 We did this work because we were thinking that a lot of the work on structural transformation is being done at the macro level and we were seeing a very more nuanced picture when we worked with household survey data. So we thought okay let's look at this question of structural transformation from the point of view of households and household portfolio and see what kind of picture that gives us in comparison to what you get from the macro side. So I think there's been a lot of discussion on structural transformation. I can probably skip this overview except here's the point is that it's usually measuring focusing on sectoral productivity. And so our approach is to really look at the non-wage sector because that's where most of the employment is and the classic idea of structural transformation from Lewis was that people move out of the family production or non-wage sector into the modern sector with wage employment. And it's true that in many countries, sub-Saharan Africa, especially non-resource rich countries that have had rapid growth, the wage sector is expanding rapidly but it's coming from a very low base and so and the labor force is growing rapidly so it's not really employing that many people. So it's likely for the next 20 years that the non-wage sector is going to continue. And does that mean that kind of structural transformation in Africa is doomed? And our answer is no. That structural transformation in Africa will incorporate the household enterprise sector as productivity rises in these household self-employment, household enterprise, micro enterprises. Now this will not provide the same spillover effects, for example, as private investment in large enterprises. Large enterprises are still, the modern enterprise is still one of the most efficient economic organizations known to man and woman. But our argument is that it will allow people to access higher productivity opportunities in the economy and it will lead to improvements in income and inclusive growth. So Uganda is a country that has had two decades of growth and output transformation. And so during those two decades, aggregate productivity in the economy grew substantially even though the labor force was growing at 3% per annum. So yet output per worker increased by about 70%. So 60%, something like that. So it was really a good period. Here's what it looks like in the trends. This is the overall trend of economic growth. And you can see agriculture over this period declining as a share of GDP. What's interesting is that after about, let's see, after really about the early 2000s, you know the kind of the picture got pretty confident, I mean there are a few bumps which may really be related to terms of trade changes or something, but the picture is kind of pretty much the same, which is sort of interesting in terms of shares. What that means is that all of the, and yet there was a solid growth. So what that means is that all the sectors were growing at pretty much the same rate, but there was productivity improvements. Okay, so then employment transformation also happened, but it's not quite as dramatic as the output transformation. This is not a surprise because we know that from work by Timmer and others that the output transformation always leads the employment transformation. So actually the absolute number of people working in agriculture did increase over this period even though the share declined. And this was a solid growth, this share of people working in industry doubling. And this is not mining because at this time Uganda has an oil sector coming, but at this time it didn't have it. So this is real, this is construction, this is a small manufacturing sector and then of course served like many other country services group, private sector services and public sector services. Now what kind of jobs were created? So we broke this down and our analysis shows that mostly these were not wage jobs. Most of the new jobs outside of the agriculture sector were created by households. So by people working for themselves or their families, employing very few workers, something like in about 2000, only less than 20% employed any workers at all outside of the family even on a casual basis. In industry and services there was substantial growth in the wage sector. The manufacturing did not absorb that much of the labor force, but in the non-wage the growth was generally higher. And here's a picture of that graphically. So what did that mean in the end for the economy for income growth? Okay, my goodness, sorry about that. I don't know, at this point I don't think I can get rid of that upper corner of the rural areas showing twice. It looked okay on my tablet. But basically what you really see, we can focus on the rural areas because this is the same as this. What you're seeing here, maybe it's clear in this color scheme, is that what people were doing was adding an income. So in Uganda about over 40% of people reported a second job. Many of those people reporting a second job worked in agriculture, either as their first job or their second job. So these numbers over here that we saw, this is primary employment. This is only, assuming they have only one job, this is primary employment. And it's these numbers on primary employment that are underlying all of the macro numbers that were presented in the previous session on structural change. So all the numbers from Ghana, all the numbers from Maggie McMillan, all of those aren't counting people's second job. But when we start to look at it from a household level, what we see is that over this, here this is 15 of the 20 years, we see that basically those in rural areas especially, those people who were in farming stayed in farming. This is nationwide, you see the same thing down here. So many, many households stayed in farming. Some got out. This is a share of total households. So what you have is perhaps some households moving in as well. But what they did is they added an income. They added a wage income. They added a non-farm enterprise income. They added a public sector, although the public sector wage employment is much smaller than private sector wage employment. It's much less important in household incomes. It's much less spread out among the households. There was also a growth of people working in agriculture wage. Agriculture wage is the lowest paid activity. People earn even less there than in family farms in Uganda. And this is generally true around the world that agriculture wage is not the best activity. You can see in all of Uganda there was a faster growth of private non-agricultural wage here and even bigger growth in the non-farm enterprise. Okay, so this is a picture of people's portfolios, their household portfolios. So the way this was constructed is if anybody said either I have a primary or secondary activity in farming or non-farming or whatever, it got counted as an activity. So in this type of household, everybody works on a family farm. Nobody even does agricultural wage labor. So they're totally specialized in agriculture. This represented a majority of the households in 1992. But by 2005-6, it was down to 30% of people that were specialized. Interestingly enough, that 30% was a very kind of a mixed group. There were people who were specialized and quite poor. And there were people who were specialized and doing reasonably well off. So there was some transformation going on in Uganda. There are debates about agricultural productivity growth. Lots of debates about that. Labor productivity growth, about the date and whatever. But it's clear that there was some commercialization. So you've got specialized farmers and some not. But then you have all these other sectors here that grew a lot. Like this one is agricultural wage and family farm. And that grew with the expansion of agricultural wage. And this one is a family farm and non-farm enterprise. And this grew as well. And then, and all these other small ones are different combinations. We have, obviously, with four different types of work, we can get 16 different combinations. So you have a few slivers there and then. But this is in the rural areas only. We showed it just in the rural areas. We have it also in the urban areas done as well. And actually, in the urban areas, there's still a number of people who have one foot in agriculture and one foot in household enterprises or wage work. OK, so what did that do in terms of incomes and poverty and structural change? Well, we argue that this was an important part of the poverty reduction story and that it represents this portfolio shift into more activities represents an important part of the structural change. In the first place, we show that two jobs means less underemployment. Now, these data are just on the rural areas. Average hours worked in the past month. And what you can see is that if agriculture is your only activity, you tend to work a few hours. You tend to only work a few hours, right? But if you have another activity besides that, not women, because many women who are working in agriculture are also doing a lot of household work and so they don't have the hours to add any more activities because they're looking after kids or whatever. But men, when they add another activity, they work 28% more, although this is still not full time. Those who are employed in self-employment who have one only self-employment, these are specialized in self-employment, they work the most and they tend to be doing quite well. And the ones that combine it also work a lot of hours. So if this person could have a self-employment, these people could have less under-employment. So that's another aspect of the structural change. It is, if you will, part of a Lewis transformation in that people who are under-employed are now more fully employed. Even if these are low productivity activities, the earnings, especially because they can work more hours, the earnings tend to be higher than in self-family farming. OK, so then we tried to analyze the impact on household welfare and poverty and what we did was we did a consumption regression. We didn't look at income per se because there are a lot of errors in the measurement of non-wage income. And our argument is that those errors are actually correlated with your outcome of total income So if you have total income on one side and then you have type of income on the other side, your errors in the total income are correlated with the type of income. So you don't have a regression that you can use. So we used household consumption where the error shouldn't be correlated with the type of income you have. And so we did a regression. The whole regression is in our paper that's on the website. And I'm just going to show you the part on household sources of income. We controlled for the level of education and household and the household demographics, the age of the head and the number of children, et cetera, and where you were located and whether you were in a conflict zone and some other things we used. We controlled for what month the data were collected and adjusted, et cetera. And the consumption was adjusted spatially. And so all of that. OK, so now the red numbers are the numbers that I'd like you to focus on. So the first thing is we just had a dummy variable yes or no, what kind of income you have. And we included remittances here on this one. Now what this shows is that, first of all, starting in urban areas. In urban areas wage income, this is all wage income, formal and informal. So this is your cushy public sector job, your non-cushy public sector job, as well as your day labor job and your private sector job. So here, what's interesting is that when you control for education, people actually do better, some people actually do better in household enterprise sector than in the wage sector. And that's because there are many people that don't have the minimum education to get a wage job or if they get one, they can only get a very poorly paid day labor job. So they're actually better off here. So our argument is that the development of this sector, and you see that in rural areas as well, although this wage income, non-farm wage income is pretty low. But the agriculture is quite low, especially farm wage income, quite low, although we have some in urban areas that's not significant. And so basically, and this is even better than receiving a remittance. So this already shows that to the extent that earnings are both productivity and hours work, this shows that this kind of structural transformation that we observe is making a difference. Now here, we have some of our portfolio categories, like I showed you, and here the base category is just the family farm, about one quarter of the people who are just family farmers. So here, the non-farm enterprise story again is good in rural areas and urban areas. But actually, what I call our super portfolios down here, family farm and household enterprise and everything. I guess this number's not coming out, but it's 448. Our super portfolio is pretty good. And actually in urban areas, family farm and non-farm wage income does better than just wage income. Just everybody in the household getting a wage. So that's kind of interesting. And also family farm and household enterprise does pretty well as well. So one reason why this one does well is, of course, if you have a wage earner, then it's easier to get access to credit. So I think the wage earner effect being combined in the urban areas, especially perhaps giving you access to credit, although that needs to be explored some more. So basically our conclusions are that if your expanding higher productivity sectors do not absorb a lot of labor, the employment transformation is gonna be slow or non-existent. So if, for example, you're in a resource rich economy and your industrial sector is expanding because it's primarily mining, which is not creating any jobs, you can have an increase in aggregate productivity and industry, but you're not actually getting the employment transformation. And that's one of the reasons why inequality is increasing in many of the resource rich countries. On the other hand, in East Asia, actually, in some countries in the initial period, as they were like in Vietnam, for example, when they were going from state enterprises to the more labor-intensive industries, in the initial period, actually the aggregate productivity in the industrial sector fell as employment grew very, very rapidly. And then it started to grow along with output growth. And in East Asia, during that period, you also had households shifting into household enterprise sector. What really struck me when I started looking at the aggregate numbers is that what makes Sub-Saharan Africa different from East Asia is not necessarily the share of the labor force in agriculture. It's the aggregate productivity of that labor force. Vietnam still has 50% of its labor force working in agriculture. And you see in East Asia, this household enterprise sector being an important part of the labor force, an important contributor to employment and earnings. But people don't talk it down the same way they talk about it in Africa. And I think that if, I think it's quite possible that some of the pictures that people show about, well, productivity is not expanding in these sectors. This is a problem. I mean, it might happen because the labor force is growing so big, but at the same time, the shift effect will be strong and you'll see aggregate productivity growth in the economy. And if you're only focusing on primary employment, you may miss the shift in productivity which is occurring if you look at primary and secondary employment. So our argument is that if your focus of structural transformation policy is only on the wage sector, your policy is gonna exclude the majority of households and your analysis is not gonna really capture what's going on. So we argue that it's important to combine the macro work on structural transformation with micro work at the household looking at the portfolio to see what's really going on. Okay, that's it.