at the municipality day level of the biggest public private bank in Brazil, which happens to be the biggest private bank in Latin America. We got all the online and in-person transactions. So we're going to use that to show that the effects are not only in terms of mobility, but also in terms of spending. And then to measure support, we are going to use the Bolshero Bolsonaro at the municipality level in the lectures, the presidential lectures of 2018. Okay. The practical strategy is quite simple. We basically estimate the dynamic difference in the different model in which the treatment is going to be government support at the municipality level, which is the Bolshero for Bolsonaro. And then we are going to interact the treatment of the municipality, which each period of effect, which in this case are days. We are going to control for day fix effects, municipality fix effects on state times day fix effects, because we need some variability within state. And we are going to basically these try to test if there are pre-trans and post-treatment effects. And let me show you just a few results. This is the main result, the baseline result. The outcome here is social distancing that goes from zero to 100, and the mean is 30%. Again, the treatment is Bolshero. So this is one percentage, one percentage point increase in Bolsonaro has this effect in terms of a social distancing index. These are the two events. This is zero is March 14, and this is March 25. There are no parties and divide pre-treatment. So basically social distancing was not significantly different before the important events. And then there's a clear sudden drop on the first event and then another sudden drop in the second event. This is robust to many specifications and I'm not going to talk about that. That's on the paper, but this is quite a robust result. In terms of the magnitude, this is one. So increase one percentage point, the vote for Bolsonaro, and this has an effect of 0.1 percentage points in terms of social distancing. Considering that, for example, if you take a state in the south or one in the north, the difference could be up to, I don't know, 30%, 40% in terms of support for Bolsonaro. So that would mean that the effect, the difference in social distancing would be like 3 percentage points out of a mean of 30% is quite a sensible effect. What happens in terms of economic activity? If you believe me that there was an effect in social mobility, then we should see some kind of mirroring effect in terms of activity. And this is exactly what we find. So no pre-treatment effects. And then there's a sudden increase in the average amount of in-person expenditures. We exclude pharmacy because we want to be sure that we are not capturing essential trips. And this again, big effect after the main event, which persists after the second event. So people are moving more and people are spending more money in person. Now, we try to analyze a little bit of, I mean, what could be happening in terms of mechanism? I'm going to show you just two results. Then we do a couple of additional things, but I think these are the most important ones. It seems to be the case that media is particularly important in Brazil to kind of propagate these type of messages and in particular social media. So we actually analyze the effect, the potential effect, and this is just the evidence of the differential effect between municipalities where media penetration is very high and municipalities where media penetration is not very high. And we do this with traditional media. We also do it with social media. So for traditional media, we're going to divide municipalities between municipalities where there's at least one local TV broadcaster and municipalities where there's no TV broadcasters. And for social media, we did the following. We livestreamed tweets in the whole country every 20 minutes for a week approximately. And then we located every tweet in Brazil so we can have a measure of the intensity of Twitter, the Twitter usage in each municipality in Brazil. So we're going to divide the sample between above median and below median in the intensity of use of Twitter in Brazil. So this is what happens with Twitter. Municipalities with high usage of Twitter are those that are driving the fact completely. There's no effect among municipalities with no usage or low usage of social media. Okay. And something very similar happened when we used to be broadcasting. So municipalities with a TV broadcaster are driving the entire effect. And municipalities with no TV broadcasting are not really showing any effect. Okay. Then again, in the paper, we show quite a few robustness checks, but these are these are the main results. So just to conclude, it seems like an obvious conclusion. But nevertheless, I think it's important. This anti-scientific rhetoric is not free. It's costly. It can have important effects in terms of public health and it can trigger undesirable behaviors. And this is basically because leadership matters and not, I mean, it not only matters in terms of intensive regulations. Actually, leaders' words and actions can have a significant effect on people's behavior. Okay. So that's it. Thank you very much. Thank you, Nicolas. That was very interesting. Do we have any questions? There is one question in the Q&A. I think Carlos already answered this, but let's read it so everyone can hear. This is from Robert Duval to Carlos. Does the model allow for separate labor disutility shocks by sector as to capture that some informal workers prefer flexible schedules as to tend to housework? These workers would presumably to exit the labor force and informal sector. Oh, did I skip something? Exit the labor force and the informal sector as a result of COVID. I already answered in the chat, but just for everybody, the model only has a common labor supply shock. However, it does have informal specific labor demand shock via productivity. And one thing that I was mentioning is if we really want to discipline labor supply and labor demand shocks as in any supply and demand framework, we need to look at prices. We need to look at wages in this case. So that's something that we are adding now to the new version of the paper in order to discipline how much of that is really people choosing to leave the labor force or it's just that labor demand is not there in both sectors. But it's an extension that we are currently working on. Thank you, Carlos. If anyone else has a question, you can click on the audio or video or the Q&A if you want to type it. Otherwise, we are coming to the end. So apologies for there not being a proper chair for this session. But thank you very much to all the speakers. It's very interesting. Thank you very much. Thanks a lot, Lina. Thank you so much. You were an excellent moderator, so don't worry. Thank you. Thanks a lot, Nicolas and Bridget. It was great to hear your talk as well. Likewise. Thank you so much.