 Felly, mae gennym ni i hynny'n gweithio'r panorama yn y gwagiau sy'n gweithio'r gweithio'r transformatio mewn gwahanol. Badaeth eich bod yn ymgyrchu, dydd yn cydwylliant gyda'r chyflodau i'r pethau am sylwedd ddechreu, ddim o hyd i yw pethau, yw ddiddorol, a ddiddorol yn ôl yn gyflodiau a i'n ddiddorol i. I shall start on this side of the room with the lady in front, please. Thank you. Mareike Rheborth University of Antwerp. I was wondering whether the... Can you speak? I mean, at my age I have the excuse of being somewhat deaf and I really should have a hearing aid. So please speak up. Okay, okay, sure. Thank you, please. Tell me if it's not loud enough, okay? Mareike Rheborth University of Antwerp, Belgium. I was wondering whether the finding that consumption growth is higher when initial poverty is higher. So the finding that you see in Africa is due to the fact that Africa is the continent where people rely most on rain fat agriculture. And so it's actually due to some rainfall shocks, maybe in the base here. And then you see like a mean reverting process going on. So that's the first question. Secondly, I think the finding for the 19th period, 2001, 2005 has more to do with post-war catch-up because the social protection programmes only kicked off after 2005. And actually between 2001 and 2005 poverty reduction was really low. And those areas or districts which were really poor in 2001, they were poor because of the legacy of the violent conflicts. Okay, thank you. So of the room for the moment, the gentleman in the grey shirt there please. Alan Thomas from the IMF. Question, asking about your data sources for your labour data. Louise Fox and myself have just written a paper looking at 30 countries having a consistent methodology for establishing the definition of employment looking at it sectorally. Because we've been a bit concerned about the quality of the ILO data, not understanding your previous association. And we do find, like you though, quite a big change in the sectoral distribution of employment. As well, the paper is also sort of realistic in the sense of going forward. So we relate the employment then sectoral employment to sectoral output development using IMF data. And so the glass half full that you mentioned earlier, I think we've poured a little water out of that, given our conclusion that there's not much happening in manufacturing and employment in the next 10 years, even though they have very strong growth rates. Okay, well at least Africa has more of a glass these days. Okay, could we take Morton Yevan on the right there please, thank you. Sorry, and could you identify yourself for the audience? My name is Morton Yevan, Simon Fraser University. Thank you Eric for a very bold framework for us to think about. I'm a bit surprised that you do not address the problem of data availability and the quality of some of the evidence for some of these claims. For instance, when we talk about inclusive growth and you for instance quote Saleh Martin's paper, which is based on correlations with, well actually driven by changes in the national accounts data, whereas the inequality data and therefore the poverty lines are derived from a very small sample indeed. For instance Angola is not in that sample. We don't have data in that particular samples for Nigeria since 1996, so to then draw trends from 1996 to 2000 when we don't have data on these countries and others, which Andrea McKay and others have pointed out. But that's one thing. Also the other thing, which is more like an invitation, an observation about paradiseation. At the beginning you said there has been 3% growth per annum 2000 to 2010. That might well be, but it also said that there has been 0% growth in 1960 to 2000. That is true on average across those regions, but you could also say there was about 2% to 3% from 1960 to 1980, then negative or zero, 1980 to 2000. What I'm driving out is that this is a question of paradiseation, and if we are interested in the underlying institutional anatomy of growth, distributional effects, issues of land rights, structural transformation, could we not maybe then learn something from the previous period of growth, because Africa has had growth before, that is 1960 to 1980, or even in the case of colonial Ghana from the 1890s. Thank you. I'm moving steadily across the room, and hopefully if I can come back, I will do. I'd like to go straight to Martin Revalian. Martin, if we could have the microphone to Martin please. Thank you very much. Thank you, Eric. I'm a bit worried about, in the AER paper, I had a test of regional effects. Speak as loudly as you can. My apologies. In the AER paper, I'm going to buy you some hearing aid. In the AER paper I tested for regional effects and reported the results, and there was no sign of it, but there are regional effects in the intercept. I never tested for regional effects in the slope coefficients, and I guess my feeling is that it would be very hard to identify those effects. Essentially what you're doing is taking just Africa in that model, you're essentially truncating the sample in a way that's heavily skewed towards low-income countries with high poverty rates. Unfortunately, that's the initial conditions we're looking at in sub-Saharan Africa. I'd be kind of sceptical if you could identify the relationship in one region, and in such an unusual region relative to the rest of the developing world. That said, my hunch would be, echoing the previous comment, that what you're actually picking up with the headcount index there is near-classical convergence. The point I was making in the AER paper is that there are two effects working in opposite directions. There's a near-classical convergence effect, completely understood from the growth empirics literature, that countries with a lower initial mean will tend to have a higher growth rate, and there's a distributional effect. The poverty effect is a relative distributional effect, because you're controlling for the initial mean to factor out the near-classical convergence part. Unless you control for the initial mean the right way, there's a real risk that the poverty measure will start to confound the two things. It's like it's got a distributional aspect and it's got a near-classical convergence aspect. The issue then would be, have you adequately controlled for the near-classical convergence effect in order to identify the pure distributional effect? The point is that the poverty measure is there as a purely distributional effect. I think that there are a couple of other minor points that we can talk about later. Can I ask? Yes, that's fine. Let's have some responses now from Eric and then we'll take some further questions. I can see Sam wanted to come in with a question. Eric, please. The first question from Mareike. To the extent that I understood your points, I think it's very clear that while agriculture was more than exploited until fairly recently, it was the cash cow of the politician. Instead of trying to extract a surplus and investing this surplus in social overhead capital, it ended up in Swiss banks or wherever these people could put money. That has changed. This is, again, good news. I think much more needs to be done to nurture agriculture in countries at a very early stage of development. It's the only possible engine of growth at a very early stage of development. Of course, once they start growing, the evidence is that productivity has gone up. Incidentally, I looked at agricultural productivity. There's a good, if we study on this in Africa. There has been no acceleration of agricultural productivity in the last 10 years in Africa, which is somewhat disappointing. Again, this calls for supranational research institutes and what have you. That's on Mareike. On the gentleman who asked a question about labour data, that's a very good question. What I should have said is that after we looked at the data set that the World Bank had been used for the World Development Report, we found that, first of all, that data set was no longer available, but secondly, it had not been updated. We had a problem, and the problem was the following. The World Bank data were based on the share of agriculture in the labour force. The only time data that we could get from 2000 on was the share of agriculture in total employment. Now, I confess, I didn't have the time to look exactly at the difference that it makes in terms of definition, but it's absolutely clear that the area of data quality, data improvement, continuity in data sets is absolutely crucial. Again, if there's something that wider might be able to make a contribution in working on the quality and the continuity of data, questioned by Morton, we all know, and of course you've been pushing that idea since you published your book, that national income accounts in Africa are subject to enormous measurement errors, and this is one of the reasons, incidentally, why the people like Solaimartin who see the glasses half full and people like Alvin Young, to some extent, because they rely, well certainly Solaimartin relies on national income consumption data, and he shows a much rosier picture. But when it comes to inequality and poverty, I think I would, my own position would be quite different. I would say that the quality of many household surveys, the quality of many of the demographic and health surveys, on average is quite good, and poverty estimates do not come from national income accounts, they come from surveys data, inequality figures come from survey data. So I think we have to be careful here that we don't generalize from saying that because national income accounts are not very good, nothing else can be relied on in terms of poverty estimate or inequality estimate, but even then I would fully agree that quality can still be improved. Then I think you had a question about can we learn something from the pre-2000s periods, and here I would refer you to the book by Ndulu et al, which was one of the more successful collaborative projects of the African Economic Research Consortium. It was a book that had something like 20 contributors with, I don't know how many country studies, but maybe 15, 20 country studies, and it does go into a rather deep analysis of the causes of very low growth in the context of this whole period. Martin, I must say I was very surprised because I needed my hypothesis to build a case for pro-growth poverty reduction, because if I could say Martin, the world expert on poverty analysis, finds that high initial poverty retards growth, then of course it enhances or it bolsters the argument for working directly on schemes that will reduce poverty. So when I found that within the context of sub-Saharan Africa this relationship did not hold, it became a paradox to me, and again I'm trying to resolve it and I think much more work needs to be done to understand what it is. But I think it has something to do with the fact that over time some of the countries that had high initial poverty suddenly had much better governance, so boom, they grew faster and vice versa, that countries like Côte d'Ivoire that had been high growth countries after conflicts so they had low poverty suddenly grew less. OK, thank you very much Eric. Now we've actually run ourselves into the coffee break unfortunately, so I think I'm going to have to ask Sam and any others who would like to ask and discuss with Eric further questions over the coffee break. But what I would like to say is that Eric has given us a marvellous keynote on the second day of the conference. He began by pointing out that as one moves through one's career in development, creativity and experience blend together. I can certainly say there was an immense amount of creativity on display, indeed an immense amount of inspiration Eric particularly for the early career researchers who are just starting down this road. As a last point, I'd like to say that there's a multi-volume, multi-special issue study by Eric and Magicona Sankar who is in the audience today on globalization and poverty. Many of the papers are on the website as well in our working paper series and I would urge you to take a look at that because it's pretty much the definitive statement on this issue. Without further ado, I'd like to thank you Eric and convene the closer session. Thank you very much.