 Great, thanks everyone for coming and thank you Kunal for introducing the project. So as you have heard in my presentation we are going to look at the case of Ghana. So we basically going to take stock of what are the evolution of inequalities that we can see since around 2005 and what have been the drivers of these changes in inequality. Just quickly on the data, so we have been working with the Ghana Living Standards Survey, mainly the Waves 2005-6, 2012-13 and 2016-17. So we have three waves of data that we are looking at and those are basically the ones that we could have comparable information on the labour market on. We focus on two different samples. We are on the one hand looked at all workers only dropping those that are basically not in paid employment, so dropping domestic workers appendices, but then for the main analysis we are going to focus on those that are in non-farm, that are either in wage employment or in non-farm self-employment. And we have been reweighting our data to basically correct for missing earnings information for the self-employed. We are going to be working with two different measures of routine task intensity and I am going to discuss later on a little bit more what we mean by those, but just to start with, for you to know, we are going to work with the ONET measure, which is basically the measure that has been used in the US context and Konal has been referring to this literature and then we are going to work with the country specific measure of RTI, which in the case of Ghana comes from the step survey which is conducted by the World Bank. So just a few notes on the country context. So Ghana, it has been basically pictures as one of the most notable success stories in sub-Saharan Africa given relatively strong economic growth over the past two decades and having attained lower middle income status in 2007. However, this growth has been largely attributable to the discovery of oil and gas in the country, so which we are adding to the main exports of gold and cacor. So basically we have economic growth, but we don't know yet what is the labor market implication of this given that it is largely resource driven and we also have been seeing a slowdown in growth since about 2013. We see, are we going to see a shift away from agriculture to services, but with agriculture still remaining the main source of employment and we see in the service sector workers mainly moving into the informal service sector basically, which will be partly driving the results that we are going to look at. And over the period that we are looking at, we have seen a significant reduction in poverty and some modest decline in inequality, however, with important differences by sub-period. So for the first sub-period, we basically have proper growth and a reduction in inequality and in the second period, we have more pro-rich growth and actually an increase in inequality which leveled out the previous reduction. So just what this looks like in terms of a growth incidence curve. So this is for the first period from 2005 to 2012 and you see a slight pro-poor pattern. So growth rates were larger in the bottom of the distribution and lowest at the top of the distribution and this changed quite dramatically in the second period where you have a clear pro-rich pattern basically leading to this overall. So as Konal introduced, the main question we were trying to answer in this project is what drives inequality. And we basically started with the usual suspect that we know and that we have been studying for a long time, which started with skill bias technological change basically. So we were looking at what happened to education levels and what happened to the education premier. For Ghana, what we are seeing is both if you look at all workers and paid employees in one form self-employed is basically an improvement in education levels. So we see an expansion of workers who are tertiary secondary levels of education and a reduction relatively in the share of workers with lower levels of education. And this increase in the supply of skilled workers has basically translated in a reduction of the skill premium, which is what we are seeing here for both male and female workers. And what we see is that drop in the education premium was more pronounced in the first period, which was the one where we saw a reduction in inequality. And in the second period, this kind of leveled off a little bit more. So we still have falling education premier, but the gradient is less steep. Okay. Moving to the next potential determinant, it's occupational change. So basically what we are going to look here is whether we see what Konal has described before. Do we see workers moving out of this type of middle income occupations towards those at the bottom and the top, or do we see different patterns? So what we have been observing is in the middle, you have the share of workers being in agriculture, fishery, and you see it's basically the main sector still for the labor force in Ghana, but the share has declined substantially. And where do we see workers moving into? So we do see some increases at the top. So you have here professionals and managers, which are kind of those high-skill occupations which have been increasing, but you also see some increases more at the bottom. So you have elementary occupations, you have plant and machine operators. So we see this kind of pattern across the distribution. And if you look at it at the aggregated level, which is kind of what we have on the side here, if you look at all workers, so we include agriculture, we basically see a strong drop in those low-skilled occupations and an increase in medium and high-skilled occupations if we include agriculture. So the moment we exclude agriculture, the pattern is a bit different basically. So we see for the first period actually increase here in the low-skilled occupations. So we have a situation where we have basically declining inequality, but a kind of polarizing trend in occupations. So this is one of the regression tests that we have been doing, trying to see if you see polarization in the labor market. So here we are regressing the change in employment and the change in earnings and trying to see where it occurred in the distribution. So we regress it against weekly earnings and the square of weekly earnings, trying to see where workers have been moving into. And what we find is for employment, we see basically an inverted U-shaped pattern in employment. In the first period, and then a negative relationship in the second period, and for earnings, the same. So we have an inverted U-shape and then a U-shaped pattern, which is though not statistically significant. So the first period seems more equalizing and the second some extent of evidence for polarizing pattern, but it's not statistically significant. This is what it looks like when we plot it. So here you see where workers have been. So this is a skill per center range and then here we have the change in employment and you see basically the largest increase in this middle range and some drop in the bottom during the first period. This one is for employment. So what is the role of routine by a technological change in this? And here comes our RTI measure into play. So basically what the RTI tries to capture is this different task that workers perform in their occupations. So we have a measure for non-routine cognitive analytical skills, which is basically what you imagine in high-skill service sector occupations, for example. So you have analyzing data, information, thinking creatively. So this is something that is a non-routine task that requires cognitive thinking and that's very hard to automate basically. And then we have this non-routine cognitive interpersonal skills, which is kind of coaching and developing, guiding people. So anything that has to do with communication, which is also very non-routine and very hard to automate. On the other hand, we have those more routine type of tasks. So we, for example, have routine cognitive types of tasks, which is kind of manual tasks that are very structured and relatively easy to automate. So they're more easy to be done by a machine, essentially. And same for this routine manual. So it's operating vehicles, doing kind of this dexterity work. And this is a more routine spectrum. And then we aggregate it into routine tasks in the team measure. So this is at the occupational level. And basically we follow the author and Dawn literature here and we create this RTI index on the one hand from the owner data. And this is an expert survey in the US. So you have a number of experts who basically, for each occupation, decide how routinized or non-routinized the job is according to those different categories. And then at the same time, as a second measure, we use the country-specific one which comes from the World Bank Steps Survey. So this is a worker's survey. So here we are not asking expert, but we are asking the workers what kind of tasks they're actually performing. And something that we've been seeing in the global papers that Kunal was referring to is we have been trying to see how do those differ across countries. So we use all the survey data that's available from STEP and PIAC for a number of different countries. And we regressed those against the number of country characteristics. So we looked at GDP, we looked at the involvement in global valley chains, et cetera. And we tried to see how does this vary across countries. And we found quite striking patterns. And I think it's not that surprising to things that a manager in Ghana might be performing quite different tasks than a manager in, say, Germany. And this is what we are trying to take into account on this project by not just using their own net measure for all countries. So for Ghana, we were lucky enough, so to say, to have the actual survey information, but for countries where we didn't have the survey information, we would use an imputation based on our regression framework to get closer to what might be the reality for this country. So what patterns do we see? So if you look at, so this is for each occupation at the two-digit level, we assign them the RTL value, and then you see how the shifts in occupations that we have been looking at have been changing. And what does this mean for the average routine task intensity in the country? And something that we are seeing here is for either group, all workers, or excluding the agricultural workers, we see a strong drop in RTI. So we see a kind of de-routinization, workers moving out of more routine types of occupations into less routine occupations. Okay. Doing the same polarization test essentially like what we did before with earnings, but now we use our RTI variable, S-explanatory variable, trying to see, do we see workers moving more towards higher routine occupations or not? And how does this relate? Basically, we didn't see any statistically significant effects for Ghana in this regression. However, we tried to go further, and this is what we're going to look at next. So it's just the same as a scatterplot trying to see, okay, what do we see? You see it's quite all over the place. There might be some kind of more concentration in the middle for the, in the earnings. But yes, you can, you can see why the regression didn't return any significant results basically. So it's, it's quite all over. But yeah, so we move now into, remember, this is basically the gross incidence curve that I showed you in the beginning. So the total, the black line is the same as the gross incidence curve that I showed you in the beginning, and we just aggregate it into a composition effect and the earnings structure effect. And what we can see here is that the main explanatory factor for the gross incidence is basically what happened to earnings. So this is what driving most is an equality pattern. And if we now use the rough decomposition to try to see, try to further explain this pattern that we have been looking at and what are the contributing factors in the earnings structure, one thing that we can see is basically the effect of education. So the light blue is the education premium. So we have positive gross at the bottom and a reduction at the, at the top of the distribution. So this is basically the effect of the declining earnings premium, but you also see the effect of RTI. And here's something that we are observing is that it contributed to gross at the bottom of the distribution, but you have a bit of this hollowing out in the middle. So basically in this period where we have been seeing declining inequality, the contribution of RTI was rather, at least to some extent, in the middle of the distribution inequality increasing. So this is unlikely to be the main explanatory factor for the inequality patterns that we are observing in Ghana. This was the education premium. Doing the same for the second period. So this is again the gross incidence curve that I showed you in the beginning. Again, we see that the earnings structure is the one that explains most of the changes that we are seeing. And if you look at the decomposition, we observe a strong unequalizing effect of the remuneration of RTI basically. So you see negative contribution at the bottom and a positive contribution at the top of the distribution basically. So what we basically observe is that the kind of the way how routine tasks are remunerated in the labor market contributed to the rise in inequality in Ghana in the second period. Okay, so just as a last determinant, like we've talked a lot now about education, we've talked about occupational change. And the last factor that we have been looking at is institutional factors, which should also be taken into account. And here's the main one that we have been looking at is the minimum wage. So what we can see for Ghana is that you had an increase in the minimum wage more or less up to 2012, which is a period where we have been seeing declining inequality and stagnation in the minimum wage during the period where we have been seeing increasing inequality. Just to conclude, so what have we been, what are our main findings for the case of Ghana? We have been seeing a shift toward job demanding highest guilt and less routine tasks. So a general drop in the routine task intensity of occupations in Ghana, workers moving out of routine jobs towards more non-routine jobs. The trends in inequality are still primarily explained by changes in the earning structure. What is what we have been looking at? And while the composition effect, so this is like the characteristics of workers, what they have been contributing is rather small. For the first period, we have been seeing a substantial decline in the education premium. And this has been the main driver potentially of the inequality decline that we have been seeing. So for the first period, it's basically still the traditional factors that we know of, just basically education, education premium, which are explaining this fall in inequality along with the rise in the minimum wage. For the second period, we have been seeing increase in inequality. And this coincided with the slowdown in the decline of the education premium plus in this equalizing effect of changes in the remuneration of non-routine jobs, which is what we have been seeing last. So here, this occupational change, not per se, but the way that different tasks are remunerated in the labor market has been contributing to inequality in the country. What we conclude from this essentially is that the development process had not really implied real structural transformation. So we still have a lot of workers in low productivity routine jobs. And this can, if continued, can be highly disequalizing for the country given the patterns that we have been seeing. And we are concerned that inequality could increase further if the supply side does not keep up with demand for skills. Thank you. Thank you.