 Okay, hello everyone, thank you for wider for the invitation and for the International Economic Association. Very happy to be here and present this work. So this work was originally in a wider research project on the evolution of the nature of work and inequality. And we started thinking about technology and we didn't find much in Tunisia so we started exploring other aspects and this is what I will present today. So what will be the main story of this paper is the role of public employment and wage policy. And I'll try to convince you that the revolution context of Tunisia shaped the story and this maybe creates a link between political context and economic aspects which wasn't initially the idea of the project but that we found evolving and thinking what may be the real points and we found that this one was the most important. So this is the plan of the presentation. Okay, so as I said there is a debate in the whole world on the main drivers of inequality and polarization and there are many papers on the US and other rich countries which rely on the work of David Autor and many others and which explained by the routine-based technical change hypothesis that we see this polarization where the low skilled and the highest skilled get higher income and the mid skilled get lower income because their work is routine-based and is replaced by machine computers, et cetera. So the idea was to see if this debate is valid in developing countries and some papers that will be presented in the next days will deal with this issue. So there is another hypothesis trying to explain, competing hypothesis is more structural change-based so there is a lot of work on the US and other countries including transition and developing countries trying to explain that it's not really a technology story it's more structural change where the change of sectors may explain the evolutions observed and a very important one on education premium here also in rich and poorest countries. For the rich countries we have an increase in skilled premium because the demand for skills is higher than the supply of skills and at the opposite in developing countries we see declining premium because there is a spectacular increase in education especially tertiary education but the demand for the skills is not as high and so this results in declining education premium. So the objective of this paper was to add another hypothesis to try to explain the evolution of earnings inequality which is the role of public employment because in many countries like Tunisia and in MENA it's very important the role of the public sector is really huge it's an important part of the economy it plays a role in our distribution it's used in some countries also someone talked early in the previous session about political economy aspects in countries like Lebanon for example the different communities have their everyone has its chair in public employment so public employment is really central in politics and in economics in many countries I don't know if it's the case in all developing countries but at least in the region that I'm working on it really plays a very important role and also of course there is work on this because we know that in public wages they are less dispersed because they are fixed by the government so in general less inequality and so the share of public work may have an impact on whole inequality. We adopted this paper label market lens so it's totally earnings inequality that we deal with and so we try to disentangle the role of the different factors and try to show that the most important is this public sector story. So Tunisia I take the case of Tunisia I think it's an interesting country to analyze such a story because Tunisia is a country that for almost 20 years has been characterized by almost 5% growth rate and despite this had around 15% unemployment and mainly due to a youth bulge and spectacular progress in tertiary education so the employment was mainly a youth unemployment and particularly for the graduates. The 2011 revolution that came in Tunisia and that started the Arab Spring which was also later had an impact on other countries started mainly due to this label market this bad label market outcomes but also of course political discontent, rising cronism I mean it's an economic social and political story at the same time. The consequences of this revolution at least for us here what's interesting is a very high increase in cost of security when I talk about increasing cost I don't mean just the budget I mean also the employment so the employment of security forces increased a lot because there were terrorist attacks there were a lot of strikes there were lots of demonstrations and terrorism so the country had to invest massively in security forces and this had an impact on the whole shape of the employment in the country and also public employment and wage policies to attract social peace because when there are demonstrations the government started hiring just to buy the social peace in the country. So our analysis is based on 20 years of labor for service we take a survey from 2000, another 2010 and another 2017 the methodology is based on recenter influence function and so we assess the contribution of public policies against other determinants. And so our main finding is that lower public-private wage gap after the revolution is the main driver of lower inequality in the country there is also a lower sector wage gap that I will explain later decreasing education in premier which is something very important in many developing countries and some other factors that I will explain later. So the data I explained the tax content measure are taking from the database developed by author and then we use the classification the ISCOG classification and adapt it to the national classification. So here we see the Lawrence curve of Tunisia it went to the left but before the revolution then afterwards so here which one, yeah here we see that it went from here the inequality decreased more between 2000 and 2010 then between 2010 and 2017. Here we can look at the gross incidence curve and we see that the changes happened more here and at the middle but not much except even decreasing at the end. And maybe here it's really interesting to look because we see that the genie decreased substantially between 2000 and 2010 by four points then just by two points after the revolution and if we look at the 50-90 evolution of the we see that the decrease between the 10 and the 50 was much higher between 2000 and 2010 while the decrease of the 50-90 decreased more after 2010. So we have a different evolution and the explanation here is quite clear. You see that the change in employment for the low-skilled decrease because we have less low educated people while between 2000 and 2010 there was an increasing share of the tertiary educated an increase here and decrease here and after 2010 we see the opposite. So the share of the higher educated decreases while the share of medium educated increase. This is something that doesn't happen in many countries because it's like opposite to the evolution we expect in a society that is growing and developing. So here it's like going back because we have much more tertiary educated people but the demand is decreasing for them. And if you want to look at the more detailed level we see if we look at the nine level is co-classification that we have a decrease for employment for the main high category and an increase for many of the lowest ones. And this is also translated in terms of wage where you see that the highest increase are for the low category wage. So here this is what I said earlier you see the supply of tertiary educated and the demand. So we see the big gap between the two that explain of course the decrease of the education premium that we see here but it is much higher between 2000 and 2010 than it starts being reduced less and even for women for female labor it doesn't decrease anymore. And now we look here at the role of the technical change. So here you have the evolution of earnings, here you have the evolution of earnings. So we see that for example and you have on this one the smooth the RTI. So you see here that we have the lowest RTI for the highest skills, the highest for the medium. So this is the case for everyone. But we don't see here the U inverted curve sorry that you see in rich countries. So we see here just an L curve between 2000 and 2010 and almost nothing for the others. And when we do the formal tests of polarization so we see that for employment we don't have anything significant. And when we look at wages it's just significant between 2000 and 2010. So decrease for initial mean and positive for the squared mean log of earnings. So structural transformation doesn't play a big role. We see that the evolution is not very high. Agriculture after decrease increased again. Manufacturing also is almost to its level of 2000. Sorry, something happened. The only important thing that we see is the increase of market, the share of market service. So it's premature and industrialization that you observe in the country. And the share of non-market service remains quite high. Well, this is quite similar. This one is quite important in our analysis. So here we compare the public and private share, the evolution for the different parts. And we see that the share of high skilled labor in the public sector decrease significantly after the revolution. So this is something really new. While for the private sector, we don't have this decrease. And when we look at wages, so here we have a contrasting evolution. First, when you compare the public and the private you see that here, as I said earlier, this person is much higher for the private than for the public. However, the evolutions are different. In the private sector, we see that the low skilled part, the income of the low skilled increase while the incomes of the high skilled decrease. And for the public sector, between 2000 and 2010 it was the opposite. And then after 2010, after the revolution, we have the same situation where the high skilled and the low skilled and medium skilled increase by the same level. So in terms of share, it's much higher for the low and medium skilled. So the methodology is based on, as I said, on the estimation of recent influence function. So I will present the last results. No, yeah, this one and this one, okay. So here we look at, the first thing that we look at is what are the composition effects and what are the earnings effects? And we see that here, we have that... Sorry, it's written 2010 three times. It's 2000 and 10, 2010 and 10, 2017 and 2000, 2017. So we see that the composition effect here, we have RTI has a negative effect. Age doesn't have an important effect. And public has a positive effect. So we'll see them much better for the composition of earning structure because the earning structure effect are much more important than the composition effects here. So here we see, for example, that being private sector is the highest effect in the analysis. Then we have education, experience and RTI that plays a negative role at the between 2000 and 2010, then a positive role between 2010 and 2017. So for the composition effect, I will not explain much because we see that there are almost compensation. So effects that increase like education, for example, education that has here, you see that the more you have education, it has a disequalizing effect, but there are many other effects that compensate. However, on the earning structure effect, I think what is the most interesting here, for the earning structure effect, we see that this public private story is the most important in terms of equalizing, especially between 2000 and 2010. So here you see that it decreases income for the richest part of the population and this is why it has the highest effect. On the industry part, here also it increases for the poorest part of the population and it increases income for the richest. So it plays also an important role in terms of education also plays a role here positive and here negative. In 2010, 2017, the public private explanation has a less important role because we have this evolution that I showed earlier where we have a closer difference between public and private after the revolution. And finally, the industry story is becoming more and more important and we have also education is distribution of education is quite similar now on almost all the categories and it's negative. So to conclude, we have in Tunisia a high decrease in inequality by six point in almost 20 years. We have an L shaped polarization so it's not similar to what we observe in rich countries but similar to what we observe in countries like China so they've been work on China showing that there is this L shaped polarization where incomes of the poorest part of the population increase but we don't see a similar increase of the highest part of the population. There is an ambiguous evidence on our BTC so we have this L shaped but when we look in detail we have the routine index that increases between 2000 and 2010 and then decreases slightly after 2010 so there isn't a clear effect like in developed countries where we see a decrease in Tunisia it's almost stable and many people explain this evolution by offshoring, by various hypothesis that rich countries may be exporting their routine activities to countries like Tunisia. We have an earning structure effects dominating the composition effect and the inequality change so the main reasons are, I mean reasons it's not a causal analysis of course it's correlations here so we have hypothesis at least of declining of the public private wage gap that plays a significant substantial role in declining inequalities declining trends of sector premia then excess supply of tertiary educated that pushes education premia and an increase in marginal returns to low wage average RTI jobs falling returns to experience that have been shown in Brazil to be the main factor in a paper by Shizco Ferreira in Tunisia is the fourth factor so it's interesting to study different countries because different settings may explain different aspects and finally we find also decreasing regional wage gap.