 So we will stick into this inequality of opportunity framework. Today I'm talking about a country that some people in the room here know better than I do. So it's with all humility that I will be talking about Egypt. Why? Because Egypt is a good player on this. They have data. I'm originally from Lebanon. No data, so I cannot talk about Lebanon. So I will talk about gender stratification, social class at birth, and inequality of opportunity. So our objective here is to look at the evolution of inequality of opportunity and to split it into two components. One due to gender stratification and the other one due to social class at birth. In order to do this, and I hope you'll cope with me, we need to present a measurement framework. So there will be some methodology. I will keep it as soft as possible. I will display the math, but I will talk in words. And I've prepared a few graphs to explain what I have in mind. So if we think about stratification, sociologists have been talking about stratification for a while. In economics, it's relatively new. So if you look at the Journal of Economic Literature, June 2022, so there is a complete issue on race and economic literature. And within this issue, there are many papers on racial stratification. And if you think about racial stratification, it comes from the black radical tradition in the U.S. And it has been imported into economics by Sandy Darity, who proposed a new field that is called economics of stratification. And this field has two components. One is theories, how these stratifications emerge, how they are explained. And another one is measurement. So I'll stick to the measurement part, okay? Because it's important to be able to measure in order to test the theories. So Darity argues essentially that unless we make strong assumption that marginalized group behave and they make choice that go against their own interests, we should consider differences on identity or inequality that are based on identity as unfair. Stephanie Seguino borrowed this idea of racial stratification and brings it into the gender sphere. And she defines, as she talks about gender stratification, and she defines what in her mind constitutes gender justice in this idea of gender stratification. And she says we achieve perfect gender justice if the probability of achieving each social position or each social outcome is exactly the same for each identity group. We've been talking about inequality of opportunity all the day. For those who were in the same session than I was, this is linked with inequality of opportunity. So let's go back to Romer, 1998. So the first paper that brought this philosophical idea of equal opportunity into economics. And in this paper, Romer distinguishes two types of inequality. I should not insist on that, but one that is due to responsible individual decision on effort. This is called accountable effort. And this is considered unfair. And the other inequalities, no, this is considered as fair. Sorry, I'm inversing. So there is a reward principle to effort. And inequalities due to the birth lottery. And this is considered unfair and inefficient in addition. Romer in this paper shows that if the distribution of residual luck is independent of initial circumstances, if it's the same for all initial circumstances, then perfect equality of opportunity. And this is the important part. We choirs that the conditional quantile function is the same. Conditional on circumstances of, sorry, at birth is the same for any possible circumstance at birth. So if we look at this condition and we think about Seguino's condition, if we include identity, gender identity or racial identity into the Romer framework, both are the same, essentially. So what do we want to do? We want to provide a framework that allows us to measure intergroup identities in equality related to identity-based stratification. So now I'm moving out of gender. And I talk in general about identity-based. And when I will come back to Egypt, we'll look at gender stratification. So what we do here, we build the link between, sorry, this literature on measurement of inequality of opportunity and the literature on the economics of stratification. What do we do? We provide this measurement framework and that can split this identity into identity-based stratification and the other part that is due to social class at birth. And we apply this framework to study the evolution of gender stratification in Egypt. And if you can cope with me, bear until the end, it's very interesting. What we don't do, we don't offer a theoretical explanation of the emergence of these stratifications or of the change that I will present. I will present some changes. If you have some theory to explain that, I'll leave it to you to do that. I'm not doing this. I'm not and we're not because my quarter is in the room. I should not say I because she will get angry at me. So we adapt the inequality measurement framework to allow for this kind of decomposition. In order to do that, we borrow on the idea of Temkin. So in 1986, Temkin publishes a paper in philosophy and public policy. And it defines inequality as an aggregation of complaints. So we'll use this idea and what Temkin says, it says individual who have the same merit should have the same outcome. We bring this into Romer's framework and how do we define merit? The same level of responsible effort. It's very simple. It's kind of the best framework to put this. So here, keep in mind that we are not focusing on individual outcome. As you will see, once we start the estimation process, we don't care about individual outcome. We care about this conditional quanta function that we are trying to recover. We develop distributional properties that these indices should obey and derive for those who know us a little bit better, of course, dominance condition to allow for the ranking of the distribution without picking a specific index. And the empirical application is our contribution. So we study the evolution of gender, stratification and equality of opportunity in Egypt. And it's fun. The story is fun. So the measurement framework, so we borrow on Romer. So essentially, Romer has a production, an income production function that depends on initial circumstances. We have two special ones, age and identity group. The important one is this one. We split the society into a dominant group and a marginalized group. Age is just there to compare the individual outcome with the person in the same age cohort because we don't want to compare 25-year-old income with someone of 50 years old. The difference may be justified. And so the big and this assumption is very important and it is Romer assumption. This production function that is a function of the raw level of effort, age, all other circumstances at birth, group identity and residual lock, this function is strictly increasing in effort. And this is one of the key assumptions. Romer argued that individuals' ability to produce this raw effort, like what will generate an income? Investing in education. But investment in education, as Rene explained, there are already inequality of opportunity and access to education. So what Romer says is part of this effort, there are inequalities of opportunity and the accountable effort will be the position in the cumulative distribution of raw effort conditional on initial circumstances. If and it shows that residual lock is statistically independent of initial circumstances, then the position in this cumulative distribution of raw effort and the conditional cumulative distribution of income are exactly the same. You can define raw effort as the position in the conditional cumulative distribution. And this is fun because when we will talk about quantized function, so the P or the Q that you find in the quantized function is now interpreted as raw effort once it's estimated. It's important to note here that in the empirical application, it's impossible to identify the effort level associated with one observation. Why? Because effort is unobservable and residual lock at the individual level is unobservable. But we can identify and estimate and you all know that the conditional quantized function and this is what we are going to do. What is inequality in this framework? So Temkin's give many reference points in order to define a complaint that is the one that we are picking is in comparison to the best of person that has the same merit. So what is this best of person that has the same merit in this model? It's essentially the upper envelope of all conditional quantized function with the graph it's easier to explain. So assume we have a society with four different initial circumstances. Here you have the four different quantized function and as you see between zero and E1 the green is the first. So it's the quantized function of the first group that is at the top between E1 and E2. It's the second group that is at the top and between E2 and E1 it's once again the first group. The envelope is the dotted line that takes always the maximum value. It's easier to program anyway with this assumption in your R code. How do we define the complaint? So here I'm assuming one particular level of effort in order to explain. It's essentially the distance so for an observation or if you have initial circumstances A, X and G that are particular value. If this is the quantized function and at E1 here the distance between your quantized function and this upper envelope is your complaint. We transform this in relative term and of course we aggregate over all possible level of effort. We assume a weighted average. We don't make any strong assumption on this function. We just say this is non-negative and it should sum to one. And think of whatever. The standard average is one of these functions but you can put some weight on the upper end, on the lower end, it will obey that. And the index of inequality of opportunity is just the average of that in the society. We define by omega here the set of all possible inequality of opportunity index that you can imagine that fits into this framework. The interesting thing if you define the things like this you can split this distance here into two distance. So assume that you are in the marginalized group. Then assume you produce another quantized function, conditional quantized function. Everything is the same except group identity. Then you have another quantized function. So the distance here in blue is identity stratification and what remains is social class at birth impact on inequality of opportunity. And then you can decompose your index into two components. What we do in the paper so we define curves. So we love stochastic dominance conditions. So we have curves. So we have the complaint incidence curve that gives this average value of the complaint at each point of effort. And the interesting thing if they do not intersect the curve that is the lowest has lower inequality of opportunity. So you have this condition for inequality of opportunity overall and you have similar condition for social class at birth and identity stratification. Of course, sometimes this is not sufficient to get a ranking. We say let's assume a little bit more structure to this weight function. We consider two possibilities. So one possibility would be to put more weight on lower level of effort. We call this pro-poor because we didn't find any good name here but it's not necessarily we're not talking about the actual poor. We're talking about people that put a low level of effort for all possible combination of initial circumstances. If you have a pro-poor view this function will be decreasing. If you have a meritocratic view and what is an meritocratic view remind all these stories about glass ceilings and things like this then you have a meritocratic view then you have more weight on the upper part, on high level of effort but you're still averse to inequality of opportunities. And we define also curves that if they don't intersect then the curve that is the lowest has the lower level of inequality. I go very fast because I don't have a lot of time. We use standard kind of dominance test I don't want to enter into this but if you're interested we can send you a paper in a few days we're finishing polishing it. Now let me talk about the Egyptian context so it's enough for the math and the method now I'm coming to the real idea here what is the story I want to tell. So correct me if I'm wrong but there's a small history of Egypt so you have Gamal Abdel Nasser and it's the period of Arab nationalism it's a centralized economy then you have Anwar Sadat that starts liberalization it's accelerated under Hosni Mubarak and this is the important story here it's accelerated and it takes more and more the form of crony capitalism so an economic system where if you're the friend with someone in power then you get rich of course if you have an inequality of opportunity view this is not what you aim for 2011 the Arab Spring if you look at a paper by El Rafayi and Vol 2020 support for redistribution in Egypt increased from 22% in 2008 to 59% in 2020 so there was something here in the public opinion then you have a short elected government Mohammed Morsi, Muslim Brotherhood and not directly but related to and then after a few protests again a new regime, military regime Abdel Fattah al-Sisi so this is where the story is framed we use the term where the story is framed we use data that is made available by E.R.F. they're doing a fantastic job I'm not an employee of E.R.F. they're making these data available to researcher and I'm excited and so we use four waves of the Egyptian labor market panel survey we use outcome variable labor income and we have a set of initial circumstances education of the parent type of employment of the father, region of birth, year of birth and gender the ELMPS has a very interesting question it's probably because of Raghiaz I'm not sure they have information on the labor the activity, the labor kind of activity that is called unpaid family worker and it's an important one in the case of Egypt and Arab countries labor economists so understand Assad and Kraft who studied the labor market in Arab countries have this extended definition of the labor market that include this unpaid family worker if you take this definition it allows you to bring women into your study of inequality and of course the interesting thing with the Chernozukov-Fernand-Delver and Mele approach it allows you to estimate these distribution with probability mass so the first thing that will come to the mind of someone is okay include the non-labor market participant at zero, we don't do this here and we just refer to the authority Assad, Kraft, Romer and Salih is honey in 2018 they explained that this is mainly rich people and if you do this you distort the story of inequality of opportunity we follow their lead this is the complaint incidents curve 2012, 2006, 2012 remember if one is lower and here you have the confidence band and we run a statistical test also on this you have lower inequality of opportunity in Egypt in 2012 compared to 2006 it was going down before the Arab string this is consistent with Assad et al also we compare all year so essentially the story will go quickly when you have omega all indices if you have omega m meritocratic view indices omega p pro pro view indices so if you look at the general story between 98, 2006, no change, no dominance it means you pick an index and tell you it increased, you pick another one it tells you it decreases between 2006 and 2012 so if you compare 2012 with 2006 you have a significant reduction then if you compare 2012 and 2018 no dominance so one index will tell you it's increasing the other one it's decreasing let's look now at gender stratification 2012, 2018 it's increasing, pro-poor view the complete incidence for curve it's not significant but if you take a pro-poor view it's increasing if you take a meritocratic view it's increasing if you look at the preceding period 2006, 2012 it was going down the complete story between 98 and 2006 no dominance between 2006 and 2012 it's increasing it's decreasing and between 2012 and 2018 it's decreasing and in 2018 it's still higher than 1998 which is bad if you decompose for social class at birth it's going down between 2012 and 2018 down, we have the same dominance test I'm accelerating if you look at the story for men since inequality of opportunity due to social class at birth it's going down for all years and the men, the dominant group are only impacted by this channel the story is very good for men it's going down all the time and even between 2012 and 2018 women are affected by both channels so we are unsure what is the impact on women as you could guess no change between 98 and 2006 between 2006 and 2012 it was going down and between 2012 and 2018 it's increasing confusion so you give me one extra minute so we develop a framework to measure inequality of opportunity and to decompose in gender stratification in social class at birth component we do this, we provide the dominance condition and we use standard econometric method to estimate and run our test we offer an empirical application on gender stratification in Egypt this 20 years period and let me summarize is interesting because we had this Arab Spring and the Arab Spring the main request I explained it was we want more equality and I'm not saying the government I'm not saying anyone the social structure changes in a way that faced with a demand for more equality it seems to adjust by decreasing the burden of inequality of opportunity on the dominant group by increasing it on the marginalized group so this is a particular story but I think it's a very sexy result and of course this can be applied to any other identity marker if you have the right data and if you have many identity markers with a good data set with enough observation you may want to look at intersection of these identity so race and gender together and that's all I had to say thank you for your attention