 What I'm presenting is the result of a research that Arun in Cape Town and we people in Florence are carrying out on behalf of UNDP with some assistance from for data and other things from the Friends of the World Bank. And I think that Kathleen has focused on measurement. I will mention measurement, but basically my own question is not whether inequality in Africa is high or low, but whether it has changed. And I think that in doing that, how do I press this one? Okay. By doing this one, first of all, we try to measure the trends, not the levels, but the trends. And to do that, we built a database on purpose, which has many of the shortcomings that Kathleen mentioned, but I think there's so far this is the better instrument one can use. And second, we try to explain, and we try to explain using eight categories of goods, of variables. Number three, we have done some macro-pundle regression on the determinants of inequality, and we know some change for the better, some change for the worse. And finally, we conclude. Now, this database is what we call it, the World Income Integrated Inequality Database. And what basically we did, we selected all the databases, and on the basis of that, comparing all the statistics, selected the one which made more sense. We have a protocol, and basically we selected 29 countries out of 44 or 48. So we don't have all of Africa, but we have 90% of the population in GDP of Africa. And that is about 280, 240 gene coefficient between 1991 and 2011. Now, then, of course, we, for the country selected, if we had gaps, and if the trends are smooth in the observed points, basically interpolated, because in the end we want to do regression analysis. Now, the four countries, depending on the shape of their own trends, were aggregated in four subgroups. Now, if you measure inequality for these 29 countries in this period, basically you find sort of a surprising result. Inequality falls both with the weighted data and unweighted data, but this actually conceals more than reveal. And if you divide it by the four categories, depending on their own trends, you come up with very different trends. And here, this is why I started talking about bifurcation. Now, I mentioned that because before this work, we did one on Latin America, which shows a sort of a universal decline in income inequality in Latin America. Africa is different. Africa is bifurcating. Now, if I reduce the period to 2001, 2011, basically you have something like 60% of the countries which have falling inequality, but 40% of the population, and 40% of the countries which are rising, and 60% because there is Nigeria alone. So basically the whole argument is that Africa is very different from Latin America. Some countries are doing better, I think perhaps of Ethiopia, and others are doing worse. I think of Nigeria and South Africa. Now, the first argument is to say is, now we try to explain growth. Okay, if you look at all the sample, no relation. The interpolated line is completely flat. If you look at the falling inequality, there is no relation. If you look at the countries with rising inequality, you do see that there is a little bit of an increase in inequality together with the increase in growth, which is not what you would expect. Normally, we would expect sort of a downsloping line. So growth per se is not a good predictor. And so then the pattern of growth is what basically is important. And here I come to a point that has basically been commented by Aaron. And if we look at the growth pattern, we look at the changes in the value of the structure. So here you see that the dark blue line is mining and utilities in the middle, and you should read it on the right scale. Basically, you see mining and utilities are increasing. The orange one is manufacturing, which is declining. Then you have services, which is rising. It doesn't look much because it is on the lab scale, but it is declining. Then you have construction, which is rising. And then you have agriculture, which as you would expect is declining. So the bottom line is that agriculture is declining, and this is fine. Mining is rising, and this is, well, I don't know, and manufacturing is falling. So basically, we are starting to think of we are in a pattern of growth where there is a change in the structure of GDP, which is what we may call suboptimal. Now, if you look at these countries, you will see, just look at the two left. These are countries where the share of GDP, depending on the export and mining, the export and mining of oil and mining resources, is more than GDP. So for these eight or nine countries, between 90, 2000 and 2010, basically the share of mining goes up tremendously. Then if you look at the second part, basically that goes up substantially as well. So this is the rise in oil and mining, which is driving the share in GDP, but which is also, for the reason that Arun mentioned, capital intensive, political rents, capital flight, and so on and so forth, drives inequality. Now then what we did, we took a big data set, and we looked at the relation between value-added shares and inequality. The one on the left is agriculture. The more agriculture you have, the lower is genie. The second one is trade and restaurant, also egalitarian. Then construction. Then you have manufacturing, which is basically neutral. So if you create more jobs in manufacturing, you are not running into the risk of increasing inequality. And then we look at the bad guys, finance, insurance, and real estate and business services. The more you have it, basically the higher will be inequality. Mining, we knew that. Now community and social and personal services as well. And then, surprise, surprise, government services. So the more you move your GDP towards the sector, the higher will be inequality. And this is what has happened in many countries of Africa. Now there is another variable which needs to be looked at. This process of changing the structural pattern goes quite a bit with urbanization. Just read only the first two lines. Now the first one is the less developed countries. So there is the rate of urbanization. So in 2010, that 46% of people in urban areas. Then you look at SSA, and you see that they have 35%. And underneath is the change in relation to the prior 10 years. And you see that during the 70s and 80s, Africa more or less urbanizing at the same speed. But then you see that from the 90s, 2000s, 2010, basically there is a very slow pace of urbanization. Which means that if the pattern of urbanization continues as it has been occurring during the last 20 years, we'll have more people in public administration, services trade and so on and so forth, which are unequalizing. So this pattern of urbanization may become, may increase further inequality. Now then what, so one first point is that basically the pattern of structural transition is suboptimal. And then we come back to that why. Now the second one is what happens to the within sector inequality. In here, there are some good news. Now I've been looking at Africa for many years and I remember that food production per capita been declining between 1960, 90, what, 1980 or something like that. Now here, the countries, these are a group of countries and if you look in relation to 2004 and 2010, basically food production per capita has been rising. And it has been rising basically because of increasing yields in agriculture. And this is the pattern of growth we do want. Now these are both from, I mean you see these are increases which are problematic so they are not straight lines. But during the last period there are increasing land yields. And in some other countries like Burkina, and here we have Mikhail Grimm who has pointed out that, basically output has increased per capita has increased but basically by expanding the land surface. Now the pattern of growth in agriculture I think is central to the inequality in Africa. And I think the question is that first of all it depends on land distribution, aren't we, didn't we say that for the last 50 years? Okay, so there is this chart confirms that in Africa land genie and inequality genie and income genie or consumption genie are correlated. So the question is what happened to land distribution? I think there have been a lot of titling program, land titling program which have been going on. And there has been a drastic land reform in 75 in Ethiopia which has been confirmed and many other changes in that country. Then there has been land certification but then there has been also a very fast increase in land concentration in population growth in rural areas that where the land frontier is exhausted basically they lead to increasing land concentration. And then there is the issues of land grabs which is debatable but can be desequalizing. Now one issue is nobody has mentioned that but what happens to the demography? Just the top. Now let's develop region 226 to 133. SSA 238 to 265. Growth rate of the population. Niger which is a country which is two or three times. The growth rate of the population goes from 279 to 385. Population doubles in 13 years. And I think that when we try this in the panel later on we find no effect because you see the scatter. You see the population growth rate which are on the horizontal axis. They are more or less all the same while the genius are different. So on the macro data is irrelevant but on micro data is quite different. I mean the richest households have lower dependency rate than the poorer households. So then you have a huge pressure on poor families due to continued fast population. So I think that the population issue ought to be brought out. One of the reasons I don't know. Sorry. Yeah. Now what happens to second set of variables? What happened to macro variables? Well macro variables I mean I think that there has been a general stabilization in Africa. There is a nice book by Ben Ndoulou and companies. So you see that during the last decade average import tariff. So there's been a massive trade liberalization. Balance budget is minus 0.7. Low inflation and real exchange rate which has appreciated. Now why as what is the fact of trade liberalization on production structure. Okay this is Malawi which is two or three countries with the micro decomposition. So you see that the share of manufacturing on the right scale basically correlates very closely with the tariff rate. Basically if you liberalize you are exposed to steep foreign competition and you are losing share in manufacturing. So one of the reason why manufacturing doesn't grow is because of trade liberalization. I mean this is at least in Malawi. Now what happens to schooling? Is human capital formation progressed? Well I think that again I don't show some of the data. So there is huge improvement in primary education which is on the lab scale. And there is a much more contained improvement in secondary education. Now if secondary education improves very little it means that every year of education of the labor force will be small. And we know this very well established relation between rate of number of years of education and the genetic efficient of the human capital distribution. And up to about six years to nine years depend on the countries considered that this goes up and so if the gene of human capital goes up then the gene it goes up as well. Now and then furthermore this secondary education is poorly distributed. The green line this is Chikos. Thank you. Basically this is the enrollment rate of the rich and the broad enrollment rate of the poor. So which means that not only there is not enough human capital but the human capital belongs to the better of families. Now there is altogether there is some evidence however that the spending despite what Chikos showed I mean you know that there is some evidence that the increase social spending the orange bars this for Africa as it also doesn't tell you much and they tend to bring down the genie. So the social spending has had some positive effect on it seems. Now how about why is Africa growing so much. I don't mention that seven out of ten the countries are fast growing country are from Africa. Yes but why are they growing. Well one is China terms of trade. And so the terms of trade and then the question is it are terms of trade equalizing or unequalizing. And the answer is that in the case of Africa we see then by regression you see terms of trade increase not only for oil but basically also for the producer of agricultural commodities. Basically they tend to be equalizing because the many are agricultural products for these countries but not for the non-agriculture countries for the mineral. Now then everybody talks about democracy Africa is becoming more democratic and then here you have a peak in ethnic wars. So actually political conditions have improved revolution in wars where the law and also adverse regime regime change have been changing and been declining. So this seems to be to have a positive effect but you will see we don't find none. HIV AIDS you see at the bottom HIV is highly unequalizing. All the micro studies show that very clearly and then we see that from around 2000 that it falls not much in our region so we want to test that as well. And finally technological shocks these are people with cell phones and the internet and the arguments say well they are market integrating and therefore so the poor may also do better because of that. Now the relation with Gini is very weak and then what we do then we try to do a test on a macro panel. We try to test all these six or seven groups. Now the test is done for the moment with LSDD estimates we are proceeding with the GMM and some of the estimate which we some of the data I mean I don't want to but basically our basically growth and growth pattern the first five. Then human capital and the second block then macro variables the third and shocks 13, 14, 15 then external factors 16 to 19 and then governments. Basically what we see is that except for those who are marked in blue this means that the results are what we expected and now we did not expect all the variables to be significant so with three stars because some variables on macro panel for instance like the percentage of the urban population or age dependency ratio they tend to be not significant on macro panel but they are on micro panels you know. So by and large what we do find seems to be supporting to a good extent the hypothesis which we formulated. Now the conclusions is basically is that one and here we agree with Aron that basically there is a major problem with the structural transition in Africa. I mean there is initial of reprimarization so Africa is losing manufacturing is going back to the mines and Africa is only part of Africa there is an increase in productivity on the land and there is a premature terrestrialization of the economy with many people being poorly paid. So GDP per se is not a good thing. Now the second variable which is important is that the secondary education has increased but not enough so it tends to increase the inequality in the distribution of the human capital. Now there is an issue of population I think there is no slowdown except in southern Africa and that actually I think is highly equalizing on the given position. So pressure on the land, the stress urbanization, dependency rate, wage rates all these are variables that will be suffering and this is an issue which is hardly ever discussed. Now macro policy I think macro policy have been equalizing the inflation has been equal I mean the foreign inflation has been equalizing the real exchange rate is appreciating during the last decade and that is unequalizing. The distribution I think that despite the limited extent of social programs in relation to Latin America we do find that is progressive except in not in mineral enclaves because of the political economic factors which have been mentioned. The last one I think that the changes in external conditions which were not particularly favorable for Latin America they've been the terms of trade they've been equalizing but not for the mining of countries because remittances have been equalizing against all theory I mean most of the theory says that remittances are unequalizing. FDI have been mostly in the resource sector so they've not been equalizing. AID is not significant but AIIPIC has been. SHOCKS the fall of HIV and war intensity has been equalizing the diffusion of cell phones and that techno shocks not significant and democracy we don't know how to measure it. So by and large there seems to be a pattern of explanation which might be useful to explain what has happened in the region. Thank you very much.