 Good afternoon. So this paper, the good thing about this paper is all the co-authors are here. So if there are more questions, I'll always be rescued by the authors. So this is my outline. I'll introduce, as you can talk about the background to this paper. And then I'll talk about the trends in poverty and growth in Malawi. And then given the trends in growth and poverty, you will see that actually there's a need for a reassessment of poverty in Malawi. So I'll talk about the new methodology that we've used, and then discuss the results, and then I'll conclude. Basically the goal of this paper is to estimate poverty lines for Malawi using an improved methodology. The details will be clear as we move along. This is, so essentially using two datasets. They are both national representative and they have very extensive consumption modules. So we're using the integrated household size A2, program known as IHS2, and the integrated household size A3, again known as IHS3. These are done every five years. Now this paper is part of a project, a uni-wide project, the growth and poverty project. It's been done in a number of countries, and Malawi was one of the selected countries. So if you look at the trends in poverty from Malawi, the picture is kind of very depressing actually. Since 1998, this is the first integrated household survey. The poverty rate was about 52.4%. There's been a very little discernible change in poverty over time. The only change that we notice here is actually between 2004 and 2009, but then these were based, these are official figures by the way, these are based on the welfare monitoring survey which doesn't have consumption expenditure data. So this was based on imputed consumption data. And there was hope then that actually maybe poverty is declining. So it was a decline from there up to about 39%. And then when the integrated household survey three data came, after upgrading the poverty rates, we actually found that if poverty rate is up again, it's about 50.7%. So if you look at the IHS 2004, there's very little change between 2004 and 2010. Now, why would you expect a change in poverty over this period? There's been a lot happening in Malawi over this period. And this lack of change in poverty raised a lot of questions. Is it true that poverty has not changed? What is going on here? So just to give you a background to that, so you see why this puzzled so many, this picture puzzled so many policy makers in Malawi, you see what has happened. There was a massive, and the next presenter will talk about this, there was a massive input subsidy which basically started in 2005, coinciding with the first or the second integrated household survey data. Its aim was basically to target small order farmers. And it was basically a massive subsidy. It's allowed 74% of the agricultural budget and 16% of the national budget. So it's a massive undertaking. Now, given the structure of the Malawian economy, you would expect this massive undertaking by the government to have some impact on poverty. Why? Because most of the people in Malawi are employed in agriculture, about 80%. And because the subsidy was targeting maize and tobacco, maize is also contributing about 25% to agriculture GDP. And the tobacco contributes about 67% of the export revenues. So it's a major exporter. So given this, in the light of this picture, you would expect that actually poverty should have declined. With the input subsidy as the next presenter, Malawi experienced almost double an increase in the yield of maize, double actually over the same period. And economic growth was about 7% over the same period. So this is a massive increase in economic growth. It was a massive increase in agriculture growth. Actually, agriculture GDP grew by about 16% over the same period. And agriculture, like we have shown earlier, agriculture is a major component of the economy. So any growth, any movement in the agriculture sector affects the overall economic picture. So 16% increase in agriculture growth led to higher levels of growth in Malawi. Actually, it was 8%, economic growth was 8% the highest in 2008, because at some point in Malawi, I was one of the fastest growing economists who kind of gone down a little bit. So it is in the heart of this background that you compare with this picture, this puzzle, why is poverty not declining with all these massive or impressive figures. And that's where this is a motivation of this, essentially this study, is to re-evaluate poverty measures, to come up with new poverty lines. So what do we do? Well, we use a new refined method, which is based on the standard cost of basic needs approach, which is exactly what the official measures are based on, but we try to refine it. How do we do that? There are five changes that we make to the official measurement of poverty. One is that we calculate regional specific poverty lines in the official analysis, it's just one national poverty line. And in coming up the food poverty lines, which are using calories and these will vary with your age, your gender, whether you're pregnant or other stuff, we take into account these, the age, gender and the poverty rates. In the official analysis, they only focus on the age. They don't take into account the gender differences, they don't take into account the pregnancy difference. So that's another dimension that we introduce. Again, we allow for the fact that the consumption bundle will change over time. And you also follow Ravarian 2009 by including an iteration procedure. Again, the official analysis is no iteration. And then key part of what we're doing here is ensuring that the poverty lines are iterative consistent. Again, the official analysis is not like that. So iterative consistency is iterative consistent over time and iterative consistent across space. So just to say something about what the official poverty figures were derived actually. So from 2004, what the National Statistics Office did was to use the standard cost of basic needs approach. So that was standard. Of course, mine are some of the things that I've mentioned here. 2010, which is the current data set, they did not do that. All they did was to just inflate the 2004 poverty lines with an inflation figure of 128%. This inflation, this figure was applied to both the food component and the food component. Where this number comes from is not clear actually. If you look at official CPI data, you can't find it. It's not clear again where this number came from. And we'll see later actually that this number is significantly slower than what we are getting from the data set. Okay, so I will spend a bit of time talking about this idea of iterative consistency. Some of the things that I've said we have changed are kind of standard. But this is based on a paper by Antion Chimla, Simla 2010. I think it was a presentation yesterday on this paper. So maybe some of you have seen what the methodology is all about. So the idea is to say, okay, utility can be assumed to be stable or constant. But the cost of getting that level of utility to be changing over time. So what the NSO is doing is to assume that as you move over time, the cost of attaining that utility, even though it has changed, they'll just multiply by the cost, the 128%. Which doesn't solve our problem. The problem is that you're saying utility is the same, but the cost of attaining that utility maybe has changed over time. What we do in this paper is to say, okay, well, we're doing this over time and across space. So the same bundle or a bundle can have a different cost across space, across regions, or over time. So what we're doing in this paper is to then say, okay, over the two periods, 2004 and 2005, has this changed, or over space, has it changed? How do we do that? Well, we conduct some reviewed preference tests. These are based on maximum entropy expansion. It's discussed in the paper, similar and anti. So we conduct these tests. Based on the tests, we found actually, for some of the provided bundles, we found that actually they failed the reviewed preference test. Now, condition on failing the reviewed preference test, you must adjust the bundles, right? So actually, most of the bundles, quite a number of them failed the test. So we had to adjust the bundles, and that's the final thing that we do. You adjust the bundles. After adjusting the bundles, what you end up getting will be the poverty lines, food poverty lines that you treat are consistent, essentially, without going into the technical details. That's essentially what we did in the paper. It involves some usage of guns and all that stuff. So after doing that, I will show you a picture of what we get compared to the official figures. Now, notice here that there are gaps here. Why? Because the official figure is just a national, these are poverty lines per person per day. So there's nothing on all the rural areas because it's just one poverty line. But what we're getting, like I said, we're creating poverty lines for each one of these areas in our study. So what you see here actually is the fourth component is about 27 for the official analysis. It's about 27, well, 28 quarter, right? It's just a little in terms of donors, actually. And then, that's plan four. If you move to 2010, of course it goes up, as we expect. But the non-food component, because our poverty line was just a sum of the food component and the non-food component, where we get the total poverty line, which is official figure here, you see that actually our figure is slightly lower than the official figure. If you look at the national level, again, if you move to 2010, you see that the national figure is slightly higher, the national poverty line is higher than the official figure. Of course, I mean, this is just a reflection of the fact that there is substitution over time and all that stuff. Now, if you look at the, I've discussed this, if you look at the food poverty line, the food poverty line for the two periods, it's fairly similar compared to the official figures. They're fairly similar. The difference comes in when you move on to the non-food component. The non-food component for 2004 is fairly similar. The difference will be noticeable in 2010-2011, which is, again, a reflection of the fact that things have changed over time. Actually, if you then move to the inflation, inflation for the poor, we find that actually non-food inflation is significantly lower than the one that was calculated by the sadistic surface. Overall, the national inflation for the poor is, the national inflation for the poor, which is this, is by the lower, significantly lower than the official inflation rate, which was used to adjust the poverty lines. So the overestimated inflation by adjusting by this, our own figure that we're getting from the data set is this one. In the paper, we try to figure out what is going on here. Is it the inflation component of it? There are two issues that are probably driving this. There's this non-food component, which is lower, and the issue of using higher prices in the official analysis. We discuss in more detail these issues in the paper. So it's just based on the non-food component. So all we are doing is to divide... Yeah, yeah, yeah, sure, yeah, sure. We have unit values, not prices of unit values, we are dividing the quantity, the amount, the total expenditure of the amount purchased. Not necessarily the prices of unit values. Yes, not prices and prices going to the market and getting the prices. We're just calculating what our core unit values are from the data. The issue that we said here is actually, as I've already mentioned, it's a non-food component, which is showing a market change in our results. But the non-food component, whatever, if you look at 2010, 2011, it's kind of lower than the official figures. And actually, we see that, actually, that's one of the drivers of this change in the poverty results that you see later. So we said, okay, could this be an artifact of our choice of the poverty line? So we did some robust analysis. We tried different non-food components. Actually, what we're finding, actually, it doesn't matter. For 2010, the non-food component is always lower than the official non-food component. So it's a robust finding. It's not something that is sensitive to choice of poverty line. Okay, now, given these poverty lines that I've just seen, we've added new poverty figures, poverty headcounts. We simply focus on poverty headcounts. We don't talk about poverty garb and poverty severity. But we can easily do that. But for the purpose of this presentation, I'll just talk about the poverty headcounts. Actually, what you see now is that these are the official poverty headcounts. These are our headcounts. So for 2004 and 2010, we have the differences here calculated. For the official figures, it shows that actually poverty declined by 1.7 percentage points. A very significant change. When you do our analysis, we show that actually the decline is minus about 8.2 percentage points. A much, much bigger decline in poverty. Remember what we said earlier that we talked about this massive intervention in the agriculture sector with the input subset. So it's kind of consistent with what we'd expect in the light of what was happening. Again, what is of more interest here is if you look at the rule of poverty figures. The rule of poverty figures, where most of the intervention in the agriculture sector is targeted, will find that the official figures show that actually in the rural areas, poverty actually increased marginally. Our analysis shows that actually for all the rural areas, north, south and south, poverty actually went down. Again, which is consistent with what you would expect given the intervention that we've talked about. So there are two things here. We're showing that actually poverty has gone down and the decline, the magnitude of the decline is much, much bigger than the official magnitudes. The other thing that we did was to say, well, can we just fix the Bando, the 2004 Bando, 2004-2005 Bando, apply that to 2010 and see what happens, because that's essentially what the official figures are based on. If you do that, and this is the last column, you are using a fixed Bando for 2004. What do you get? Well, the figures are slightly similar to what you get from the official analysis. Essentially, again, we have another thing which tells us what is driving the results. It's the fact that the official analysis assumes that the Bando is the same over time. Our analysis is saying, okay, the Bando, these things may be changing over time, and because they're changing over time, the results will also change. So there are two things here that are driving our essential results. It's this thing where the landfill component has gone down and the fact that the landfill component is going down, persistently in 2010, and the Bando, if you're using a fixed Bando, the results will be similar to the official figures. Using a flexible Bando actually leads to the differences in the results within our results and the official results. So are these the results robust to choose a poverty line? Actually, you see that actually, if you plot the poverty headcounts over all different poverty lines that you choose, you see that actually there's a change, there's a decline in poverty, whatever poverty line you choose. And these are mostly significant. Using the 95% confidence interval Bando here. Quickly, I'll just conclude there, say, okay, well, I mean, there are two things that are driving the results here. It could be, as I've just mentioned, the poverty decline that we're seeing here is driven by the shift in the landfill component, or it's partly driven by the fact that we have lower food inflation that we've estimated. Now, can we say that the fifth, the food and agricultural subsidy was effective, maybe, because this is kind of consistent with what most people expected, that poverty should have declined and that you was targeting the agricultural sector. And most of the agricultural sector is basically a rule. So our results are kind of consistent with what most people expected. Thank you.