 Good afternoon, everyone. First of all, thank you for being here and thank you to the local organization committee and the committee in general for this invitation. It's a pleasure for me to present this job called Gender Labor Earnings Gap in Costa Rica over the last decade, what drives it and the effect of the COVID-19 pandemic. As you may see in the title, this work has two parts. One is measuring the gender pay gap during a decade and the decomposition, identifying the main determinants of this gap. And the other part is related with the effect of the COVID-19 pandemic. This work was conducted with three more colleagues who are in Costa Rica and in the World Inequality Lab in Paris, France. Okay, regarding the motivation and contribution, as you may know, gender pay gap is a reality in the entire world. According to the last report of the ILO, there is a gender gap of 22%. The salary, the pay is 22% higher for men than for women. There is bad literature assessing this gender pay gap in developed countries, but it is really scarce in developing countries. It's scarce and at the same time, there are least literature analyzing inequality between groups. There are more literature analyzing inequality in the entire nation, but not between groups. Most of this literature is concentrated in countries with largest populations, like India or China, are more based in cross-sectional data. There is also more literature using traditional models like mincerian earning functions and the composition methods. And also, despite GDP, GPG gender pay gap has been decreasing, as you may see, is remaining in favor of many in the entire world. As part of the motivation, there is a concern about the effects of the COVID-19 pandemic in gaps in general, in socioeconomic gaps, but definitely one of the main concerns or one important concerns is about gender labor gaps, specifically in my country. We have a problem, a structural problem, because women have less access and less pay than men, and it's been a situation repeatedly during the last two decades with no solution. There are previous studies doing this kind of studies analyzing the effect of the COVID-19 pandemics. One of them is in India with an OLS model, and they do an analysis pre- and post-pandemic, and they analyze the differences. There is another study, recent study, analyzing this gender pay gap doing micro simulations based on previous data set they have like two to three years previous the pandemic. What is the contribution of this work? First of all, we are adding new literature to the major and the composition of the gender pay gap in developing countries by using a large-scale data set in a decade and using a vast set of human capital and job-related predictors. We have a really good data set, it's a survey, as you may see in the next slides, and we have fortunately good controls to conduct this research, and to the best of our knowledge, for the first time, we use panel data to try to estimate the effect of the COVID-19 pandemic in the gender pay gap. Definitely it's good for you, also it was really good for us to study the context. What is the context not just in the pandemic crisis, also during the period of analysis? What is the evolution of the market, and what is the evolution specifically in the differences between men and women? As you know, in the majority of countries around the world, the pandemic produced a truly difficult situation. One of these situations definitely is the decrease in the production, especially in some specific activities which depends entirely from, for example, tourism, which is an activity which depends on a broad tourism, also on other activities like commerce and transportation and storage, which are activities who were really affected because of the lockdowns we faced. Sorry, you may see also, this is just informative, the first part, maybe the most important part is the decrease during the COVID-19 pandemic, similarly maybe to most of your countries. And this is more specific to the labor market in Costa Rica. Our country has been facing definitely different structural problems. One of them is the unemployment rate, another one is the under-employment, the rate of informality, and also differences in participation, not the optimal participation, like the natural rate and unemployment definitely, and maybe the most important problem is the differences between groups. We are talking about women, youth people, people who live in rural areas, people who are employed in occupations with lower levels of skills, etc. And this is really important for us because most of these variables we have to take into account to understand the results we obtain. What about the data? As I said before, we have a cross-sectional dataset from 2010 to 2020. This is around 1 million observations and our final sample of 300,000 observations. Also, as you want to measure the effect of the COVID-19 pandemic, we took advantage of a panel dataset we have during this period 2020-2021, specifically the first quarter of 2021. With 100,000 observations, 140,000 observations, and our final sample of 40,000 observations. As you may see, most of these observations are concentrated in men, not in women. And as you can imagine, this is because our dependent variables are salaries. And this is the people who are into the labor market, and in Costa Rica, as you see. Before, there is an important gap in the interest to the labor market. Because of this, there are more men than women. Oh, sorry. Well, about the empirical strategy, we use also two different measures, two different empirical strategies, but related. Because the first we use is the Oaxaca Blinder, the composition, which is, as the majority of you know, well known, the composition method to analyze gaps between socioeconomic groups. And also, we make a correction using the Heckman correction bias term, which corrects for the characteristics which influence the entrance into the labor markets between the groups of analysis. Regarding the main predictors we use, which is this component, the Z, we use some of the predictors already the colleagues explained, like the skill levels, we use occupations, we use, if the person works in a public or private sector, we use the traditional mincerian function predictors, education or schooling, we use experience tenure. And we also use another particular predictors for our country, which is the difference if a person is working full time or part time is important, particularly in our labor market. We use another controls also, like if the person lives in a rural or urban area, etc. And regarding the second method we use is an extension of the Oaxaca Blinder, the composition. It's a really up to date command in 2020-21 around August was released. And we took advantage of this new command to, because it is specialized in measuring cross sectional data, analyzing pre and post event, and it's also suitable to analyze panel data sets. Because of this we use this method, which is almost the same that the Oaxaca Blinder, the composition method, but this method adds this term, which interacts the group variable with time variable, specifically with the time zero or period zero in which we suppose is the chalk in this case, the declaration of the pandemic crisis. And we use another predictors of variables who affects this relationship. Based on previous literature and also in the characteristics of our labor market, we use if the person were working in a public or private sector on coving sensitive or not industries full or half time. And after that we obtain the results for both parts, the analysis during the period of during the decade, and the part in which we analyze the effect of the COVID-19 pandemic. This is more the equations properly. And you can see maybe just to put attention on the meaning of the Oaxaca Blinder, the composition methods, with the differences in the means between the salaries of men and women. Women is the base group. And then this method analyze which part of this different is explained and which part is unexplained. The explained part is related with the endowments or characteristics of the population. And the unexplained part, there is a debate in literature, but it's more related with productivity and also with discrimination. There is a discussion, but maybe both of them are important to explain or to, yes, to explain the unexplained part. And related to the identification or the main, yes, the main point of this equation is that we suppose that group zero has the same characteristics of the group one and is the same with the endowments, with the vectors, the vector of coefficients, which we suppose exactly the same, trying to have like a counterfactual. Perfect. Regarding to the results, here we have the evolution of the estimated earnings of men and women, as you may see. There is a gap during all the period analyzing favor of men. And this is of around 0.105 lock points, remembering that our dependent variables is the logarithm of the salaries. And this is the result of the first part. We have a consistent gender pay gap in favor of men. And in the second part, we have the results regarding the effect of the COVID-19 pandemics. As you may see, and this is important to mention, to highlight, based on these results, there is a reduction on the gender pay gap during the COVID-19 pandemic. But not necessarily there are good news, because most of the explanation behind these results is because the women who maintain in the labor market wear women with higher salaries. And all the women with less salaries were out of the labor market. This is the explanation behind these results. And all these results is especially significant during the periods closer to the initial shock, which is also an expected result. Here there are some rubbleness checks we use. We compare the results of the part of the effect of the COVID-19 pandemic with the conventional WAHACA blinder estimation. We also make some estimations splitting our sample in specific groups who may influence the results. Fortunately, we have that these groups didn't affect the results because their results were pretty similar. And also, we have pretty similar results using the WAHACA blinder, the composition method, the traditional, which was a good news for us as well. Regarding the conclusion and discussion, definitely these kind of words contributes to leadership, especially for developing countries, and also for public policy. Because in Costa Rica during the last, I don't know, maybe five years, we are talking about the bridging of the gender pay gap, but using the data not necessarily corrected for controls who change completely the results. At the beginning, we have the same results, but then correcting for the predictors the necessary controls, we have the reality. And maybe this reality is underestimated because, as I explained before, the more vulnerable people in the labor market in Costa Rica during the pandemic were women or youth people. Also, it is important to understand that this is not the final. We have to analyze more in depth what happened later. We have more data, and we are conducting new analysis trying to understand if this bridging of the gender pay gap was something in a specific period. This is our hypothesis, and then the gender pay gap returns to this situation, which is a consistent gender pay gap. Just to finish, I read this phrase last week, and I want to read for you. The gender pay gap is rooted in systemic inequalities. Equal pay is essential not only for women, but to build a world of dignity and justice for all. Equal pay for women are nothing less. Thank you.