 Good morning all, it's a pleasure to be with you this morning. My co-authors are also in the audience, so they can answer any questions that I can't, which is highly likely. So the South African case study is an interesting one in this project, because South Africa is very different to other African countries. So the focus is not going to fall on agriculture and food prices in terms of trade, but nonetheless it turns out that actually prices are quite an interesting and under-explored part of the South African story. So this is a country, the post-apartheid South Africa, which has grown 3.2% on average, 1.5% per capita. That's not really roaring along, but it is a positive growth rate. And the received wisdom is one that it's characterized that we've had some decreases in poverty in South Africa over this period, but starting at a really high level of inequality, we haven't made any progress at all in reducing that. So the first part of the paper makes sense of that story. There's a lot of research, we've got quite good data in South Africa, and it's mostly the income data that's been used to tell that story. And I'll give less attention to that in this particular presentation. A gap in the knowledge then is the effect of differential price movements across the distribution. And what has that done to this poverty and inequality story? And that's the main contribution in a sense of a new contribution. Let me briefly summarize the received wisdom then on the post-apartheid poverty and inequality story through a graph, through some kernel density estimations on real household income per capita from 1993 at the dawn of the post-apartheid period to 2010. And I'll just pull out a few key trends. We put a poverty line on here that is the Statistics South Africa's poverty line. That's in 2010 RANS, everything's here in 2010 terms. And very briefly, one can see if this is 93 and this is 2010, you can see the distribution pushing to the right, which then does manifest in this decline in poverty that we see in the income data. And if you plot the cumulative functions, you can see this very clearly. It's not a huge reduction in poverty, but it is, it's definitely there. And you can see the distributions piling up a little bit higher in the middle. And then you can also see some action out here. And that's the problem, right? So if you then plot the Lorenz Curves, you see that they're pretty much sit on top of each other over the post-apartheid period. And don't forget that South Africa is starting in the top tier of most unequal societies in the world, and we haven't done anything about that. So the paper spends a bit of time just documenting that using the 2010 data to splice it into the story to make it contemporary. Then we also tease out mostly through a review of the existing literature, the non-money metric story. Because it is quite interesting, it's a much more positive story. So work by Harun Barat and the DPRU people, as well as by myself and Arden Finn, et cetera, tell a story of asset poverty that's declined significantly in terms of access to water, electricity, sanitation, and housing. That's clear in all the data. It's pretty much a consistent story. If you do some multidimensional poverty work, which we've done, that then splices in access to education and access to health into the story as well, along with some other things, and there's also very strong improvements in access to health and access to education. There are, I'll return to some issues about that. That's access. So we've got some improvements in this non-money metric side. But then given that, and given the income story, how do you reconcile the two? How come has the income growth been so thinly inclusive, if you like? And there's all sorts of stories that I can't dwell on too much here. One is that it's too sluggish to begin with. 3.2% is just not enough to get going. And most of our sort of planning models we've used in South Africa call for 6%. But obviously you can call for whatever you like. That doesn't make it happen. We haven't had six, but it gives you a sense that we're short of the sort of momentum that we're looking for. And then there's a lot of complicated stories. There's some decomposition work in the paper that shows that basically the labor market has to kick for any of this to work, and it hasn't. South Africa has very high unemployment rates, and they've increased over time. And that's a huge problem. You don't get a return on any of your investments in education or any of your investments in better sanitation, et cetera, a longer run, a more dynamic return, except through the labor market. Then there are some stories about the fact that access to education doesn't mean quality, and their concerns are of a quality of education, quality of health, the quality of social expenditures generally. Our social grants have been expanded rapidly. They are responsible for the reduction in poverty. If you look closely, they are responsible. But even there, they're not generating any of these dynamic effects that we heard about in the lecture this morning. That's actually gonna get people out of poverty traps and on a different trajectory. So that's the story. It's sort of a received wisdom. What about the role of prices? Well, this was a painful process. Let me just say that at the outset. Because having worked on income data and being very comfortable there, I was, and our team had to dive into the weeds of expenditure data. And it was, I'm very pleased we're out of there for now anyway. So we looked really, really closely at the expenditure data. Monet in particular is probably South Africa's expert on those, these data. And we had, potentially we had great data, 93, 95, 2000, 2005, 2006, 2008, and 2010. But there's a lot of variation in the format of the expenditure data. We were quite happy to work hard on that. And we reconstructed the aggregates in trying to do the best we could, fine. On the price data, there's also some issues. Perhaps I should make a key point though here. We don't have prices and quantities, so we don't have what Channing has. So we couldn't do the toolkit thing, we've got expenditures. But we're still doing what we can do, right, with expenditure quantities. So expenditures were tricky. Price data, there's all sorts of things going on. The coverage changed over time, there were methodological changes, and they're not really any rural price data. Fine. So we spent months there to emerge with a finding that actually, on the expenditure side, we only had three data sets that were consistent in the sense that we needed them to be consistent. So 2005, 2008, 2010. With those in place though, let's take a close look at what they did to the poverty and inequality story and see whether that's useful. So starting with a percentile consumer price index, we created using the bundle of goods consumed by people at the particular percentile, as weights, you can calculate the price index relevant to that percentile. And you can sum across all of those to get the percentile-specific consumer price index. This is tricky work, actually. So what do we have here? We have a CDF of consumption expenditure per capita in 2005, and the blue one is just the regular CPI CDF. When we do these percentile-specific things, you can see that actually, people spend less, which in the CDF sense means they poorer, but clearly not, right? I mean, what this means is they're spending less for the bundle of goods if you price it at their prices. So starting at the base year, 2005, the percentile-specific ones led to an upward shift of the CDF. If you then do 2010, which is how I'm going to tell my story, 2005, 2010, they're sort of on top of each other. So there's been a change over time. Now, at the percentile-specific, they're spending the same amounts as with the CPI. So clearly, there's been an increase in expenditure over time. And you can see that very clearly if you do the growth incidence curves, it shows that spot on. The blue is the CPI, growth incidence curve. The red is the percentile-specific growth incidence curve, and it's much higher. But then the key question and the sort of heart of the issue is, okay, well, how much of that increase in expenditures is really a good thing, like it's actually a representation of increased well-being, versus how much of it is just the price effect. That's not a good thing at all. So we do two exercises that work hard on that, because clearly then you have to distill out the price effect. So there's some work to be done in the paper. And we follow a cluster of papers. The famous Dr. Volley on sort of decomposition of poverty into growth and redistribution effect has been extended by Gunther and Grimm. And Grimm's here, if you've got any questions about the technique. To decompose poverty changes into a growth effect, a redistribution effect, and then with Gunther and Grimm also a price effect. And what do we find? So we find, of course, that growth. We had some positive growth and it did lower poverty. There was a redistribution component too, which also lowered poverty. In fact, it was quite strong in lowering poverty. And then the poverty line component, which is the value added of the Gunther and Grimm approach, shows was positive though, it increased poverty. So the price changes actually were indicating that in terms of the poor, the price changes were anti-poor. So we have an improvement in poverty, but the price changes themselves were not helping. They were actually dampening the impact. Sorry, five minutes. Oh, thank you. Yeah, I was trying the chaining technique of ignoring you and then you sort of claim an extra minute implicitly. Okay, so then moving on to inequality. There are similar techniques that are asked to use the data that we have. So they aren't as sophisticated as what Channing was doing, but they do enable us to try and break apart the expenditure effect into its price effect and a real changing welfare. Goni, following Goni. We decompose now inequality. The inequality indices in particular into a nominal consumption effect. You've got a nominal change in inequality and you decompose that into an effect due to changes in inflation or prices and then in a real inequality component. And there's some subtlety here in the sense that each household, you're valuing the household again there. You're creating like a price index for each household based on their consumption bundle in this technique. It's not a percentile specific, it's actually a household specific thing. I just show you this through the Lorenz curves. Just so that we can see what we find. So the bottom line then is 2005 at 2005 prices. That is the Lorenz curve as conventionally plotted. The top line closest to the line of equality then is the 2010 basket at 2010 prices. So inequality improved marginally in that sense. And then in the middle you can see if we look at the 2005 basket at 2010 prices. So that's the attempt to take, to account for the price effect. You can see that it pulls, it does increase, improve inequality and take us closer to the 2010 situation, but not all the way. And the impact of the Goni decompositions really show us that the effect of real inequality does change and the actual real inequality effect dominates the price effect which I'm sure we would have expected, but it does turn out even here that the price changes were anti-poor in the sense. So two pieces of evidence, one a poverty decomposition, one an inequality decomposition to say that the changes between 2005, 2010 were, the changes in prices were not helpful. We're not real increases in welfare. In fact, they detracted from the straight expenditure effect. So what's driving these results quickly? Well, obviously there's different levels of exposure of households to high and low inflation terms. Yes, that's just defining the problem almost. Here are some results of the 14 items with the highest price increases. In other words, greater than 8% per annum. The poorest 40% of individuals have relatively great exposure to 10 out of the 14. So that's the sort of the kick. Some of that's obvious. Some of that's what things we just believe is development economists. So food items, for example, where they're consumer-grade to share, we all know about angle curves. That's part of our law. But so, and necessities were indeed very, very important. But there's some very interesting things that I'd like to finish on. One is like electricity here. So electricity, looking at housing carefully. If you remember my story about the non-monometric asset side, access to assets is one thing. But if the price effect is really kicking really hard on things that we're supposed to have provided access to, that's a serious problem in actually turning access into real development outcomes. Thanks. Okay, very good. Thank you. Other quick questions regarding the presentation? Again, we have time for this question in here. Yes, there's a gentleman at the back. Murray, thank you very much. My name's Karl Poe from IFPRI. What did you do about only having open prices? Did you just assume that those apply to rural areas or did you do some kind of adjustment? And let's suppose you only did this analysis on urban areas only. So restrict your household samples to urban households. Do you think that it would have given different results? So good answer. I mean, we looked hard. So statistics, I don't know if it has started collecting rural prices more recently. We look quite hard at them and we don't have a good answer to your question, but basically it doesn't look as though there's a massive difference between the rural and the urban price indicators. So basically we didn't really do much about that at all, but we did try and address it in the paper in the sense of saying, okay, how much of a distortion is that? And it doesn't seem to be a big deal, which is almost incredulous in a broader African discussion, but in South Africa it's not implausible, given where people buy their goodies, they buy them from the supermarkets, even in rural areas, et cetera. There was a second question that I sort of forgot. Right, so we didn't do that. I don't think we did that. We could do it, and that's interesting. We'll pack that one away for doing. We'll go back into the weeds and do that. Maybe another question if there's one. If not, then we thank Marey and proceed to the next paper, which is given by David Stiefel from... Yeah, okay. That's a good one again.