 It's an honor and a pleasure to be presenting this work here today. What I'm presenting is a chapter in a forthcoming report on African data on poverty and revisiting the facts. And just to set a little context, as Chico described the economic growth and poverty situation in Africa, the purpose of this report is to revisit the data with a more kind of careful and fine-tuned lens on what do we know about monetary poverty and non-monetary poverty and the quality of the data. I'm not going to spend much time on this slide, but there are a number of reasons why we care about inequality in Africa, which I think is why many of you are in this room here today. The outline of my presentation. So I think what's nice about this session for me is that I've been in correspondence with the other presenters, and so we've in some way kind of organized ourselves in terms of the content in our presentations. So what I'm not going to present to you today is a story around poverty, inequality trends in Africa over time and what explains those trends and the determinants of inequality. Rather, my presentation is going to try to do two things. One is step back and look at the big picture on the data we have for inequality in Africa from a monetary measurement perspective. And then secondly, it's going to try to look beyond the genie in terms of how we think about inequality in Africa. So the starting point for our chapter was we wanted to kind of set the stage for thinking about how do citizens and people in Africa view inequality. And what we found when we looked to the data we have on perceptions from sources like the World Values Survey, Afrobarometer, and Gallup was a really mixed story. This graph basically replicates what was also presented in the Equity World Development Report from 2006, Chico is the head of that report. And what it shows is that you don't find a kind of universal perspective that all aspects of inequality are viewed as bad when you ask people about inequality. You see a range of answers in different countries, and over time that changes slightly when we have repeat surveys from World Values, but generally it shows you the same story. There's not an overarching strong message that people view all types of inequality as bad. This is also true when we look at perceptions of government and adequacy in terms of narrowing the income gap. Now here what we see from the Afrobarometer is a large share of respondents will say they want the government to do more, that the government's not doing a very good job at narrowing the income gap. But the level of response is not correlated with the genie measure of inequality in these countries, which is the red dot on that slide. So now we're going to deep dive on the data side. What we did for this report is we looked carefully at the set of household surveys where we draw our genie data from, our genie measures and other inequality measures. And what we see when we look at the landscape on data in Africa is one is that there's been a big increase in the last 15 years in both the number and quality of surveys that are available. But one of the problems that is pervasive is comparability in surveys conducted within a country over time. And two examples that we cite are genie and molly, which each have conducted four nationally representative household surveys for measuring poverty and income. And no pairs of those four are comparable within these two countries because they change fundamental aspects of the questionnaire in terms of diary or recall method, recall periods, the season of the year when the data was collected. That effectively make them not comparable. So when you look at the data ability from a comparability lens, what you find, and this is the map on the right, is there are a number of countries where you have two or more surveys with which to judge poverty trend and inequality trend where the surveys are deemed to be comparable. But there are a lot of countries where we have maybe one survey. And there are a handful of countries that maybe have a survey that we can't get our hands on or haven't even conducted a survey. So now I'm going to talk about data, a range of data matters when we think about inequality measurement. One of the things we looked at in this report was, so what if you have no survey? What do you do in terms of saying something about inequality in a country where you have no recent data? And here we did a very simple comparison with the standardized world income inequality database, which is an effort to impute genie coefficients across all the countries in the world. And on this, in this graph, you see that the red dot are the survey-based genie estimates coming from our micro data. The black dots is the mean of the 100 imputations coming out of the SWID database. And the blue dots are the 100 imputations themselves, which give you a sense of the standard error around the mean estimate of the genie from the standardized world income inequality database. And what you basically find is the efforts to model or impute a genie give you huge errors on both sides. At the end of the day, you really can't say much, we believe, if you don't have an actual survey to base a genie estimate on to calculate a genie from. Second approach would be to take something like the Gallup, which collects a self-reported monthly income question. So it's not actually a consumption question. The Gallup estimates are advertised as being one way to understand income inequality in Africa and certainly help you bypass the need for a complicated household survey that takes a lot more time to collect. So we look at the Gallup. Now, the Gallup comes up with estimates of the percent of the poorest households holding 20% of total national household income and the richest 20% holding the richest share that hold 20% of national income. And their estimates in the bottom row here are 63% and 2% respectively for Africa. Now, we do the same comparison with consumption. Consumption's not income, which I'll show you another slide that makes that point. The point of this slide is to say when we take a consumption household survey perspective, we come up with very different numbers. And at the end of the day, we don't know if these numbers, these shares are different because it's consumption versus income. If it's different because the income measure doesn't include the valuation of home production or if it's different because of the mode of asking these questions is fundamentally different between these two questionnaires. So speaking to the consumption versus income inequality, one of the constraints we have working in the Africa region is most of our household surveys generally don't come with what we would consider to be reliable estimates of total household income. And so often, usually almost entirely, we use consumption-based inequality measures. And it will matter because when you go to something like PovCalNet, a lot of countries in other regions, particularly in Latin America, are basing a genie measure on income and not on consumption. So here we have from a study that looked at three countries in Africa, and I show you Tanzania results and Uganda results, and actually did use detailed LSMS data to construct an income aggregate and compare inequality looking at income perspective or consumption, and not surprisingly, you find that inequality looks much higher from an income perspective. But that might give us some sense that the overall average and or high genie that we see in Africa that you'll see in the next two presentations may actually really be quite higher compared to other regions where it's an income-based measure of inequality. Okay, within country, going to the comparability story, I have two examples here. So you take Malawi, which has had three national household surveys, and if you know the design of these questionnaires, it's widely recognized that the 1997-98 integrated household survey on the consumption measurement side is not comparable with the subsequent two household surveys. And this makes a huge difference if you want to tell an inequality trend story for Malawi. Likewise in Cote d'Ivoire, 1992-98 Cote d'Ivoire household survey deemed not comparable to 2002 and 2008. It doesn't show as dramatic a change in trend, but none unless you would want to recognize that not comparability and potentially not make the mistake of inferring trend from two surveys that you should not compare or we think you should not compare. Okay, so we take the surveys that we have. We have about 117 micro-datasets from household surveys in 45 countries. Excluding the not comparables, we're down to 24 countries and 58 surveys. And so what do we find when we look at inequality trend? Well, we find there's a mixed story, and this will be, I think, repeated in the next two presentations. On the other hand, we find that it doesn't matter a lot if you restrict yourselves to the comparable surveys. So the trend story, even if you take all the surveys, all 117, won't look dramatically different when you eliminate the non-comparable surveys. The second aspect I wanted to raise beyond the questionnaire design aspect is deflated or not. In the course of doing this work for this report, I learned all kinds of interesting details about what kind of data we have and how it's computed that I wasn't aware of. And one thing I was not aware of is that the genie coefficient in Paw-Cal-Net is based on a nominal consumption measure within the country. So urban households and rural households, there's no price adjustment between the two. In a very small number of countries in Africa, there will be a price adjustment. Interestingly enough, it seems to be a historic artifact of how the data was funneled into Paw-Cal-Net and not a strategic decision. But by and large, the surveys that you find in Paw-Cal-Net for Africa anyway are nominal and not deflated. And that might matter hugely for saying something about inequality. So here, we managed to get a deflated version of the consumption aggregate, not necessarily deflated in the same manner across countries. So that's a caveat to this graph. But what you find is the nominal, when you take the, so here, my inequality measure here is simply taking the mean household consumption in the richest region to the poorest region across four countries in Africa, where region, for example, in Ethiopia is region, but in Nigeria, it's state. And what you find is that ratio as an indicator of spatial inequality in these four African countries goes down when you adjust, I mean, not surprisingly, I think, when you adjust for price differences that urban and rural households face. So our, you know, much of, in fact, before we did this exercise, much of our story about between and within inequality had a heavy emphasis on the spatial dimensions. And we had to kind of deflate that story a bit, if you will, in light of the fact that while that story is still there, it doesn't go away. It definitely has less magnitude when you use a deflated consumption aggregate to look at between and within inequality in any country at any point in time. Surprisingly, we found that generally, in terms of ranking countries, based on their genie and alternative measures of inequality like mean log deviation or PAMA, they generally held. And so we tended to focus then on the report on genie, but not to say you might not get subtle different stories from using different indicators. All right, shifting gears a bit, one thing we wanted to do in this report is say something about African inequality. So take the continent as a whole. Pulling together all the micro data. And what we find from an African inequality perspective is that Africa-wide inequality is going up. It's gone up in our estimate from 52 to 56, circa 1993, circa 2008, because we have sort of benchmark years for our surveys. On the other hand, when we look at the population-weighted country genie, we don't find it systematically going up. And that's part of the story in the next two presentations, I believe. And we find a greater share of African inequality is explained by these gaps across countries. So you have a bifurcation in terms of rich countries with richer households and poor countries with poor households and a widening gap between the two. Which is different than say, I think Bronco Milanovic's work on the story about global patterns of inequality. And not surprisingly then, what you see when you look at the entire set of households in these multiple country data sets is the share of the richest African households are largely in rich African countries. And the red here in this graph shows you the population residing in upper, middle, and high-income countries in Africa. I should have said also that Africa throughout this presentation refers to sub-Saharan Africa. The second part of our chapter then looks at unequal opportunities taking an alternative view of inequality and thinking about things like inequality of opportunity. How do circumstances at childhood influence adult outcomes? And if we can take information from surveys on parental background, the region of birth, for example, how much of that explains levels of inequality? So this is challenging for a number of reasons. One of which is we don't have many surveys that ask these background characteristics and we don't have a consistent set of background characteristics across countries. So not every country has the same set of circumstance variables and what we find. And the second caveat to that or third caveat I suppose is by, this is a lower bound estimate of inequality of opportunity. So this is what we think the minimum share of inequality is driven by unequal opportunity is. And it comes out to about 15%. But keep in mind the caveats that I just mentioned. Finally, we can also take kind of a similar perspective and say, well, okay, if we know about parents' education, can we look at how parents' education influences the level of education of children? And likewise for occupation, to get an intergenerational persistence, which is what we think is a core feature of inequality transmission over time. And so here we have work that's actually being presented in a poster session. So if you want more details, I encourage you to go and seek out the poster that Eleni Yitbarik is showing here. This is one example from Uganda. So we have cohorts of people over time. So we have Ugandans who were born back in 1945 and then Ugandans born in 1990 on the far right. A simple regression of child's, my own schooling on my parents' schooling gives a purple coefficient, which is the beta, which can be by some accounts a measure of intergenerational education persistence. So it's falling, which I think we think of as a good sign. The correlation coefficient on the other hand is flat and the standard deviation ratio between parents and children is actually going up. So in some sense interpreting this kind of depends on which indicator, well it does depend on which indicator of intergenerational education persistence you rely on. And here are graphs for a whole host of countries and you can see that you, there isn't one clear obvious story except to say that in most countries the beta on that regression is going down. Finally, we can also look at occupational persistence between generations. And here we have very simple figure on the fraction of sons of farmers who themselves became farmers by cohort. So again, from left to right, older people to the youngest cohort on the right. And in some countries it's fallen by a lot and in Cormorosa it's fallen. Ghana kind of bounces down and then bounces back up and in other countries like I would say Rwanda it's fallen a bit but generally there doesn't look like there's a huge amount of occupational transition. Now some of that will be driven by structural transformation but we find that even controlling for structural transformation and kind of the some sense economic growth of certain sectors doesn't remove the fact that there has been real progress in occupational mobility between generations. The last part of this work I still think is the most interesting aspect because we but it's also the one that's hardest on the data side. And that is telling a story about what's happening with extreme wealth. What's happening with the kind of inequality that our household surveys are never going to capture by design. Here a lot of stories about billionaire wealth Oxfam had a very big report that launched last year and so we wanted to look at that issue from an Africa perspective. And what you find is strong evidence that billionaire wealth is growing in Africa. And here we have information on the aggregate net worth from the Forbes list of world billionaires for Africa and you see an upward trend starting around 2009. Compared to other countries it looks like Africa moves like other countries in terms of the percentage point change in GDP per capita growth and the increase in the number of ultra high net wealth individuals. So these are not the billionaires these are only the millionaires if you will. Still these are hugely rich individuals but it generally it doesn't look like it's moving unlike the rest of the world. Okay so bringing all this together is a challenge that we're still working on but what would I say based on this? The obvious thing, 48 countries. There's a lot of diversity in terms of the inequality experience in Africa. Data matters a lot. The national statistical agencies and private entities who are collecting surveys have made a lot of progress but we still have a long way to go. In terms of equality of opportunity we're seeing signs of increasing mobility between the generations. We don't have as much data as we would like. We have very particular set of countries there. We face the same data challenge with extreme wealth holdings. We also have very limited information on the details around how ultra high net wealth individuals actually accumulated their wealth and to what extent it's coming out of rent extraction or political connection. So that's an area I think for more data and more research. Thank you.