 Right, so my name is Hayford from Ghana. My question goes to the second presenter. I guess that's Soya. Sorry, Dante, the Chile presentation. I want to find, I'm a bit curious about the curriculum that I used in the different, you mentioned the public, the virtual section, and then also you also mentioned the private page. I want to know whether it's the same curriculum because you are comparing performance. Is it the same curriculum that is used across board? Then you also indicate that you have data on family characteristics, family background. And I want to know who are those who actually send the awards to which of the categories of the schools that you're looking at. So is it the people who work in a private sector who prefer to send the awards to the private page or you have a mixture of those depending on the region someone may find his or herself in? Hi. I had questions for the last presentation. I was not clear on the interpretation of the sex of the child variable. It was negative in the first regressions, which meant that boys gained less. So if you could clarify that. So how do you, all you said in the last slide that you plan to differentiate between the impact of user fees versus the impact of, say, in a society if the importance of education has increased and hence people are sending children to schools more? So as well as there's also impact of having more schools in the neighborhood because government programs might have increased the schools. So how do you differentiate between, say, these three impacts or other impacts from the user fees impact? And lastly, do you have a sense of does economic status of the household have an impact on your results? Very quick questions. So first question is to Liliana. You didn't mention the problem of tax evasion. So is that something that seriously affects the quality of your data in this, in particular in the upper centile and the top centile? Then one question to Danta. I didn't very well understood the thing on the test scores. You have panel data and then you say that you can identify some outreach change in test scores on the earnings. Or I wasn't sure about that. I mean, I think I saw that this is something which is measured once in your educational life or so, but it seems that you have repeated observations on test scores. So maybe you can just explain that a little bit. Then to Somya. You mentioned that one implication that we could have a conditional cash transfer. But if we talk about health shocks, maybe it's even more obvious to think about health insurance. So I wonder whether you consider this as an option for India. And then a question to Kenneth. At the very end, you discussed the effect of having or not having biological parents. And you find this negative effect of not having the biological parents. But I wonder whether the effect that you observe is not more the effect of the shock that led to the status of an orphan than the fact that now you live with your non-biological parents. Because you say it's mainly a channel that is through the transmission of education to these non-biological parents to the children. But it could also be still the effect of the shock that led to this status. Thank you. Any more questions? Yes? Hi, I'm Tuli from Alta University and UNU-Wider. I have a question to Liliana. I was also wondering if you could kind of discuss a little bit like what kind of people, besides tax evasion, might not be covered in your tax filers. And also if you could, you were covering the results quite quickly, so maybe you could kind of reiterate on what was the main result regarding the mobility among the top. And perhaps some explanations, what explains, and in the graph that you showed, there was some, a dip in the probabilities to stay in there around, was it 2007 or something? So maybe you have some idea of what is happening there around in the top income mobility. Thanks. Okay, maybe let's have the response, and then we'll go to another round. Can I ask a question? You are there also to ask a question. Okay, maybe let's have the quicker response, and then you'll be the first one to ask the question. So Liliana, you go first. Okay. First, please, about tax evasion. For the top, when we come from the top of top series, we come from, by control variable of total income of that population to avoid this problem. So series for the top are related to a total full population. Total full income, the methodology was proposed by Piketty. It's not, top income series are not constructed relative to the tax filing population, are constructed relative to a total income controlled total population. But I am not able to use, to follow the same methodology for the entire distribution, tax distribution, because as I said, if I use control variables for the total income and total population, I only able to capture, for instance, in 2011, 25% of potential tax units. So for the total, I don't have problem of evasion. But then for when I use tax, the entire tax database, I focus most in the middle of the distribution because I know that the bottom, there will be problems of evasion for the bottom of the distribution. But I think that tax data, tax data provides a better picture of the middle and on the top of the distribution because we have all former employers who are there or people who are required to keep an account in books and self-employment are there. So I think that provides a better picture of the middle and on the top of the distribution. For the bottom, that's why I'm not making an allies for the bottom because I think that for the bottom it's better works with the household service. And for the second question, very quickly the result is that we have mobility at the top is very low. The probability of a stain of people who are in the top is almost 60% at the end of the period. For the middle of the distribution, we find that people are moving up over different periods of time. Almost 80% of people who are in the fourth and this fourth, this fifth are moving to a higher income groups. Also for the top, when people who are in the top are more likely to drop to the top 5% than to drop to the bottom 90, 95%. So if people in the top move, they stay on the top. And then the probability of decline. I actually analyzed this subject because in that year there was, there will be a tax reform. A tax reform, so maybe this is a response to this tax reform. Maybe, so I'm not sure of that. And then finally I find that the people who are, who holds an academic degree are more likely to experience other movements than don't know movements. Especially those people who are in the middle of the distribution and who holds an academic degree. Yes. So I didn't present this because it's in Spanish but it's gonna be useful in order to answer I think both of the questions that I received. So the main story of this paper is you have, I mean, we have many test scores for children, okay? Who are using this one? This one here, all right? So we are able to control using this test score, the future of these guys, okay? Connecting the data with level-mounted outputs. But the story is the following. The new one is, I mean, this factual test score that's between public school, public school, and private high school. So there is a significant gap here. You follow the same guy over time. The whole educational process, the gas remains, okay? So we are using this and we are using this but now we are collecting all the test scores in order to monitor the whole life cycle in the educational system and also in the data format. The point of the paper is that the inequality in education is transmitted to the inequality in, okay? There's no significant gap between people attending public school and audio school but there is a huge gap between these two guys and this group, you see? And the final important issue is that when you compare child when they're born, they are quite similar. So now the question is why people do not, you know, place their children in this type of school because they are so expensive. So you are rich, you are going to follow this path of rich, you are going to be working in the future but if you are poor, you are going to follow this path. So what it remains? That's the big question and issue. So it's not related to curriculum, okay? I mean curriculum is quite the same in all schools. I mean there is a minimum curriculum in the country and every school has to, you know, educate the kids with this kind of curriculum. The main difference are infrastructure, the quality of the teachers and social capital of the parents. So that's explaining the gap of the time. Right? You're not using any within variation in schools. That's what I understood from the presentation. No, no, you see it in good places. Yeah, okay, thank you. So may I? Yeah. The question is on why not explore the options of health insurance. Actually, under British government has a health insurance scheme. It covers almost 80% of the population and even in my survey, 80% of the households are covered under the state health insurance scheme. In my previous paper, I found that the health insurance scheme does not have a significant effect in reducing the coping strategies like borrowing, sale of assets and so on. I think why this is mainly happening is maybe the indirect costs are more than the direct costs. In that case, when there is a loss of income, even if there is health insurance, the families are anyway going to send the children to school. So that is what maybe is driving my results. I had also used health insurance as an explanatory variable. But I did not find anything. Okay, Dennis. Okay, first with respect to the effects of the child. So it's true that for this particular model and for this age group, boys seems to have less education than girls. This sometimes changed over our models. For example, if you looked at the impact of the father's education or of the household heads education, so this is not consistent across all the samples, but I forgot to mention this. And we also looked at the socioeconomic status of the household. Unfortunately, the DHS does not contain any information on income, but we looked at assets. But this is, of course, also correlated with the education of the mothers and fathers. So in this model, we wanted to show you the model without the socioeconomic status, but only the pure effect of the sole effect of the mothers education without controlling for these other effects. And could you please repeat your second question again, I did not get it fully. So the question is, there are several impacts on why the child's education would have improved. It might be that the society, there's more awareness in the society about education or there's more supply of schools. Or one of the reasons which you are exploring is user fees. So how do you plan to differentiate between these impacts? Okay, good question. So far we bypassed all these issues. So we planned to also study the impact of the introduction of user fees on enrollment rates. However, then we need to look at different age groups, of course, but we have no country information on all these other effects since we are doing cross-country comparison. So it's a good point, but we bypassed this at the moment and that we should take this into account in our discussion, thanks. And I could just to clarify, so we found a positive impact of the non-biological parents on education, but it was lower than for the biological parents. It was not negative. And the reason that it is negative of course can be simply due to the shock children lost their parents. However, this is a bit mitigated because we looked at children at older age groups, so 15 to 18. And to take your point into account, we should look at what happens actually if children lose their parents with respect to enrollment of children. So if we find, we should find a significant impact on this, then we can take this discussion into account when interpreting these results of transmission of these non-biological education and non-biological parents' education on non-biological children. But you are right. No, the first time we would try, I mean, first of all, was to be direct but just for any education, we're not even interested in this transmission, it's lower. In that would be the effect of shock, not so much transmission to these non-biological children. Exactly, but I guess that the effect should be much higher for enrollment than for actually achievement for these older age groups. But of course, drops in enrollment rates are then translated into these limited achievements. Thank you, we have 10 minutes, we can have another short round of questions. Dante, you wanted to ask your colleagues a question, we'll start with you. I have a question for the audience. Well, I mean, when we compare, as you mentioned, the probability of staying in the top 1% is about 60%. How comparable is this one with the other countries in which we have data on that? And the second question is again, with this idea of mobility in the middle class, according to the World Bank in Latin America, middle class is highly vulnerable in the sense that we can move forward, as you mentioned in your presentation, but also when downwards. So there is a lot of flotation that according to the World Bank in Latin America, so how you reconcile your data with this evidence from the World Bank. For the last presentation, I have one conceptual question and one empirical one. The theoretical one is when we talk about, or we discuss about the intergenerational transmission of education, we might think not only in terms of quantity of education or year of education, but also the quality of such education. So your paper is just about quantity, right? And I believe that for the future, I guess, not only quantity is relevant, but also the quality. For instance, six years of education in Finland is equivalent to 10 years of education in Chile. So that gap is quite significant. So that's the first one. And the second, related to your question as well, I think you mentioned that there is a positive relationship between mother education and child education, right? And then free education is positive related with that. And then the interaction is negative. So can you elaborate a little bit more about that? I mean, because you are controlling for fixed effect, for control level and everything, so I cannot follow how come you get this negative interaction, coefficient. My question is from the, oh, okay, you have a question. I have a comment. Dante, actually, my comment is he showed that, he took the sample and he showed that there was a lot of attrition due to zero data. So, and then you were comparing the means before the attrition and after the attrition. So in that case, I would suggest you to do a regression and show that there are certain characteristics of these are not being, because that gives credibility to your story. Let's go. One question from the floor. Okay, so, okay, you try to respond to some of the questions raised, and then. Related to the probability of a saying at the top of the income distribution, there is, there are not empirical evidence for South American countries. At least, you know, first, this literature on top income is very recent, but I find more or less the same results for the probability of a saying at the top for countries like France, Canada, and Germany. In France and Canada and Germany, the probability of a saying at the top of the distribution is also almost 60% after one year after. So, we find more or less the same results. I am not able to make a cooperation to South American countries because there are not empirical evidence on top income probabilities. And then, concerning the vulnerability of the middle class, in the paper, in the last part of the paper, I made an analysis regarding tax data and household survey. And I compute in absolute terms the middle class using tax data. So, for instance, if I take people who hold an income of 10, between $10 and $50, which is actually what is proposed by the World Bank to people who hold this amount of income, are considered belonging to the middle class. And if I made the same thing to the database, it corresponds more or less to the decimal three, to the decimal nine. So, in my database, people who hold an income of $10 and $50 stays between the decimal three and the decimal 10. So, that's why I always, that's why I focus on the middle decimal. So, I can identify middle class in absolute terms also. Okay. Thanks. I think you are totally right with respect to the comment or question on the quantity, whether it's quality issue. We really would love to have information on quality of schooling for cross countries in Africa and other developing regions. But in this data set, unfortunately there's no information on quality of schooling, but it's a valid point. It's discussed in the literature. It's discussed in our paper and it's also discussed whenever someone uses the DHS to analyze dynamics in education in developing countries. So, I mean, there are, fortunately some countries that now have made information on test scores in schooling available. So Kenya, for example, and it's a good point to take these data on board and try for a couple of country cases to find whether there's an effect on intergenerational transmission on mother's education or actual outcomes with respect to quality of education. But so far we are not able to do it because of data limitations. And regarding your second point with the negative effect of the interaction term. So we found an impact of 0.5 years of schooling for mothers with no education. And then with each additional year of schooling of the mother with respect to none, this drops by 0.08. So, first of all, well-educated parents are not affected by school fees. And now this means that one school fees are introduced, less educated parents benefit more from the introduction of these primary education than better educated parents. However, the overall effect remains positive. And so even for mothers with 12 years of education, the total effect with free primary education and the interaction term is about four years of schooling. So we did find that we can improve the distribution of education. And we also find that even for mothers that have a lot of years of schooling, we can be quite sure that there is an effect, a positive effect of free primary education at all, so that we can argue that there is a widespread political support when it comes to the discussion of free primary education. Okay. And the one who was the enemy on the comment... No, I fully agree with you. I mean, you're absolutely right. It wasn't in the presentation, but it's in the paper. Okay. Thank you very much. We have come to the end of the session with a two or three minutes bonus. So allow me to thank all the presenters and yourself, of course, for active participation. So let's clap.