 Well now we have ample time for questions, so I think what we'll do is to take four questions and then Let the speakers respond to them There's one person over there Do we have somebody with the microphone? So there's somebody over here Does somebody have their hand up over there? Oh with Francois Hello They gave me the mic I'll go ahead. Thanks. This is on the presentation by Arjun on a Consumption and income inequality The first set of issues would be why the difference Between consumption and income inequality is it taxation is it savings? Is it remittances? Is it credit and so on and so forth just to explain that that difference But even more critically for me would be the issue about The utility of either of those categories for policy Which which would be more useful for policy makers is it income or Consumption or expenditure in inequality And then lastly the issue about consumption and life expectancy This is just to pose the question whether There is a certain level of consumption that might in fact be negative for life expectancy Depending on what people consume. Thanks Okay, thanks very much Francois Yes, thank you very much for the Very nice presentations I would like to say a few words about the two presentations on global inequality The first thing that I I was very Pleased to see that there was so much interest for this issue of global inequality And this suggests to me that maybe it would be a good thing to Try to organize some meeting where people interested in that issue could exchange views Because there are many different issues there And we are very far from a complete agreement on what is going on And for example today we had an example of A study which was based on survey means Versus a study which is based on GDP per capita And we know that this has been one of the most important discussion In in this area As a matter of fact, there is a recent paper by anon and sigil in the handbook of income distribution I was mentioning yesterday Which is Discussing that again, which is proposing a new Estimate of global inequality There are now estimates new estimates by milanovic. We'd go to 2010 In a forthcoming book. I have some estimates which are also going until 2010 And I think it would be a good thing to try to make sure that everybody is more or less on the same page And that when we say That inequality is going down In the world, there is a truly an agreement. So Really, this is an invitation maybe to wider to organize such Such a meeting One remark only on The presentation by By wider Which is on this issue of the fact that lately It would seem that the within country inequality when you decompose using the tile number two Is going down It seems to me that it is very much due to the fact that you are using the tile number two Which is a mean logarithmic deviation. If you were using tile number one Which as a matter of fact is decomposable unlike what was said in one of the slides Then you would not find that the result simply because in that case Countries are weighted by their income and not by their population. So china would weigh much less In that case than the united states and the increase in inequality in united states was definitely Cancel the drop in inequality in in china the last word about david san the presentation I'm not sure about the way in which we have to interpret the result that you present I would tend to say that this result that Today the progress in health is progressive Is a sign of inequality Because what it means? What does it mean? It means that in the progress in health In the population in those countries Rich people who were the first to benefit from this progress And what you observe today when you look at the last 10 years Is that the progress Is made in at the bottom of the distribution The first people who benefited for example from the better education or from secondary education in those countries Are people coming from which households? And 10 years ago 15 years ago 20 years ago The children in the other household were not going to school So when you look at the evolution over the last 10 years, of course, it is progressive But the real issue is why was it okay that right at the beginning more health care was Delivered to rich people than to poor people. This is I believe what those curves are showing Thanks very much. Is there another question? Yeah, I have a couple of questions the first one Is I didn't get your name. I walked in when you were already speaking. My name is Rahul Yes, you presented an example about Mexico where you predicted the consumption shares and that's a country one of the few that has Both income and expenditure. So Why didn't you maybe you have done it? But why didn't you show? Use a Mexican example to show how well you predict or not The the consumption shares with your method because you can do it for Mexico for every year They've consumption expenditure surveys since the 1980s Up to 2012 now. So you could use that case for that purpose in Brazil. That's too and David I was kind of puzzled Something one of the curves. I mean, it's a very Very pedestrian question, but you had a negative growth in height For people That are rich in one of your curves and how's that? I mean, it's it's peculiar. Did you notice that? in one of your health incidents curves because it means that They're shrinking and since that the rich I didn't understand why Okay, and one more This is for Miguel and possibly for you it might have some implications for your work as well On the decline in the between group component of the genie coefficient global between group component It's probably susceptible of a rather simple mathematical explanation Not least because of China and India which have very low basis of mean incomes And have registered very large rates of increase in per capita income so it's inevitable then that If you're starting from a low base and have and register high rates of growth Then a relative index of inequality will decline and And this is also entirely compatible with absolute differences in means actually increasing So the the the tendency for inequality to decline might actually be reversed If you were to look at an absolute index of inequality or More reasonably a centrist measure and in this context, I think I should draw your attention to two recent pieces of work One is a paper by Atkinson and brandolini in which they show very clearly that it is only relative indices of inequality Which have displayed a secular decline In the global distribution of income unlike Centrist measures and more recently a paper which appeared in economics by decoster deconk and and Post months So I think this is an extremely important issue where the protocols of measurement themselves determine what kinds of trends you're getting And with respect to the health Trends which you see I think really it's it is it should be regarded as being somewhat unremarkable that people at the very Bottom of the of the income distribution should be registering large rates of increase Simply because of the base effect A given absolute increase will translate itself into a very large proportionate increase for those who are lower down the Income distribution scale that's not necessarily a case for celebration. It's what one should normally expect When you don't see it happening as in the case of growth incidents income growth incidence curves That is indeed bad news But I don't know that what you call good news is necessarily good news. It's just Unexceptionable. It's unremarkable But the other thing is really bad news Okay, thanks very much. Well, we've got a good collection of questions and comments there. Rahul, why don't you start you've got So thanks for the questions first Why consumption and income difference? Consumption and income are quite different concepts. You'll find that consumption tends to be far less concentrated than income and People tend to mix them, but we wanted to keep them separate because The trends in inequality as well as the levels of inequality are different like there's a recent paper In the u.s. Which looks at both consumption inequality and income inequality shows that consumption inequality hasn't changed much But income inequality has increased too much So there might be differences in terms of levels and patterns. So we want to keep the concepts different Um on the utility and policy, um there the income One can conclude by looking at consumption consumption inequality that It's not changing much, but income inequality might be changing a lot which might Result in tax policy or other issues. So they have different implications for policies On the part of consumption consumption profile and negative life expectancy That example was just an illustration of how the database could be used for partial ordering Or any other uses It could be a negative correlation. I'm not very sure. Thank you And one more question. I forgot. Okay. The mexico example for 1989, which I show from consumption to income Again, that is just an illustration of how For for in our database currently for 1989. We have Mexico only has an income survey So as from the data's data, we have we might The more data we include we might have both consumption and income, but this wasn't just an example of showing how For other countries also, we convert the income into Consumption of isaac and how it would look at the various stages Well, I think I we got more suggestions than questions. So I we just want to thank for your suggestions So, of course, we will look at the alternatives of Looking at for example, the TLT Thank you, Francois and so we'll thank you for the references. We will take a look at this And on the water meeting, perhaps, I don't know whether Thing wants to say something, but uh, yeah, I think it would be very interesting to see whether we can organize A meeting on these issues. Yeah, thank you David I must start with the Norris question since it's easiest At two issues one is firstly if you look at the tails of the distribution where you noted that in a couple cases There are large standard errors at those tails. So Those numbers that are minus point zero one don't mean anything give at the at Given the size of standard errors. Also at the tails of the distributions Uh, we use a relatively narrow bandwidth in terms of estimating the regressions We can use a broader bandwidth in those extremes, uh, or those drops or Precipitous increases will disappear. We didn't do that. So anything Shouldn't really say this but anything in the top five percent or bottom five percent you can kind of ignore for now But again, that's kind of represented in the large bandwidth. Um No, well, you know, they could but uh, not not too quickly. Um But of course these aren't people right, you know, it's an ordering of the distribution. Uh, so so it is plausible. Um, so, um Let me go back to uh, what you think is unremarkable. I think there was something it may be unremarkable There was some things clearly misunderstood though. These are not proportional increases. These are absolute increases So if it is somewhat remarkable given that they're absolute I would argue But it's certainly a different story if it was proportional, but just let me reiterate all the improvements in health were absolute terms okay, um so, um then, um François, I'm not quite sure how to respond to that You say the bad news is that things were unequal and the good news is that things are less unequal I I think that's right. I mean it is maybe bad news that very poor countries 20 years ago were the poorest countries that the well-being not only at an income, but especially in areas like health or education were Uh at at that state were particularly poorly distributed um But luckily that's history and as history is evolving things are unambiguously getting better Uh, and I think that's the that's the main story But let me also make a point about and it somewhat relates to your point even if you want to think about absolute increases You when friends while when you said well that the rich people got better health care for 20 years ago Well, let's remember that 20 years ago and even today The rich people are of the top five percent of these distributions. I mean in many of the countries particularly in africa 50 65 percent of the children Are malnourished so the vast You know, we're looking at these distributions. We can't look at these distributions like we do in obviously in france of the u.s At least in terms of health These kids are have very high infant mortality rates Up until the 75th 80 75th or 80th percentiles. They have a very high probability of stunting So even at the upper ends of these distribution These kids are not in great shape. Okay So the the increases though are concentrated amongst the worst off kids, but it's not like You have a whole bunch of kind of you know middle-class kids whose health is also improving everybody almost everybody in these populations Is really at a very high health risk Okay, um Let me let me stop there. Oh, I want one thing. I didn't say anything about the first comment about life expectancy So I was talking about two measures of health Just to be clear infant mortality and Heights of children You did ask is there a level of consumption at which Life expectancy will be adversely affected The short answer is not that I know of there is no evidence in any country that while there are Changes in terms of the causes of disease as countries get wealthier On balance despite, you know, the emergence of diabetes chronic disease and other things like that in sub-Saharan Africa or in East Asia Richer as countries get richer their health does improve So I don't think that's really something we've got to worry about at this point And I don't think it's something we really will have to worry about at any point in the future. So thank you Okay, thanks very much. Uh, well, let's see if there's more questions Two questions one for Rahul Can you give us a best estimated data as to when the the data series will be available publicly? Uh, then for david um I feel slightly uneasy about your your paper because One of the things you haven't mentioned. I think it's admirable to look at this long as long a time series as you can But one of the things that's going on over this period that you hadn't referred to is urbanization And I think if you analyzed your data separately for rural children and urban children, you would come to very different conclusions Because all of the every study I've read over the last two years points out that the height figures for rural children Because of a chronic underlying health condition in asia latin america and africa the height Outcomes for children in rural areas or universally worse than in urban areas So part of what is happening. I think in your results is simply the degree of urbanization Uh, that is going on over the time periods that that you're examining. So I would urge you and steve to Uh, perhaps take a back up and and take a look at this with What's going on with urban and rural children? I think you'd find quite a different different picture Hey, could you just pass the microphone across the aisle to know this gentleman in front of you? Thank you Hi, thanks. I've got a question for Raoul as well actually something I'm really curious about is that and I Apologies if I missed it and I don't know if you whether you've looked at it or not in your project, but um So differences, let's say by decile decile shares of consumption versus income something I'm curious about is if there are any um, uh, any trends for example Across especially for example across decades, let's say, you know whether those the sort of differences Um remain constant over time and also whether there are any special things or things related to level of development level of GDP and so on Okay, and uh fin Thanks a lot. I I have a question for david or first an observation and and and a slight marketing stunt and then on the other hand A question the first one is when you say all of that literature just looks at one fruit and then you sort of Change it to the majority. I mean you will find actually 20 wider working papers emerging out of a growth and poverty project That that have I mean that has specifically tried to exactly say it's not just one fruit um The the question really is it emerges out of conversation with a key central bank governor in Africa He basically sort of says I'm fed up with donors because We've been doing what they told us Focusing on public provision of social services We've used massive amount of money including aid for that We've seen substantial improvements in these indicators which you are referring to Not quite a substantial An improvement in the income or consumption based and then he sort of sits there scratching his head and he says well but the public provision Of these services is part of government expenditure Which is not captured by the consumption based measures. There's just sort of wondering. I mean does that give any Sort of do you agree with that? Is he right? And and and what is it explainable Because of this observation of of of private consumption versus public provision. Is that what's looming in the background? Okay, let's take some answers and then we'll have more questions. So, uh, you had a few On data availability As we as three of us Tend to say that we are still in our alpha version and not even the beta version. So we are not very sure On exactly when but we plan to at least in the next year give Get out some data there The second Thing on deciles shares and income shares by spatial distribution So in the regressions, we tried to add both regional dummies as well as Decade dummies and see but it did not have much of an impact But that is also because we have limited data where country years have both the consumption and income survey So, uh, it's not conclusive Okay, and uh, David you had a couple So I I am certain that um, you're correct in the sense that part of the story is part is in part the dramatic movement of populations from remote Areas that are poorly served by public services Where markets are thin where prices fluctuate Where access to food is limited where there's greater Vulnerability to famine and so forth to urban areas. So there's there's no question in my mind that a large share of the health improvement May be attributable to the structural change in the economy where people live and They're access to services. So I think that's correct Um, and it is correct that urban health indicators are better than rural health indicators Um, yeah, I mean, I think it's worthwhile making the case You know and going back to some of these countries and doing urban versus rural But I can tell you unambiguously that the secular improvement like urban populations were better off 20 years ago Like they are now But rural populations have also seen dramatic improvements in their health indicators So I don't think this is um A story it is a story in part of movements of populations and accessibility But I don't think this is a story of only rural populations are getting better often and benefiting from kind of the the uh investments in public infrastructure and uh increases in their um general level of well-being Um Finn I think your question Or well firstly, uh, I stand corrected on on the first point. Um, I did not Okay, so let me leave it at that in terms of the wider papers my um Commendation to wider in general for being one of the few institutions uh To have a very broad perspective and and looking at these questions, uh, and Your second question though is is is actually a really I Or comment is an important one in my mind. I mean, I think one of the unherald unheralded um An amazing success stories in development is the improvements in health indicators And to the extent that you're suggesting that you know donors are our government officials kind of are um expressing a sense of frustration that this that that success is not fully recognized or internalized by donors and others I think it's an important issue, but it's also an important economic issue and I remember sitting and having probably a quite similar conversation with with uh Who was the minister of finance in Ghana several years ago and he was complaining about them being uh criticized by the bank for a lack of improvement in their poverty numbers for a long period of times gone as poverty numbers despite kind of it was one of the poster children for economic growth and good governance and so forth um Were quite intractable the the income poverty numbers And then he said to me and and you know, I'm not sure The sales pitch may have been overstated But he said but look at our health indicators. They've improved dramatically and that's because we've improved our tax systems. That's because we Measure income and consumption after taxes, you know, it gave a whole bunch of reasons But he said but the primary reason is we're investing taking the money taking the income As much as we can through taxation and other means and investing in health So it's not showing this big decline in income poverty, but we're seeing a big decline in Other measures or a big improvement in other measures of well-being and he was saying that out of a sense of frustration So, you know, it seems to me that at the policy level I think one clear lesson is that the whether it's the donor community or governments They've done a great job in terms of improving health and they've done an almost equally job and good job in terms of improving education outcomes while whether donors and And policy dialogue has been equally effective, you know, in terms of improving GDP growth, for example, or even addressing income inequality is less clear to me But I think unambiguously in terms of the social service side The news is is really pretty impressive and Not sufficiently well appreciated even in countries where their economic Indicators have lagged, you know, quite significantly And I have a two quick question one to Miguel. I mean your results beside the objection made by Francois your result seems to be in jing very much for the 2008-2014 period On this sort of stabilization and then a slight decline on the china inequality And the question is it is this government statistics or are they independent service because when somebody told me that Well, you know, the chinese they started getting tired of being told that They're growing a lot, but they are very unequal So the central statistical office was received orders to improve their results Rather than the so that mean I don't know if there are other sources which do confirm that trend Now the second point is for David I think that There is the bourguignon story, which is there was a famous book in the past called the poor have to wait So first the services I've delivered to the Urban people then to the richer people in the rural areas or the more neighboring areas and And so that may be This is Francois explanation. Now the second one is the one that Finn and yourself have provided which is well much more money has gone into that And then I I agree with you is is a wonderful thing The third explanation could be that the health technologies Have improved. I mean I spent 14 years working for UNICEF and I remember that basically Jim Grant's Go BFF revolution. So so the the same dollars Spent on vaccinations basically had a much bigger impact on than a dollar spend on standard standard care So I don't know whether you are able to capture that but I agree with you that these are impressive numbers So why don't we take responses sort of question by question now? Um, so I guess the first question was again for Rojo Is that right? No, uh, Miguel Yes, um, thank you. Yes, um, the data is from the nbs. So this is the official data from china um China updated the last estimates, uh from 2003 to 2013 um And what they they did is not only to harmonize this time series of this period, but also they tried to adjust as well with national accounts to impute incomes from the The very rich so the end of the income distribution. So, uh, um, they they are They said that they did that and when I was in china last week They said they explained me that they have done that for that particular period So whether is The data is Adjusted or not. I don't know but uh, this is official later Can you make a quick comment just that um, my colleague Wen Jie Zhang Sorry, my colleague Wen Jie Zhang. She's also Been doing some work on china and noticed the same trend with the peak in inequality in 2009 using the utipi nido data So if you look into that you'll see that as well I think uh, we generally agree and just one clarification is I think it's less Technological fixes in the sense that we often think about technology And more in terms of the ability to deliver Uh Services in a efficacious manner Basically creating of creation of kind of institutions At the local level that are able to To provide These low-cost technologies. So I don't think it's been technological advances. The one clear exception Is a problem that didn't exist 20 or 30 years ago is hiv where where that miracle is a technological miracle of drugs But that's that's the one clear exception Everything else has been newt basically focused on delivery and uh and outreach and and uh dissemination of existing technologies Thanks, uh, okay any more questions Is one back here? Right back there On the right hand side All right, thanks. Um, I'd like to go back to the very first presentation. Uh Beatrice Um You said in your in the final power point the literature on inequality measurement is very messy Um, perhaps I'd like to offer a slightly more positive interpretation of what you were showing Because you showed some country cases where It really was messy because the you had few observations short time series and very contradictory Trend analysis, but you had some other country cases Where the picture was much clearer, although You had longer time series. I think you pointed out the uk As as a case in point Where you have long time series and where you have four or five six data sets that shows the same trend lines So once you have that You can and you have trend lines that would either show For four or five or six data sets Decline and then a bottoming out and then an increase or the vice or or vice versa Then you can be pretty sure That there's something there If there's consistency between the trend lines of quite a few observations And even if those observations are on fairly shorter timelines and that's the china example Um, it's looks like this. You had four or five observations that said increasing inequality From sometime in the early 1990s and then a peak around 2008 2009 and then a decline Um, that gives me quite a lot of confidence in those observations um, and Um, Miguel, I don't think you should be surprised that There was the peak in around 2008 2009. I would say um A lot of literature would actually say that once you've had uh that long period Of growth very high level of of gp growth in china and an increase in inequality over a period of 15 years It's not surprising at all that you then get a peak and the inequality starts to to decline According to those measurement criteria that was was used Thanks Yes, thank you. Um About your comment one thing I just wanted I agree with you that it's true for some countries. Um You do have enough data that gives you some confidence in Um the measures, but I didn't show a lot of the countries where it's much more messy such as the sub-saharan african cases where there's just a lot of missing data and It's incongruent. So Well, thanks, I think you're right. I mean, um, of course, um, we may expect that growth has an effect on on the long term Declining inequality, but I think the story is not only that So, um, because precisely we observed this this drop after the crisis. So yes Growth has an effect obviously, but I don't think it's the only A part of the answer so Okay, we have time for uh, just one more question at least one of our speakers has to leave Exactly at 330 for transportation reasons. So It's to you No, just a brief observation for what it may be worth Why we're raising the issue of the difference between a consumption and Income inequality Is because from practical experience for instance in South Africa recently An announcement was made that Consumption inequality had declined and so inequality had declined But when the information was interrogated further it was established that in fact it is consumption inequality Whilst any income inequality had not declined and amongst Other reasons for this if you were to look at Mine workers, for instance, many of them are Neck deep in debt So their income is low and they go to micro lenders simply to to to support their their consumption So the difference between the two I think it's an issue that needs Clarification as the data mining and and and analysis is completed. Thanks. Let's just take that as a comment And then there appears to be one very urgent matter that has to be raised here. So would you please Have the final word there. Yeah, I'll just ask wondering because I've been raising my hand from the very first round Thank you for finally giving me the last word I have a couple of questions for David and First is that I tend to concur with my friend who says it might not be a very dramatic thing But for a different reason if you look at The relationship with the the the growth incident curves for income Well, the poor benefit but Lower than what the rich benefit But when you translate that to to health Is normal that the the the the the rich the the poor will appear to benefit more than the rich Probably because the the With income the rich can increasingly get more income or benefit more from growth But when you come to health something like Hyde for example, there's a limit no matter how much income how much resources you have You cannot just The relationship cannot be linear or to perpetuity. There's a cut of point So it could rather be a situation where it is the translation of well, the Relatively little much that the poor benefit is translated to that level of health But the rich have already reached their saturation point in terms of height and you don't expect much Second question is suppose Your point is the case that there's actually a bias benefit on the side of the poor Who does the credit go to is it government policies? Is it? Is it that governments are really being very effective in Distributing the tax revenue in terms of proper services I'm asking this question if you take South Africa for example, or maybe Brazil where you have you might have dual system of health and also education I don't know. I didn't see you present those cases, but I don't know what you will find out of situations like that Lastly and quickly. Let me ask a question to Is it Rahu? I don't know to what extent your assumption when you you had your your estimated Income and expenditure shares they didn't up at up to 100 and you assume that That shortfall was equally distributed But when I look at what possibly might be driving the regressions I don't think that the relationship between income and expenditure for the rich will be the same like that of the poor You might want to think that Maybe There will be bigger error for the rich than the poor Thank you very much And David do you want to respond quickly? So your your point your first point is well taken in fact Certain measures of health inequality will tend to be driven largely unlike income equality by the shape of the Left hand tail of the distribution because there is some limit to How tall you can get or how long you can live? I think many of the countries in the sample particularly in Africa It's not really a big issue yet because infant mortality rates are still you know 80 90 100 So even at the upper end of the distribution the probabilities of dying are still pretty high Um So but your point is generally well taken but in one sense it says that Maybe we should worry more about health inequality because it's more sensitive to what's happening at the lower end of the income distribution Or the health distribution rather than the upper tails Which do we really want to worry too much about how much money bill gates and Warren Buffett have? Um The second and how that drives inequality numbers the second point in terms of who I'll I'll leave it to you and the world bank and ministers to have that discussion and figure out who should get the blame or the The praise for I I really could not answer that question in a in a Informed way, but it's worth I'm not sure that you can apportion Uh The success you know decompose the success into the different players in the game And Rahul um, so um It's just a difference. Uh, it's not equally done but equi proportionally. So the rich tend to get higher So but the four percentage gap each of the quintile will get four percent So in absolute terms the share of the rich would increase or decrease more Okay Um, I can just make one quick comment um To me one of the most interesting things that we've heard about in the conference is the Decrease in inequality That's been seen in brazil recently Now we're hearing about it in china as well. I think if mixed mr. Heckscher or mr. Olin were here Uh, they would say well, that's what they predicted and took a little while for their results of globalization to Take the the form that they would have predicted But evidently they have so i'm sure that they're happy Wherever they are Okay, thanks very much to all the speakers and to the um a group here today It's been a really worthwhile discussion And let's thank the speakers again