 This morning, as the Chairman said, I'm Liliana Canu from the University of Toulouse. And today I'm very glad to present you a part of my research, which is actually related with income inequality or with a special interest in South American countries. And today I will present a paper which is entitled, Income Mobility and Equity, using individual, using data from income tax reader. So the presentation will be as follows. In the first part, I will present the main goals and motivation for doing this paper. Then I will make a very brief literature review. And then in the second part, I will present the data methodology, the main funding, which is the most important part, and then the concluding remarks. And so while many recent studies have recently documented the decline of income inequality in most Latin American countries, there has been less attention to the analysis of mobility in this region. So with this paper, we analyzed income mobility and inequality from 2004 to 2011, using data from tax readers. So we study whether the evolution of top income shares has been accompanied or not by an increase or an increase of mobility at the very top of the distribution. Second, we study whether there is a surge of an Ecuadorian middle class. And there, we analyze the factor associated with income mobility over the three years of the period, which are the motivation for doing this paper. I have two main motivations. The first motivation is based on the growing interest in the study of income inequality at the top of the distribution. So it started with the seminal work of Piketty in 2001, 2003, how many research have constructed series of top income shares using data from tax returns and national accounts. And today, following the methodology proposed by Piketty and by Adkinson also. And today we have a series. Series has been constructed in more than 25 countries. And as we saw yesterday, most of the series are available in the world of income database. So here is Ecuador and we are, we follow this growing literature and we construct top income series using this methodology. And the second motivation is based in the study of intra-generational mobility. A recent economic report from the World Bank documented that almost 30% of Latin American individuals had experienced change in their economic status over the last years. And most of these movements, there are upward movements. For Ecuador, estimates of income mobility are scarce and mostly do because we don't have the appropriate data for doing this kind of study. And it's worth noting that the World Bank in the recent economic report, he worked with synthetic panels because panel data is not so easy to find in most countries. And we take advantage, of our tax database, which is a panel database, and we study mobility for the entire tax filing population. So the third part of the literature review, I will just very quickly. There are sociological and economic approach of mobility. In this paper, I will focus exclusively in an economic approach. Literature of income mobility is very vast and doesn't provide an harmonized framework of analysis because the world mobility may connote different ideas to different researchers. So an important review of our conceptual we can find in fields, addings on Jenkins, field of vocation, code, et cetera. With the language, three main definitions of mobility. Mobility as movement, mobility as time independence. So is the present income more or less the terminal of a future income? Mobility as a quali is of long term. If there are chains of income at some point of time could be an influence income in the future and mobility as moving, people moving up or moving down in the income distribution. Besides, there are two main dimensions of mobility. So mobility, intra-generational mobility and inter-generational mobility. Inter-generational mobility when we follow the same unit over time. We analyze the income dynamics of the same unit and inter-generational mobility as more related to the study of generational, for instance, as child's income more or less related with parents. And so because we are interested in mobility at the top of the distribution, there are, I made a little research of literature review on top income. And we don't have so many studies on top income mobility because first I think that is a recent literature. And second, data, mostly tax return data is difficult to obtain and to make this kind of a studies. So starting in the intra-generational mobility, we have a studies made by Auten and Guy for the United States. Copscute also for United States. I signed something for Canada, Landais for France and inter-generational mobility, Shaggy in the United States and George Lantz for Sweden. Most of the results of those studies suggest that the evolution of top incomes have not been accompanied by an increase in mobility at the very top of the distribution. On average, almost 60% dependent on the country of people who we belong to the top distribution reminds at the top of the distribution after different periods of time. On average, 60% in most countries. And so based on this literature review, I mobilized four hypotheses. So the first hypothesis is that income mobility trend in Ecuador has not been accompanied by an increase of mobility at the very top. The second hypothesis that I mobilized is that there is a high degree of upward mobility as we saw in most Latin American countries over the past years. Upward mobility is mainly explained by the initial position in the income distribution and the upward economic effect of education on income mobility should be more or as important as the initial position. And so, I started the second part of the presentation. So I will present the data, methodology and the main results. We work with micro data from income tax returns from 2004 to 2011. We have the universe of tax fighters. This is a data that is produced by the Ecuadorian internal service. So it's the SRI for the Spanish acronym. We have information for every tax filer. We have information on labor income, on capital income, returns to capital, other kind of income. This mainly comes for three different tax forms. The first is 100 as even, which presents the information on salaries and wage. The second form which presents the information of wage, salaries, returns to capital from self-employment individuals. And we have also this 102. We contains information for individuals, also for labor and capital and for individuals who require to keep an account in books. For instance, in 2011, we worked with 2.3 million of tax filers. The unit of observation, the tax units in these countries, individual as in most South American countries. And of course, we are working with anonymous data. As we saw yesterday, there are advantage and disadvantage of working with tax data. The main advantage when we have a special interest on the top of the distribution is that tax data provide a better picture this part of the distribution than most hospital surveys. We have a composition of incomes where we can make a clear analysis of income from labor, from capital, from other sources of income. And we are working with a real panel database. There is advantage, of course. We have a problem of evasion, elution and tax reforms over time can change the definition of income. Also, we work with information on individual characteristics of sub tax filers from the Ecuadorian Civil Registry. For instance, we got some information on age, on gender, on marital status, level of education and the region of origin. We have that information for tax filers from 2008. And we were able to merge this information from the three last years of the period. So in the last part of my analysis when I am studying the factor related with income mobility, I use the disinformation. Those are my control variables. Sixth exploratory control variables, so the initial position in the income distribution, age, gender, marital status, level of education and geographical region is worth noting that we have very detailed information of level of education. But I just separate of individual with high school and more and less than high school. And so, the methodology. I organize the methodology in three parts. In the first part, I construct top income shares following the methodology proposed by PK, the top income shares, but related. So the amount of individual tax returns, which is the numerator of my share, to a comparable control for total population, which is the denominator of the share. The income definition is before personal income tax, unemployed payroll tax, as is usually proposed by top income literature. I constructed first the top 1% and then I constructed a series for smaller fractals. So the top 0.5%, top 0.1%, et cetera. To control total income and total population, I work with household surveys. I construct a total income from Ecuadorian household surveys, which is more or less 65% of GDP at the end and I work with population age of 20 and more. Those are my control variants. Normally, for people who are familiar with top income literature, control total income comes from national accounts. In a previous working paper, we work with national accounts to construct control variables. And in this occasion, I work with household surveys to construct those variables. Once we have construct top income series, I analyze the persistence of top individuals. So I compute the probability that I remain in the top after one, two, three years after. And then using transition matrix, I examine, I am a study the movement into the economic elite. Second, I made the same thing for the tar tax filing population. It's worth noting that here, I have a methodological difference in the second part. In the second part, I don't use control variables. All transition matrix are relative to the tax filing population. And this is mainly for one reason. If I use control variable for the entire tax filing population, I am only able to capture, for instance, in 2011, almost 25% of the population, of the total population, potential population. So for the top, I use control variables as proposed by Picetti. And for the entire population, I don't use control variables. So movements are related to the tax filing population. I will do it quickly. And factor to study mobility, I use three models, first accounting procedure, then a multinomial logit model, and then a generalized order logit model. Logit model for 2008 and 2011, we have 1.2 million to 2.3 million. And so we are able to control by initial position for people who are present in both years of about 1.4 million of observation. We have control variables, information of control variables for about 7,000 observations. In the third part, so when I analyze factor related with mobility, I have a limit, a methodological limitation, how many centers, how many income-saving people have moved. So I use another multinomial logit model to assess our downward movement of at least 10 centers. And then I make a transformation, a logit transformation of my dependent variable to measure the change in the person's position of individuals. So the main point, we go quickly. So we have construct top income series. For instance, people need an income of almost 70,000 to belong to the top 1%. In the top 0.001%, we have 94 units. Those people, to belong to the top 0.01%, people need about 2 million of income, 2 million dollars of income. Those amounts are expressed in dollars. We have the series of top 0.01, consistent with a prior empirical evidence based on household service. We find also a declining trend in income concentration. In 2011, almost 20% of total income goes to the top 1% of the population. It's consistent with some Latin American countries for which we have information. For instance, in Colombia, I think that is 25%. And Uruguay, 12%, and now in April, we compute almost 20%. Then I just make, I decompose top in smaller fractals, and then I compute the probability of staying in the top 1%. On average, 55% is the probability to remain at the top of the distribution. After one year, 56% after two years, and almost 50% after three years of the top one. The same thing for the top 0.01%, the probabilities on average is 30%, 20% and 15% after one, two, or three years. We saw that probability of staying of the top 0.01% is a smaller than the probability of the top 1%, which is normal because maybe those people are very transient over time. So I don't know if it's obvious, so for that I was wondering if people who are in the top are more likely to move among the top or more likely to drop to the bottom 99% or 95%. So to know that, I'm studying movements between the economic elite. So the diagonal entries of my transition matrix show the stay in groups. The rows correspond to the top percentile of origin and the columns to the top percentile of destination. So we can see that for instance the top one, people in the top one almost 82% had moved by 2011. Most of this movement, there is another movement, people, the results is people in the top are more likely to remain on the top and if they are movements, they are most moving until the top 5%. People who are in the top and who drop to the bottom 95% is almost 20% and there's more practice. Factors associate, I will be very moving, counting, multi-generalize with the three methods. I find that people who are in the middle of the distribution experience another movement. People who are in the middle are movement are almost 75% moved to an upper desi by 2011. The same thing when I use the multinomial model and the same thing when I use generalized ordered logic model. Which is very interesting is that people who have level and who holds a high school diploma, a high school degree are more likely to experience upward movement. So education plays a key role in income mobility. And then the last one is we have, we analyze the probability of experiencing an upward movement and downward movement relative to not moving. So the same thing, people who have an education are more likely to experience an upward movement. And we model, when we model central effects, we saw that the initial position, people who are in the first and the second and the third and the fourth desil are more likely to experience a movement of about 45 centile, sorry, centile, 25 centile, et cetera. When I put all the control variables, people who have an education, a level of education are more likely to experience an upward movement of nine centile. And which is more important, woman, which is very important, a woman who have a high school degree are more likely to experience an upward movement of about 11 centile. Thank you very much.