 G enjoyable morning and thank you very much everybody, for coming into this presentation, this as Michael said is another country that was part of that UNU wider project looking at growth and poverty reduction in Africa I was involved in a case study in Rwanda we actually had two studies in Rwanda. That one focused mostly on poverty and the links between growth and poverty reduction. I've been working on poverty and Rwanda for about 15 years, on and off. And so, of course, when one does poverty calculations, one routinely computes gini coefficients and looks at the evolution of inequality. But we never really properly analysed the case of inequality. So I wanted to try and look and say what's happening. Now, if one looks at some gini coefficient data, and of course, WID is the source for this, not the one that Finn talked about that was updated this morning. This is the one of a few days ago. If we just compare countries in the East African community, Rwanda has the highest level of inequality. Now, Kenya's about 45, Uganda's about 43. The latest figure for Rwanda is 49. Burundi and Tanzania have much lower figures. Rwanda's got the highest inequality in the East African community, but it's also got higher than most other countries in Eastern Africa or even West Africa. Ethiopia, Malawi, Mozambique, and so on. There is also in WID data from Rwanda from 1985, it says four there, but it's five, with a very low number. Exact providence of that, I'm not quite sure about. I don't have the data. Rwanda is a high inequality case for sure. Yes, there's countries in Southern Africa that are more unequal. Central African Republic has also got a higher value in WID, but this is a high inequality case. Now, what I'll give you is just 30 seconds, some features of Rwanda. I want to talk about consumption inequality, which is a sort of basic preferred measure which is generally used. But I want to also look at income data and see whether that gives a consistent story. We usually prefer consumption data to income data. It's better measured and so on. But I think it's important to look at the income story as well. It's also important to look at the income story because it helps to explain. Because income tells us how people get their livelihoods. Consumption doesn't tell you that. It just tells you what people spend and what people eat. It doesn't tell you where it comes from. Land is obviously an important story, so we'll talk about that too. Most of the data I'm going to use for the first 10 minutes of the talk are from 2000 onwards. But at the end, I'm going to try and say something comparing back to 1990. Rwanda is a special country. It's not a typical African country at all. It's got the highest population density in Africa. Like many African countries, it's dominantly agricultural. Like quite a few African countries, it's got a long history of conflict, culminating in the 1994 genocide. But a lot of conflict before that as well. It's a special country in many ways. It's also a country which has got quite good data, quite good survey data. There have been three surveys since 2000. These have good quality data. This is what's used for the poverty analysis. This is also what's used for the poverty story. If one looks at the recent history, and of course some of this is growth from a destroyed economy in 1994, but it's not just that, there's impressive growth of consumption over the 2000s, and over the 2005 to 2010, 4% per capita growth and consumption. Poverty fell significantly. Much faster than was Michael reported in the case of Burkina, for example, 59% in 2000, 45% basically in 2010. We're using, of course, real consumption per adults and so on. Those growth incident curves show the pattern of change. That shows the pattern of growth of consumption by percentile on the bottom. What one sees between the first and the second survey, one one sees in the early period is slight increase in inequality. That's between the first and the second survey. The growth is slightly faster at the top of the distribution compared to the bottom, but looking at this one one sees a different story. Faster growth at the bottom, slower growth at the top. Those charts already tell you what's happened in equality over these two sub-periods. It's gone up slightly in the first period, and it's gone down certainly in the second period. Now if one calculates measures of inequality, and there's different measures of inequality at the bottom, the ratio of the 90th to the 10th percentile, the generalized entropy index, two versions of it, the GE1 is the tile index, and the Gini coefficient. What one sees is an increase in inequality between the first and the second survey and the bigger reduction in inequality between the second and the third. So inequality at the end is lower than it was at the beginning. What one also sees in those data, though, is that these are high levels of inequality. These are high levels of inequality compared to the other African countries we saw. There's no question about that. If one tries to disaggregate a little bit and sees what is the pattern by region or by an urban rule break down. So this is the capital city, Cigali, other urban areas and rural, and five provinces. What one sees is a very big difference between urban and rural areas and between Cigali and the other provinces. The other provinces are much more rural, but one sees much higher Gini coefficients in Cigali compared to the rest of the country. If one uses a tile index that has the property of decomposibility and we can say how much of the overall inequality is due to inequality between groups, here between urban and rural or between the provinces. What one sees is that there is a massive contribution of inequality between the groups. Here, between 25 and 30% of the whole inequality in the country is just accounted for by the differences between these three parts of the country. That's massive. Compared with many other countries, that's a huge number. There's a very big urban rule differential here. That's for sure. For those that break down by province one gets slightly lower numbers, but it's still a big story. Or if one does the same thing at a district level, the same story. There's a big degree of geographic inequality here. That's consumption. That's the preferred measure. That's the measure routinely used for talking about inequality. The survey also collects information on income. We can look at different components of income and as I said, that's interesting because that can help us to understand that people are not able to pay for consumption. It's linked to livelihoods. Now, the problem is the income data is usually not particularly good. People don't report accurately their incomes and so on. There is some of this. We've calculated income in this data. The measures are not too bad and generally get better over time. So there's underestimation of income in the first survey, less underestimation in the second, and income is similar to consumption in the third. When there's a scatterplot of income against consumption, that graph in the bottom right, broadly there's a positive association and this is taking out the common elements. So things like consumption of own production are included both in income and consumption. Taking those things out, we still get a positive association. So I think there's reasons to think that the income data is probably okay. So what does it say about inequality? Well, again, the same structure as we had before, genicoefficence and tile indices, but let's just look at the genicoefficence high inequality, income inequality is higher than consumption inequality. That's not a major surprise. It's less accurately measured. It's more volatile. That's not surprising. But again, we see the same difference between urban and rural areas, perhaps a bit smaller here, and again we see the same difference between Cigali and the rest of the country. What we don't see there is a clear trend. There's not a clear trend. The story of consumption inequality increasing and then decreasing, that's not shown in the income data. Now again, that's possible and it doesn't mean that the consumption numbers are wrong. These are less accurate data. But what we do see is the urban rural gap. Broadly speaking, the features that we saw in the data I think are pretty much confirmed here. Now it's a huge agricultural economy, so therefore land is potentially a very important factor. So if we look at land inequality, now there's information where people are asked to report the plot sizes and so on. Land inequality, again we can calculate genicoefficence for plot size for area of land. And then again we see in the yellow shaded column, the left yellow shaded column, quite high values of genicoefficence. That shows the same pattern as the consumption, it increases and then falls. What we also see is very large numbers of households cultivating extremely small areas of land. Point two, hectares or less, is basically not enough to cultivate to feed a family, a routinely sized family and they need other activities. But there we have between 30 and 40% cultivating these very small areas of land. Now I want to say something about the economic activities that people do. Because I think that is an important part of the story. What we can do is calculate people's income from different things, from agriculture, from business activity, from wage activity and so on. And what we did is to try and define different activity groups. Now that's classifying people into basically 10 activity groups. The first five are where people get two thirds or more of their income from one source, agriculture, farm wage work, non farm wage work, business, transfers and rent. That's the majority of the sample but the majority of the sample is here. Agriculture only. And the other ones are people that are combining different activities. Now what this compares is the first and the fifth quintile just to broadly a transition in between those points, what we see is that the bottom of the distribution three quarters of people are reliant only on agriculture at least in the first two years. In the third year in the first quintile quite a few people move into combining agriculture with wage work. Either in farming or non farm activity. And there's a sharp difference between the first and the fifth quintile. Partly this is an urban rule difference but it's more than that. Now this movement into combining activities may suggest some sort of increase of diversification. Movement away from complete dependence on agriculture. Now if we define those groups we can look at inequality measures within those groups as well. And we can compare the levels of inequality. The levels of inequality are high in the cases shaded in yellow where basically people are not reliant in agriculture. Where people are working in other activities and business activities reliant in transfers, reliant in non farm wage. These are the high inequality activities. In agriculture farm wage work and other things inequality levels are generally lower. If one also looks at income shares by quintile agriculture is the very large source for the first four quintiles. It's only in the fifth quintile that agriculture becomes much less important and other things more important. Now the other thing one can do with the income data is one can do a sort of decomposition. The genicoefficient can be decomposed into inequality from the different constituent components of income. So here we have six components of income. This column shows the average share of income coming from that source. This is the genicoefficient of income from that source. Now some of the high values there are because there's lots of zeros. The last one shows the share of overall inequality which is due to a particular source of income. Now if we compare the two shaded numbers for agriculture the last column is lower than the first column. Agriculture is nearly 40% of where people get their income from. It only contributes 23% to inequality. Whereas these ones, the non-farm activities contribute much more to inequality than they do to income. These things contribute quite a lot to inequality. Significantly lowering inequality in agriculture land inequality seems not to show up so much in there. Land inequality I think matters for other reasons. Maybe increase the diversification. Maybe some evidence that public transfers are getting to poorer groups. I'm going to take a couple of minutes to just talk about extending back. These are surveys since 2000. Geneside happened in 1994. There were a lot of surveys conducted in Rwanda, especially agricultural surveys in the 80s and 90s. One in the 1990s collected a lot of income information and expenditure information. Methodology is different. It's difficult to perfectly compare. It's only rural areas. It's an agricultural survey. But let's see if it's possible to compare that with the later surveys. So what we can estimate from that is income and food consumption. Not total consumption, just food consumption. Now one I've tried to report there is different inequality measures for 1990 and the three later surveys. Focusing only of course in the rural areas when we're using the later surveys. The income data these numbers jump around an awful lot. 2000 figures are much higher than the 1990 figures but it jumps around a lot between there in 2010. Food consumption data is more stable and the food consumption data also doesn't show a consistent increase. In fact it doesn't show an increase in inequality over the period. The food consumption data is the data I prefer. I think it's the more reliable data. In rural areas maybe there has not been a big increase in inequality over this period. But of course remains. We can do the same socioeconomic group type analysis. We see pretty much the same composition of activity in 1990 as we see in the first two post-genocide surveys. We can do the agricultural source decomposition. We get pretty much the same message. So what in summary can we say? I presented you lots of numbers. I presented you lots of figures and stuff like that. Some of them I have more confidence in than others. What do I think we can say? Inequality in Rwanda for sure is high. The numbers that were talked about for Brazil this morning by the Minister current unique coefficient is not that much higher than what it is in Rwanda at the moment. Rural inequality may not have increased that much. Yes, the recent data showed some reduction in inequality between 2005 and 2010. Land inequality shows the same thing. But I wonder if that's sustainable. I think this is driven quite a lot by surveys that are always susceptible to the economic conditions when they're specifically carried out. The harvest in 2005-6 was relatively poor. The harvest in 2010-11 was better. Where does the harvest impact most in the distribution? We've seen it in the data at the bottom. So I think something of the pattern that we see of reduction of inequality is partly due to that. But other things are important too and transfers do seem to be reaching poorer ourselves. Reduced land inequality also may be important. The urban rural gap I think is hugely important and the gap between Kigali and the rest of the country. This is a really important policy issue. In Burkina Faso the problem was too much migration. I wonder if the problem is probably the opposite. Population in Kigali is very tightly controlled. People are brought in and poor people in particular are shipped out periodically. The government introduced what was effectively construction of new villages in rural areas from the late 1990s onwards. Another important factor. The way in which Kigali has developed I think these things contribute to this urban rural inequality. Non-farm activities are a big driver of inequality. Now, I think that it is important that the government and the activities are a big driver of inequality. Now that's not a surprise and that's true pretty much anywhere. And poor people can get non-farm jobs and we do have more poor people doing non-farm jobs. But they have more difficulty getting non-farm jobs and they get less good jobs. The non-farm jobs they do are pretty much unskilled. They may not be agriculture but they're basically manual construction type jobs. They get less access to them and they get poorer opportunities when they do. There is more diversion in 2010-11. More people are combining agriculture with wage jobs. That I think is a good thing. That I think is a sense of diversification but I think that could well be driven by a decent harvest in that year. Which therefore increases demand for labour and increases opportunities for work. It's not certain that this is a story that's another survey being completed at the moment it's not certain whether this is a story that will continue and will show up in the current survey. Thank you very much.