 All right, thank you. So did you want to introduce me? Yeah, go ahead. Now we have Joshua Greenstein, a PhD candidate at the New School for Social Research. And he'll be talking on new patterns of structural change and effects on inclusive development for South Africa and Brazil. OK, thank you very much. So yes, my presentation today is called New Patterns of Structural Change and Effects on Inclusive Development, a case study of South Africa and Brazil. This is sort of an initial step in a larger project. I'm currently working on expanding this same type of analysis to a group of about 15 to 20 other countries and periods. So the hope is that this sort of initial case study of South Africa and Brazil is sort of a jumping off point or initial look at the subject. So the sort of motivation for this study is this idea that I have here all these different people who have been making this sort of argument recently, that recent industrializers or countries that are currently growing and transitioning are following a different pattern than has previously been observed as far as changes in the structure of the economy. Most importantly, the big point that's most relevant to what I'm talking about is the percentage of employment that goes to manufacturing. So the sort of traditional story that I'm sure we're all familiar with is this idea of a transition from agricultural-based economy and then a move towards manufacturing and then eventually a transition towards services. So what these authors and others are arguing is that this isn't really what's happening in countries today, that manufacturing employment is peaking at a much lower level of GDP per capita than it previously has and that there's an early transition to services, but that this early transition to services is sort of a different type of transition to services than you see associated with sort of already wealthy countries, right? So my project here is that I'm saying that the effect of these changing structural patterns on well-being has not yet been systematically examined. So I'm saying, okay, let's say this new pattern is sort of the more common one now. What does this actually mean for how people live and how it's affecting people's lives? And again, this case study here is sort of a first step at trying to look at this more closely. So what do I do? Using a multi-dimensional measure of well-being, it's not income equality, so I'll show you what it is more specifically in a little bit, but it's basically something similar to some things that we're familiar with, like a multi-dimensional poverty index or something like that. I'm using it at creating this measure on a household level and then I'm using this term insertion of households into the economy, by which I mean the location that they're living and the type of employment of the household head. So again, I'll show this in more detail in a bit, but I'm breaking these households into categories such as rural-based agricultural workers or urban low-level low-productivity services or urban manufacturing and things like that. And then I'm looking at a period of growth, so beginning and end year of about 10 to 15 years and I'm looking at how the changes in this part has affected this part, basically. Okay, so I'll get to this later at the end, but the kind of argument I wanna basically make is that particularly in South Africa, you see this big spending on social welfare mechanisms and that in the lower performing groups, for example, rural unemployed households, you do see a big improvement but that they're still being left behind very much until they're grouped. And so the argument that I hope to build towards is sort of that something actually similar we heard about Brazil this morning as well, that perhaps there's a limit to how quickly we can improve the population as a whole through redistribution and that what's actually needed is a better inclusion in the productive economy. So what I'm doing is making this multi-dimensional indicator and then I'm using growth incidence curves and decomposition of change between groups to make this argument which I'll show later. Data is from census data, from iPhones. It's census of the entire population. It's a 10% sample. Once I lose some of observations by missing data and things like that, I still have these very large samples. 98,000 is a small one. Brazil is over a million households. And the reason why that's important I think for my question is I'm dividing these into like 15 groups so I think it really allows me to do that and still be able to say something with statistical significance about these groups which might be more difficult in a 3,000 person or 5,000 person survey, excuse me. So this is the indicator that I'm using. As I said, it's something very similar to the kind of thing that we've often seen before, the multi-dimensional poverty index or something like that. The differences between that are basically based on data availability and the fact that I've used this to create a positive indicator rather than negative. So these different types of things, again, calculated for each household, child survival rate, school enrollment, electricity, access to clean water, et cetera, are combined to make this household level indicated at a scale of zero to 100. And then this is what I'm using to measure well-being. So these are these types that I mentioned, right? So again, it's made by basically creating these categories using two different criteria, whether it's a rural or urban location and the employment of the household head. So again, I mean, I was going through the whole list, but you know, so this is the household heads rural worker living in agricultural area, or excuse me, other way around, agricultural worker living in a rural area. This is high productivity services living in an urban area, et cetera, right? Rural unemployed, rural non-labor market. Okay, so this is the first thing I'm doing. This is really a probably fancier explanation that's really necessary for what I'm doing with this part. So this is a growth incidence curve. And basically what I've done is just I ordered these types that I've just shown from the least worst performing to best performing in the initial period. And then I've calculated a growth rate for each group and then it just draws a graph. And the purpose of that is to sort of see if lower performing groups are catching up, if they're growing at a faster rate and higher performing groups. The other thing I'm doing is a decomposition of change. So this is actually a method that I've adopted from a paper by Van Ark and Timmer and they were using it for labor productivity. And so I'm sort of adapting that to do with something about wellbeing. And so the idea here is that it's related to my focus on structural change, right? And where people are living and what people are doing for a living. So this WBI is just wellbeing index, right? So this first equation is just showing, so this is the average change for the wellbeing index for the entire period for this country, right? Excuse me, total change. And this is the total change within groups and this is the change from shifts between groups, right? So the idea is there's two different ways that the total score can improve, right? Households within a certain group can start doing better or the proportion of households in a poorer performing group could get smaller and the portion of households in a bigger performing group could get bigger, right? So what this method does is decomposes the entire change for the whole country sample and assigns each portion of that change either to improvements within a specific group, which is this one, right? So this is calculated for each group and it shows what portion of the total change is attributed to that group. And then this is the shift effect which this one is calculated only for the groups where the portion of the population have gotten bigger, right? And so what we're doing is comparing the average score in those groups that have gotten become a bigger percentage of the population with the scores of the groups that are getting smaller. And so what this basically does is assign some portion of the total change, that's this part, to shifts to larger groups, right? So if those groups that are getting larger are doing better than the groups that are getting shrinking, that are shrinking, this will be a positive number, right? And if they're doing worse than the groups that are shrinking, this will be a negative number. And so when you add up all of these and add up all of these, you get the total change in the indicator. I don't think I need to do this slide in this room, especially I'm citing a lot of my panelists here, fellow panelists here, I think, but this is just a little bit, I just have a little bit of an overview of things, other things that are going on in Brazil and South Africa, which I don't think I really need to spend time on, but the really basic important point of those slides that we just sort of skipped through is that they both fit this pattern of premature industrialization or low manufacturing employment, or whatever you want to call it. And so that's why I've used them here. All right, so these are the two periods that I'm looking at here. South Africa, 1996 to 2007. These are the two census I have in Brazil, 1991 to 2010. So just some descriptive statistics first to give an idea of how this indicator works. So again, these numbers are this multidimensional index that I've created and these are the average scores. It's on a scale of zero to 100. So for example, you can see that in South Africa, 96, the average was 65, and then by 2007, then I grew into 73, right? You can also see how we could use it to show differences between groups, right? Here you have, this is the head of the household is black, this average score was 60, head of the household white is 88, right? By 2007, this gap has actually closed quite a bit, right? The white head households didn't get a lot higher, they're already quite high, but there's this room for improvement, but and the black head households have caught up a little bit, right? So these are the kind of things that you can see with this thing. Let's say Brazil, it's a longer period, but actually the scores are somewhat similar and have grown by a little bit more, but again, a longer period, so 63 to 77, you can see the same kind of disparity between groups. All right, so this is this growth incidence curve that I mentioned earlier, and so again, what I've done is these 15 groups or 20 groups that I showed earlier, the lowest performing initial one is here, the highest performing initial one is here, then these lines map the growth rates, right? So one thing you see here, which is positive, is that the lowest performing groups did in fact grow much more than the higher performing groups, right? Now this is somewhat built into the measurement since this is a scale of zero to 100, right? So if you didn't see this, it would be a real big problem, right? But it's still, even for the top performing groups, there's certainly a lot of room to improve, their scores are high, but they're not at 100 already, right? So these are all basic indicators that could go higher. So you definitely do see a real catch-up effect here, right? And it's a little bit flatter for Brazil, a little bit more even across for South Africa, you see a bigger drop-off. This is just all of these groups, we won't go through every number obviously, but it gives you some of an idea of the differences that we're looking at, right? So this is South Africa. So again, scale of zero to 100, you see like agricultural rural is 57, 96. Mid-level productivity services is almost 80. So there's these big, huge disparities in these groups. And then it gets a little bit closer in the second period, but you can see the growth rates to a bigger for the lower groups, something similar for Brazil. All right, so this is, this decomposition of change shown in graphical form that I showed the equations for earlier. So the way to understand this graph is that these are the groups, right? And for every one of these bars represents the proportion of total change that are assigned to that group. So the black bars are the percent contribution of total change to improvement within the groups. And the lighter gray bars are the percentage of improvement assigned to that group to a shift towards that group, right? So if you added all of these bars up, you would get 100%, right? So what's interesting about this, the biggest change from a shift effect is towards this urban low productivity services, right? So this is sort of exactly the story we were talking about at the beginning with it's like wholesale and retail trade restaurants and hotels, things like that in urban areas, right? So that's a positive shift because people, households in this category are doing better than households in the worst performing categories that aren't growing like rural unemployed, rural non-labor market, right? So there is this positive shift towards this sort of low level urban employment we could call it, right? However, this group, let me go back here. Yeah, so that's this group, right? But the, so there's a big shift towards this group or a big effect, the positive shift effects for this group. But if you look at the actual absolute scores, they're still doing worse than most of these other much better performing groups, right? And in fact, in many cases they're doing worse than the higher performing groups were doing even at the beginning of 1996. So the big effect of within group is in the initially poorest performing groups, right? So we see this big black bar here for rural non-labor market, this big black bar for rural unemployed and they're doing much better. That's the high end of this growth incidence curve that I just showed, right? So what's the story here? That the improvement has come from within the lowest performing groups and this shift towards this sort of better but still not the best sort of mid-level group. These are the same, this is the same thing in numbers but I think it's better to look at the graph. It's total within group is 75% and shift is 25% which will be relevant when we compare it to Brazil. All right, so here's Brazil. The graph is the same way. Here, there's almost no effect from shift, right? Almost all of the positive changes from within group improvement. In fact, some of these even have negative effects from shifts which means these are lower performing groups that have gotten bigger as a percentage of the population but here's the same group which is sort of, I guess, the most interesting one for my question which is this urban low productivity services and secondary and they're also one of the initially poor performing groups and there's this big effect of within group improvement there, right? And also for agricultural role, right? So here you see even less of an effect from shift than you do to South Africa and again you have these initially poor performing groups that are counting for the majority of the positive change. So here it's 97% compared to 75% for South Africa of contribution to total change to improve within and only 3% due to shift. So I'm calling this initial observations not conclusions because it's a very initial project here and again the goal is to be able to say something sort of about global patterns as a whole about looking at individual countries so this is obviously a first step but what I hope to be working for here towards here is to make some argument like this at the bottom. So the initially poor performing groups are doing a lot better than they were originally and that's the major source of change over this 10 year period specifically for South Africa but they're still doing significantly worse than all of the other groups, right? So the improvement of shifts in South Africa all came from this move towards low productivity urban services which again is positive because that's better than the rural unemployed or the rural not in labor market but still worse than all of the other types of urban employment that exists, right? So I think the positive way to look at these results if we want to be sort of optimistic would be to say, okay in South Africa there was this shift the biggest shift effect was towards low level urban work they're doing better than the rural unemployed or the rural not in labor market and in Brazil there was initially a lot of people in this type of work and the biggest positive effect there was from people within that group so maybe now South Africans that are in this type of work and in this type of environment will start doing better too and this story will work out, right? That's the optimistic take but I think a more pessimistic or worrisome take could be actually something that we heard a little bit about again this morning in relation to Brazil too that the reason why these lowest performing groups are doing better is because of sort of redistributive policies towards these lower performing groups but even despite this big lift there's still doing much worse than households that are more productively inserted in the economy, right? So the more pessimistic take would be that there's some sort of a limit to improvement that can come from this avenue, right? That there's a limit to just lifting up the bottom performers through redistributive methods and that what really needs to happen is that the people in these households need to become more involved in the productive economy and that doesn't seem to be in many cases what's happening. So hopefully the next time I see you we'll have another couple of countries there and I'll be able to give you a better answer of which one of those stories is the correct one. Thank you.