 Okay, so we have up to half an hour for the questions and discussion. The job of a good chair is to get you to lunch on time, even maybe a bit early. But we have plenty of time for questions, so let me invite people to ask questions. I'm asked by the video, by the person taking the video, and you're asking questions. Please stand up while you ask the question, because you're going to be videoed. Okay, so let's have some questions, please. Thanks, Perronos from SIDA. Can you stand up? Stand up, also. Yeah. Thanks, and Perronos from SIDA. I have a comment, I suppose, and a question to John Landon Lane on study on Tajikistan. Now, as you mentioned, Tajikistan is in many ways atypical because of its extreme dependence on overseas migration and on remittances as a source of household income. And the period you covered, 2007 to 2011, as you pointed out, in the midst of it, 2009, you have the financial crisis, which hit Tajikistan particularly severely through one channel, and that was through migration. So Russia sent home Tajik workers on mass, and that was the main channel of transmission of the crisis impact. I mean, that was the main channel for the crisis impact. So remittances fell, and there were lots of people who were sent back during that period, who then later returned. So I have one question or comment, I don't know, on your definition of vulnerability. Now, I understand that you define vulnerability as those who are above the poverty line, but not more than twice the poverty line. That captures one important dimension of vulnerability, namely the ability to absorb economic shocks. But there is also another dimension to vulnerability, and that is exposure to shock. And there you have, which also can vary greatly. I think, particularly in Tajikistan, and I think this period captures that very well, you have two quite distinct categories. You have households which are dependent on sending household members abroad. There's Russia to work, and you have those who do not have migrant workers abroad. Now those with migrant workers abroad are arguably a lot more exposed to shocks, external shocks. There are health shocks and other shocks, I know, but in this case, external shocks. Because they make a huge investment to send somebody abroad, and they do not know if this person is going to find a worker or not, if he's going to be expelled or not. They are much, much more exposed, as the crisis in 2007 or 2009 showed. So how do you factor that in any way, because to me, vulnerability is both exposure and capability to cope with shocks. And then two questions on methodology, which is also related to the fact that such a large part of the population is not in the country, but works abroad. One is, do you use, is it the factor population you work with, or the shure population? So who do you include in the household? Do you include migrant workers abroad, or are they left out as soon as they leave the country? It makes a difference, it makes quite a big difference. In this case also, because you had such a large number of migrants being more or less forcefully repatriated during the economic crisis. So the size of households, the size of households were larger in 2009 than it would have been earlier because husbands returned home, or were sent home. So I think that matters somehow, you need to factor it in, because you're not working with the same population during all these three periods. And last on remittances. I don't know how, remittances are part of the question, I mean are part of the money that migrant workers bring back. But much of the money, and certainly in the case of Tajikistan, where migration is typically not more than, I mean people go back every year or so, much of the money is brought back in person, it's not remitted. So presumably you would have had an increase in income as migrants come back in 2009, because they don't, some of the money is sent obviously, but a lot of it is actually brought back by the person, it's accumulated and brought back. And that means that you will have lots of variations of time on, if you measure incomes. Thank you. Thank you, there were several questions, but I'm going to take three or four questions and then give the presenters time to respond. So let's have further questions, please. Yes sir. Okay, thanks for the presentation, my first comment goes to the first presenter, that is the Cameroon case. I remember you were talking about three sectors, the private, public and then the informal sector. If you look at the sub-Saharan African sectoral distribution, one key component is the civil society organizations which can include both international, local civil society organizations that employ to some extent the higher quality type of the educated. And so in terms of the situation of workers between the private, public and then the informal sector, I think that is one of the areas that need to be considered. I don't know whether the data didn't capture that, but in terms of as people get more education, where you can find the civil society organizations or the international organizations are key in developing countries where they are educated, always absorb into those sectors. The other aspect is that in terms of the educational dynamics as more people are getting more education, it's also the quality issues in terms of accessibility, okay, is the educational expansion including the quality issues or is that access has been given to a lot more people to be in school in spite of the quality concentrations. My other comment that goes to the second one, John, I saw overlapping periods for the periods and also for the categories that you have. You have one and then 1.2 and then the next one also starts from 1.2. In the context of making comparison, I think to become problematic once you have the periods overlapping, okay, the upper band view of one category is the lower band view of the next one. In that context, I didn't see the clarity in that band in terms of the categories. Thank you. I saw that on slide 21 or 22 also in terms of the distribution of the categories. Thank you. Thank you very much. Further questions? Anyone else like to ask a question at the moment? Okay, well then let me give John first then an opportunity to respond. He had quite a few questions in the first. So first of all, thank you very much for those questions. I'll do the last question first and then I'll get on to the first question. There is no overlapping periods or categories that may have been the, I had a difficulty getting all the information into the graph but the first transition is from 2007 to 2009 and then it's from 2009 to 2011 so I don't think we have those problems and certainly for the income classes, the cutoff is one. If you have a relative income equal to the poverty line, you are defined to be below the poverty line so that's how we dealt with people right on the boundary. So I'll make sure in the next presentation I'll use the appropriate notation so it's obvious. Now so the first question is clearly, so now breakdown of labor in this paper, I'm not the data expert, that's Senya who actually put together the questionnaires for the 2011 survey and carried out the survey. So I apologize to her if my answer to you is not as accurate as it should be. The first question was about definition of vulnerability. So I apologize for if it made it sound like I was defining people to be vulnerable if they were arbitrarily below two times the poverty line. The point I was trying to make is that we are trying to make as everybody is vulnerable and that's why our vulnerability measure included probabilities for all initial income classes and their probability of falling below poverty and so I agree completely with what you said. And what I tried to show in one of those last slides was actually if you were going to use an arbitrary cutoff it looks like it should have been three and not two in that first period but even though actually it could have been everybody because everybody had a probability of at least point two of falling below the poverty line. If anybody was in the previous session where a similar paper was discussed they were using five percent as the cutoff for vulnerable. So I just want to make clear adding migrants since this is the idea of covariates which is we've worked out the theory and we just have to do the results yet but you're right we need to put in what we need to do is identify who's vulnerable and it's not necessarily some household a household with income of two point five times the probability line might be less vulnerable than somebody who's five times the poverty line based on individual characteristics and that's what needs to be in the analysis and I agree with you completely. So the interesting thing about the migrants and their remittances which you bring up is we if I didn't report this result but when we look at households with migrants current migrants so there are a lot of questions in the survey so you get asked is there a current migrant is there a recently returned migrants so we know which households have recently returned migrants we know which are current migrants. The remittance question that as I understand it and this is where my knowledge might not be perfect here is that you're asked how much money this period that the household received due to income earned overseas and that's declared as remittances so it's actually this income that came back with people or it could have been sent that is all accumulated into remittances so what we found was households who had remittances this current period were less vulnerable than those who didn't. Now we also found that households who have a current migrant that result they were not any less vulnerable than households without migrants so it's clear that you're right that the house the individuals who went overseas weren't sending money back in real time they were accumulating assets and bringing them back when they came back. So and again it's we were we didn't make a stand on what was the cause of the crisis and what caused the the vulnerability to be so high in the 2007 to 2009 transition but I think your explanation actually is very good I think that's that's the right explanation. So again I think we can factor all this in we are going to add covariates it's took a while but now it's relatively once I worked out what to do it's relatively easy to do that and we can start looking at a finer definition I think we need to have a finer definition of vulnerable it's not just your randomly less than twice the poverty line that's not the appropriate way to think about this I agree thank you. Right you know when you are carrying out let's say a study you know decomposing the various labor market characteristics would depend on how far you want to go but basically when you just want to consider manageable categories you can have the public private and informal sector then controlling as we did in this for small scale agriculture but if you wanted to disaggregate if there was reason to do that it is possible to do that but in this particular case for instance I think at something like international organization those working in international organizations were assimilated into the public sector while things like civil society organizations were simulated you know those that have formal structures were simulated in the private sector but of course if there's any reason why you could decompose you know more than the four categories that you are considering it is okay thank you thank you Francis okay we have time for more questions so let me see if other people would like to ask questions in terms of your that the final slide on basically the first presenter that I'm talking about the one the education the ways differential in the middle of that third fourth quanta that the impact of the education were high in terms of ways of that is it because of the sectoral this thing you have informal formal but if there was agriculture manufacturing or you know service sector then of where you know you have that you you'll you'll find more impact of the education in terms of wage difference yeah so so so that because the considering the only five period like there could be the least impact on the agriculture manufacturing but expect more in the service sector where the income you know its impact would be high and that could be the in build only in your your formal model that you use so I was thinking whether you know that are relevant all it's maybe more good your your like informal sector is a very big when it could be agriculture other thing but if you have a standard you know like education basically it's better on the service manufacturing and the what is agriculture then you see where the education basically the wage difference would be very high its impact perhaps as a clarification perhaps you should just answer yeah well I I hope that I understood the question because you know what we did as I said we face of all things that the the sectors that we included in our wage equation where the private sector the public sector private sector informal sector well you know controlling for the small scale agriculture you know as the then the our analysis were done according to percentile selected percentiles or quintiles you know from some 5% right up to the 90 feet percentile and so I don't know because what they do what we included in the structural equation where the inverse means ratios that we generated from the multinomial probit of the first stage equation so we included them in the you know in the second stage equation which is our structural equation but a structural equation was estimated overall and according to percentiles yeah so and of course what we presented was simply the effects of education we didn't consider you know we left out order control variables and all they like thank you for the questions yes yes no we're not yet I'm actually I have a question for Patricia I was I was very interested in in this definition of inequality for education attainment and I may not have understand understood everything completely so my understanding of this of this measure was if you look at if you break characteristics individuals down into different characteristics it's it's how it's based on their test scores and a more unequal distribution as if the distribution of test scores so it's sort of it's test score based so the my question is if you did the inequality measure based on just access to different types of quality of schooling is that are you going to get a different ordinal ranking sorry can I also ask patrice here what the what the sample is for a piece of what what sort of school is this public schools or is this any type of school okay it's small one yeah I was just doubting because the way I see you characterizing your inequality measure it appeared to be something like a coefficient of determination does our square so I was doubting whether our square can effectively be considered a measure of inequality for the last one and start from that one yes it's the variance and could be considered as a measure of inequality and the idea of you using that particular measure with this particular outcome with this particular test score is not mine is just because it has been proved that the ranking of country could change if you use other inequality index when we consider the press standardize it or post standardize a distribution so many due to the standardization procedure that I do not explain in something carried on by the UECD Ferreira and Gignu published this paper when they show that the proper index to use to measure inequality is the variance and when you have a linear approximation of the test score production function you can simply look at the air square for the your question was if the kind of schools which are yeah the sample is well there is this two-stage sampling procedures so all the schools are involved public but public and private school then there are also variable on these characteristics about school so there are both as for the quality of the school system well first if we consider access to education maybe I think it would be more correct not to look at inequality of opportunity but simply to inequality of outcome because there you cannot I think in my opinion is that you cannot really consider individual responsible or not if they have access to education and for the quality of school but still don't know how really to measure in a correct way the quality of an institution because often quality of institution is measured by the performance of students in this test so it's a kind of efficient circle thank you the concept of access is rather different if it's a private school though compared to a public school I yeah I don't have the distinction matters anyway let's see if there's more let's see if there's more questions and anyone else like to like to ask questions or you want an early lunch okay it's 22 minutes past so I think I think then sometimes some of the best interactions come from one-to-one conversations the speakers are up here at the front please come up to ask them individual questions if you want to even if you do that you still may be able to get to lunch early so thank you very much to the three speakers for presenting to us and for keeping to time and thank you very much for you all for attending and I wish you all a nice lunch thank you very much