 So I will be presenting this work that is joined with my colleague Luis Felipe López Calva, who recently just moved to UNDP in New York, also with Professor Nora Lustig from Tulen University and Daniel Valderrama from Georgetown University. The main idea of this study was to be a background paper to a recent regional study we did at the World Bank on Waging Equality. And this paper focused mostly on some stylized facts to provide some context to that study. Later on I will go over a few points that we didn't do in the study, but we focused mostly on some equilibrium results in terms of earnings as the result of the dynamics between supply, demand and institutional factors in the labor market in Latin American countries. So some context of income inequality in Latin America. This kind of data is all data, but just in general, when we divide the world in terms of the geographical areas as the World Bank does, the growth, the analyzed income growth of the bottom two quintiles of the income distribution grew the largest with respect to the other regions globally. And particularly that ratio of the growth rates of the bottom 40% was almost twice as the mean average incomes when looking at household income data, which certainly you can imagine translated into a steady decline in income inequality. And the topic of total income inequality actually has been widely studied in a region like Latin America by colleagues in Universidad de la Plata Cedlas, Leonardo Gasparini and Guillermo Cruz's, also by Nora Lustig and Lopez Calva, some World Bank and Inter-American Development Bank studies, and Courtney and others. So depending on whether you believe in some sort of averages of genies, there has been a significant decline that has been reported since 2003 across the region, either if you have unweighted or population-weighted averages of genies. We also have another diagram that includes, if you pull all the microdata from household service, you more or less get the same graph for Latin America. And Nora had some issues putting that one together because it will be like the genie coefficient of all Latin Americans, but I think that one is in the paper. But still, even if Latin America experienced such an important decline in total income inequality, it still remains as the second most unequal region in the world, just behind Sub-Saharan Africa. This graph shows the benchmarking of inequality in Latin America with respect to other developing regions. It's quite challenging because Latin America mostly measures welfare with income information, and other regions typically use consumption information, but we borrow some factors of conversion from Alvaredo and Gasparini from the handbook of inequality, and we end up with this comparison in which Latin America certainly is far away from the most equal region in the world that is Europe and Central Asia. So what is behind this recent trend in declining inequality in Latin America? So when you use a sort of non-parametric decomposition to understand those declines, there is not a clear trend in the early 90s from early 2000s in terms of those trends. I mean, few countries experienced small declines in total income inequality, others experienced an increase in total income inequality, but if you really look at the second panel that was from early 2000s to 2011, 2013, in all countries both in Costa Rica, from the 17 countries we have in the satellite database that we use, experience a decline in the total income inequality. And when you decompose, what are the factors that explain such a decline, if you look at the blue bar, that blue bar actually tells us the contribution of labor incomes, and the white bar tells the contribution of the quantities of labor in terms of demographics and the new entrants into the labor markets, and certainly the social protection and remittances have some additional relevance in terms of such a decline that is this gray bar. But overall, the message here is just with the exception of Costa Rica, all other countries in Latin America experience a decline in total income inequality in Latin America, and it has been reported in previous studies. But then the issue here from the previous graph really is that there has been a substantial and statistically significant decline in earnings inequality in Latin America, starting around 2002-2003, which when you use some sort of comparable data from ILO Global Wage Report, you find out that for non-Latin American countries there has been certainly movements in both directions, so there is not a trend that compares to what happened in Latin America. Given that this panel is also to compare other middle income countries, we try to use this Global Wage Index from ILO to include other countries' high income and high middle income countries into this graph. And we have two panels, basically the panel in the top and the differences between, in the Y-axis, we have the annualized change in labor earnings measured by the Gini coefficient, and the X-axis is the labor earnings Gini. And the top panel is in the decade of the 90s. The panel in the bottom is in the decade of the 2000s. And as you can see, all these blue bubbles represent Latin American countries, which departed for a high levels of labor income inequality, and even though there were some significant declines in the first decade of the 2000s, still the levels of labor earnings and the quality is high with respect to other countries in the world. But getting to what explains this success story, so the contribution of this story is that different to other literature that has been thinking about this before, we have a more comprehensive approach. What has, what have been done in other studies is focusing a little bit on the fall in the education premium, some papers by Manacorda and others, and Gasparini and Kornia, they focus particularly on the skill premium. And there's a lot of country studies, again, also focusing on this type of price effects, the changes in the skill premium for Mexico, Chile, Panama, and several set of Latin American countries. More recently, Messina and Fernandez applied a broader framework also to understand the effects of experience premium in Argentina, Brazil, and Chile. And we really are closer to a paper by Cebedo and others, but in that paper they have a non-parametric approach of the compositions and we try to do is to follow more a framework from those of you who are familiar with the work by outdoor and cats in the U.S., we try to follow that framework for the analysis of earnings and equality in Latin America. So the contribution of this study, we believe, is that we try to get a little bit more into the details of the main determinants of labor income inequality. And to a lesser extent, we try to contrast this in the paper with the trends with other middle income and high income countries. And we present some stylized facts in terms of which percentage of such variance in earnings and equality we can explain from observables, skills, experience, gender, urban, rural residents. And we also tend to explain which part of that we are not able to explain with the data we have. And the importance of this study is that we have a regional approach with these 17 countries in the satellite data that I understand Garry's book also used, the satellite data that covers more than 90% of the total population in the region. And we take also a long-term perspective, departing from the early 90s, and actually the last version of the paper we tried to update it up to 2016. So if I don't have time to go in detail with the illustrations of each one of our main results, here's an overview of our findings. The main one is that as shown before, there is evidence of this trend reversal in labor income inequality after 2002 for 16 out of the 17 countries we have data for, which is supported first by a substantial expansion in the real early earnings at the bottom of the distribution. Secondly, through a steady decline in the education premium, driven by a larger growth in labor earnings among the on a skill and low skill, relatively to those with high school and college education. Also we find that this is a new result and is part of the contribution of the study and a steady fall in the experience premium in which most experienced workers have seen a reduction in almost half of their premium with respect to younger workers. On this one, we don't have an explanation of the channels but some hypothesis this has to do with skills of solicence and other things related to demand. I think Santiago Levy and Felipe also had this story in Latin America in which they show that there are a lot of demand factors that might be explaining why we observe this decline in education premium and possible experience premium by not creating enough good jobs. Also, we find small effects in the gender wage gap and we find a substantial effect in the urban rural earnings gap which declined during the 2000s which is correlated with this period in Latin America called the commodity boom in which was higher terms of trade particularly for South America. And then we find that there is a key role of unobservables in such a trench in which inequality which unfortunately with the data we have we cannot explain but this is our typically associated with efforts or with soft skills and with other information which we really don't have in this labor force service. So the framework we use is the one from cats, outer and cats and the mule and cats, Kearney and an outer in which basically we try to go through unpacking the effects and the drivers of earnings inequality through the observables and also try to have something to say in terms of the unobservables. And in general, as I mentioned before, some of the shortcomings of this paper is that we really don't get into the mechanisms at play that explains those effects through the education premium experience premium and so on about the other background papers of this regional story on which inequality actually discuss some of these effects in terms of relative supplies for the effects and I mean this idea of the paradox of progress in which you have more educated people joining the labor market and then there is not enough labor demand responding to such an increase in relative supplies that actually reduces the wage premium at the top of the skill distribution. Also, there are a few background papers dealing with topics related to minimum wages, which are relatively high in Brazil and Colombia and unionization factors and so on. But those are beyond of this study so I will not have a response for those type of questions. So what we do? So by basic definitions of the data and what assumptions of this study, we just run very basic Minsterian equations in a semi-parametric way in which we define multiple domain specifications for the following, early earnings of the main occupation of full-time workers between 15 and 65 years of age. Second, education is defined in three categories, college, high school and primary education. Experience refers to potential experience which is a drawback from the data we have but we believe it's a good proxy for experience that typically is your age minus six minus the years of education and we divide that in five groups, zero to five up to more than 31 years of experience and then gender and urban are defined as collected in the survey and then we assume that this function in the Minster is a linear function so we can talk about each result as the response to workers' characteristics directly. So a few of you probably saw this graph also in this Outdoor and Cuts paper for the US but just flip in the sense that in Latin America what we observe is that in the early period of the 90s of 2003-2004 there is some sort of stagnation of earnings. It doesn't matter in which part of the earnings distribution you were, either in the bottom 10%, 90% in the median, there was not a lot of movements on real, early earnings but after that and maybe associated to the commodity boom in terms of trade in South America there was an effect in which the bottom part of the earnings grew more than the top part of the earnings distribution and this was mostly driven by what happened in South America that again are the countries that experience better results in terms of trade driven again by the higher demand of commodities such as China, India and so on and this is what happened in Central America and Mexico really there was not a lot of movements in earnings if any, there were some negative shocks that affected that, certainly the issues of crime and violence in Central America and migration to the US and then the global financial crisis that affected a lot of Mexico particularly the top part of the earnings distribution like the percentile 90. This is more or less how the growing scale looks for earnings in which it was somewhat unequalizing in the first period and very progressive in the second period and now moving into the second main result that is the education premium in general the ratios of and the gray line is either the differences in which ratios either are observed or the results of the mincerian equation it doesn't matter but in general there was a decline of about 25% in the education premium between college and primary and I believe also in the education premium between high school and primary that is reflected in this wage index for high school and college with respect to primary school this is just to show the strong association that exists between the decline in labor income inequality and education premium we try to include information from other countries including Russia from the Russia monitoring surveys from South Africa from Turkey and from the US hopefully Haroun can tell us what happened with South Africa that is a big outlier here in which there was an increase in earnings inequality that declined in the education premium and that one I don't know much this is one of the most interesting results of the paper that is the one related to the experience premium in which we really got from the early 90s there has been a reduction of almost in half on the experience premium with respect to the benchmark zero to five years of experience with respect to every other level of experience premium unfortunately we don't provide an explanation of what is happening here there is some hypothesis related to automation related to this routine manual and abstract tasks and so on but we have not been able to respond to this let me move quickly into the results for the gender gap and the urban rural gap the gender gap actually was caught in Latin America in the previous period and certainly a little bit more in the second period of analysis while there was a significant decline in the urban rural earnings gap there is a very good paper by Leo Gasparini in Argentina presenting precisely this urban rural gap and associating this to the commodity boom and how that benefited earnings in rural areas finally the topic related to the role of unobservables in the data to explain this wage variation so basically it doesn't matter basically the period we cannot explain about 30% of wage inequality with information on education, experience gender and location and one of our colleagues at the bank through a more sophisticated model a more complete model in which included interactions in all these characteristics and even a sector of activity and occupation and the most she could get out of observables was about 50% of the explanation which really leave us still an important task to understand which are the other factors that explain these declining wage inequality this is the summary of results the relevant columns are the last two which present the decline in each one of these relative returns either in education potential experience, gender and area of residence and what we can observe is that the annual rate has a negative which is associated with the declining labor earnings and particularly the last column is the one in which since 2003 such a decline has been larger so just as a matter of conclusion again we believe our main result is the one related to experience the experience premium which we certainly would like to explore further and also as I would like to invite you to take a look at the other background papers of the wage inequality report which presents a country case studies with some of these dealing with some of these channels to supply the manner of institutional factors and certainly there is a research agenda for us in terms of thinking how other factors such as self skills efforts and others are also important in terms of defining the changes in earnings inequality