 So, my purpose is basically looking at food prices that you need and inequality, and then I ask further whether inequality is underestimated in Malawi when you control food price differences. This is basically the outline. I'll talk about the motivation of the paper, the recent trends in growth, poverty and quality in Malawi, and a bit of the methodology and the results, and then if I have time, I will give my concluding remarks on what the paper is coming up with. So, there's a literature which is growing, which shows that actually there are food price differences in terms of what the poor pay for the same items and the rich. This literature basically talks about a poverty penalty that the poor face in the food market, so that they usually pay studies, actually those listed studies show that actually the poor pay more for food, right? There are different reasons for that. Of course, a good review can be found in Mendoza 2011 and Mula 2002, but one of the key reasons there is that it may be more cost to save the poor because maybe they're staying in remote areas or where they're staying it's not safe and therefore you have to impose the premium to cover the costs. And then of course the poor may face equity constraints, which means that they may buy food in smaller quantities, which means that actually they cannot enjoy discounts and therefore they need to pay higher prices. In the context of Malawi actually this is an issue that actually which was found with the World Bank in 2007 that around May-June everybody has harvested their maize and their other food crops and because they're facing equity constraints, they want to buy other things, but they will sell their maize. So they will sell it cheap in May-June and in February or next year, they will have to buy the maize at an expense actually. So why do they do that? There's a problem of reliable storage facilities, right? And they're also facing equity constraints, so they are forced to sell their food. When they have it in excess supply, they need the money to buy the things and then they end up selling the food. So we have a situation where those poor households end up buying the same food that was selling earlier more expensively. And then of course the poor may face higher search costs, right? Of course, I mean there's an argument there in the literature that search costs are invading you in income, that their search costs are lower for middle income families, they're very high for high income families and very high again for low income families. That's basically the background because it's that the poor may face inequalities in terms of the food prices that they face, and then how does that affect inequality measurement in my life? That's basically the issue. So the poor may not have implications as I said on the measurement of inequality. As you may recall, the interest law simply implies that actually in a poor country context, food expenditure, the share of food in a poor expenditure is so high, and if the poor are paying high prices, it must be actually the desecrizing effect of a poverty penalty will be much, much high actually in a developing country context. So what is implication of this? Well, aggressive food prices will need to, and the estimation of real income inequality. And in a number of papers that I've shown this in different work, why you need to control for, you see what we see in most countries what they do is deflation of consumption expenditure or income is done at a much, much higher level. And therefore this doesn't indicate into account the fact that actually households face different prices, that food prices are income dependent in a sense. And to sum up what we see in the literature, and this goes for all African countries and other developing countries, what Mola says, that deflation of warfare using regional national level prices in developing countries is a norm rather than the exception. So it is a norm in most developed countries to deflate a higher level using regional prices or even in some cases national level price indices. So with all these, first to capture the fact that actually you have the poverty penalty in the food market. So what do I do in this paper? Well, I'm trying to reexamine the inquiries measures in Malawi, taking into account the fact that actually there's a possibility that households may face a poverty penalty. So the current practice just in keeping with in other developing countries is that our national statistics when they are carrying their figures, they will just deflate their consumption expenditure using regional CPI, which will then use maybe a base for rural and urban. And then in the light of what I've just said, that doesn't take into account the fact that if households may be facing food prices depending on their income. So what do I do? First of all, I establish whether there is indeed a poverty penalty. And then next I then examine whether this has consequences on measurement of inequality in Malawi. Why is this important? Well, it is very, Malawi has been experiencing high levels of economic growth over the past 10 years. Actually, it was one of the star performance, if you read the World Bank documents. So what is the problem? Which is that actually with high levels of growth, we have experienced marginal declines in poverty. And then that has also been followed by increases in inequality. So you have high levels of economic growth and then high levels of inequality and then marginal declines in poverty. So this paper sort of tries to contribute to that puzzle. What is going on? Is it that inequality is so high? Is it that inequality is underestimated? That's what we're trying to do with this paper. So if poverty is high, you are familiar with the poverty growth inequality tri-album, which simply talks about the possible interrelationships between inequality, poverty, and growth. So we can argue that. So the findings from this paper try to contribute to this puzzle. But there's another important dimension to this start, which is that actually, if inequality is indeed underestimated, there are long-term implications of that. There are so many factors that is including forces to align, which show that actually if inequality is high, the poverty-reducing effect of growth is actually minimal. So if inequality is indeed high, which is as a result of underestimation, that has long-term implications of the impact of economic growth in addressing poverty. That's physically why this study is important. So just to talk about the growth figures, you see that actually this is for an average for 2005, 2007. Malawi was growing at 6.2 percent, and then between 2008 and 2011, it's going at 7.5 percent. These are high levels of economic growth. But at least see what happens to the poverty-headcounts. National level, if the 2.4, five years later, marginal decline. Of course, actually what you see in rural areas, there's an increase in poverty. So what is going on there? If you look at the genetic coefficient, again, it's an increase in inequality. So the gene was 0.39 into 0.5, and it goes up to 0.5, 0.452. And you never see a pattern in rural and urban areas. Okay. So what data do I use? Well, I'm using the National Survey, the second and the third integrated household survey, IHS-2 and IHS-3, which was collected by the National Statistical Office. This is an LSMS-type survey, so it's funded by the World Bank. IHS-2 had 11,000 pound-eater households, IHS-3, 12,000 pound-eater households. So all these data says collect information on food consumption. So IHS-2 had information on about 115 food items. These are categorized into cereals and vegetables and all that. And then 124 food items on IHS-3. The issue then becomes how do you... So the food consumption was collected in terms of different units. So some of the items were used kilograms and grams, or milliliters or liters, the standard units. For some of the items, they used non-standard units. So how do you convert all these non-standard units into the standard units? So the statistics office has an organization of conversion factors. So you can use these to simply then convert the non-standard units to standard units. But then these two guys, Fidozo, Galo, and Eka to Sanfo, actually found that actually there are so many inconsistencies and errors in the official conversion factors. So what do I do? These two authors propose they come up with a new set of conversion factors, and that's what I use in this paper. Mo did us about what happens to actual consumption when you use these new conversion factors. I gave it in back 2014. This is a paper that was presented here last year. So what do I do? Two-to-four total food consumption is basically composed of three items. You either purchase the food or it's coming from your own production or it's a gift. But remember I'm talking about a poverty penalty in the food market. So what do I do? My main focus is on the purchased food. In the paper, I give Mo did us about the shares between ruler and app in terms of purchased food and in the three components. How do I then measure the poverty penalty? Well, in the literature, the standard practice is to use the regression based method, which is basically to run a regression on price indices. Could be a household price index on income or consumption expenditure. So that's basically the practice by Mo and beta 2010. What I do in the paper is to take it apart from that, I use concentration indices. Why do I use concentration indices? Well, I view the poverty penalty as a form of socioeconomic inequality. That if it's possible, if there's a poverty penalty, it may be that there's a high concentration of high prices among the poor. So what do I do? I then construct a household specific glass base index. This is a formula that I use where the base category, the base is basically the price, the winning average price for rural areas. So rural or urban depending on which where the household is located. And the quantity again is the base quantity to be weighted quantity for rural or urban depending on where the household is located. So essentially I have a household specific price index that I then use to deflate my consumption expenditure, which is what we're using in the fighter measurement. I'm departing in a sense from what the statistics office does, which is to use regional price indices. So I'm saying I'm using concentration indices. How do I interpret these things? Well, simple. If the concentration index is negative, it simply means that there's a high concentration of high food prices among the poor. Essentially there's a poverty penalty in the food market. If it is greater than zero, again, the rich are paying more for food. If it is equal to zero, there's no income dependent food price in equities. All right. And the magnitude of CG, the G is actually rural or urban, the magnitude of the concentration index will be telling us the strength of the poverty penalty. If it is very high, it tells you actually the food penalty is kind of extreme. Now, it's simply a question of testing a simple hypothesis, which is that, well, the analysis at CG, the concentration index is zero or it's less than zero. That's basically what I do to check whether it's indeed a food penalty or not. So what do we find? Well, in terms of, so before I go to the results of the concentration indices, in terms of, so of course we find that actually the poor pay more for food, but then what does that imply in terms of measurement of inequality? Well, how do I measure inequality? Well, I use the standard measures and these are the measures that I also use by official statistics. I use the genetic efficient and the two generalized entropy measures, the tau L and the tau T. How do I handle my warfare indicator? Well, essentially, I have the new consumption aggregate, which is based on the new conversion factors. So in here, I have three sets. So I have no real per capita consumption expenditure, real per capita consumption expenditure, where I deflate the new consumption expenditure with the official CPI series, and then real per capita consumption expenditure, where I then deflate using the household specific price index. For purposes of comparison, then I also use the nominal consumption expenditure, his official aggregate, and then of course I have the real consumption expenditure, which is deflated by using the official CPI series. So what do we find in terms of evidence with respect to poverty then as well, all the concentration indices, so the analysis is done for all areas and in both 2004 and 2011. So I'm looking at spatial differences and intertemporal differences. What we find for all the survey periods and all the concentration indices are negative. Well, when I test that, when I test whether it's indeed these are negative, where we find that we end up rejecting the now that the concentration indices are zero in favor of the alternatives that are indeed negative, which is evidence, that is co-evidence that there's indeed a poverty penalty in Malawi. What we also note in the results actually is that the concentration indices smaller in the sense of they're more negative in rural areas, which implies that the poverty penalty is much more prolonged in rural areas than in urban areas. And then we also note that actually the poverty penalty is declining over time. So it's more concentrated in rural areas, but it's declining over time. And these are basically the results. I've left out some output here, which you can get from the paper, but basically this is what I was saying, okay, the significance of the concentration indices. Now before I look at the inequality results, just let me just give you a sense of what is going on when I do the deflation for the different consumption aggregates, the official aggregate and the new aggregator of yes, percent rate, cost that constructed. So what I do there is to create percentile specific averages of each aggregate and then see how that changes. And then what we see is that actually deflation leads to decreasing consumption of all areas and then locations and the decline is more substantial for poorest households, right? If we look at the first percentile, the decline, if we look at HS2 for example, the decline is 27.8 percent in consumption for the the decline is only 8.9 percent for households in the 99th percentile. So what are we saying here? Deflation is having a different impact on the consumption of different households, right? The poor households, the reduction in consumption expenditure is much, much bigger, right, than for the rich, which is basically what we're saying here that actually we have a footpath by poverty penalty in the market, okay? So I've said I'm using the results based on the tau, the tau L, tau T, and the genie. I will represent the genie results here. The other results are in the paper. But what do we find? Well, the genie coefficient is underestimated, right? The genie coefficient, so there are two things here. If you look at the new consumption aggregate, right, and for all the years and all the areas, we find that actually the new, if it isn't the new consumption aggregate, the genie coefficient is underestimated by official figures as well. In some cases, 10.4 percent or 5.7 percent if RHS-3 and RHS-2. So there's an underestimation on the account of the fact that we're using a new consumption aggregate, right? If you deflate by using the new price index, household-pacific price index, you also find evidence that actually is an underestimation of inequality by official figures, and the underestimation ranges from 3.9 percent to 7.41 percent. Similarly, you get similar figures when you're using the entirety and the entire area. So the poverty penalty here leads to a quantitatively substantial understanding of inequality. And essentially what this means is that official figures closely understand inequality. Here I try to say, well, I'm using unit values, which are not actually prices. They're process for prices. And in unit values, can't be contaminated by quality effects, right? High unit values may simply be a reflection of the fact that each household is buying more expensive food items, right? So I follow a detour, and then I just go into effects, which is basically, this is a quality effect, and this is the price, which is unobserved. And then when I do that, you run this regression, this zeta here basically will be a price, right? Like unit value without a quality effect, sort of, right? Then I do the analysis again using the adjusted unit values. What do I find? Well, the results are robust. There's still a poverty penalty. Inequality is underestimated. And then the difference between before adjusting for quality and after adjusting for quality is minimal. In other words, quality does not seem to be driving what I'm finding here. Conclusion, well, as I've just said, no mean or inequality, unless there's really inequality. Official figures in Malawi actually understood the inequality problem. And as I said, remember that this has implications both on why this is part of increasing high growth rates and low poverty reduction, or indeed increasing inequality. This partly contributes to the explaining that puzzle. But of course, it also has implications on the long-term path, poverty reduction path that Malawi may take. If you have very high levels of inequality, which I underestimated officially, that has implications on the reductions for poverty. Thank you.