 So this paper is written with Julia Camiotti, who is a PhD student here, and Emanuel Oriol from the Toulouse School of Economics. In fact, the idea of the paper is the following, is that when I was a student in the development economics, all the main people are actually with Hirshman, Boron, Nyaume, Meir, and all those people were thinking that norms, culture cannot be an obstacle to growth and development. And in a sense, they will be moving in the process of growth, so you shouldn't be really concerned with that. I call this the organic approach to culture and norms. But then the mood change, and I think it's changed mainly when Amartya Senke came in with his writing, saying that well-being cannot be measured only by material level of living or income ahead and things like that. There are other dimensions that directly affect the well-being of the people. And I think the talk we had yesterday by Josly in about situations in San Salvador shows that it makes no sense just to measure development in terms of income level. So a second point is that indirectly, culture matters or norms matter in the sense that it may constrain the way economic opportunities can be seized by the people, for instance by women. If you have mobility-constraining norms like seclusion norms, it's obvious that you can say we have growth and expanding economic opportunities. But if they cannot be seized by a part of the population, then there is a problem in culture matters. So in a sense, since the 80s, there has been a huge focus on the issue of what we call social engineering, by which we mean an attempt to legislate to change the situation, to change the norm, to radically modify the norm, so that victimized parts of the population or marginalized parts of the population can benefit from these changes. And the UN has been in the vanguard of that. So in a sense, it's not coincidental that this conference is sponsored by a wider institute who belongs to the UN. So the first one was in 79, the convention to eliminate harmful practices called CEDO. And later, there was in 95, the so-called fact sheet on harmful traditional practices affecting the health of women and children who listed five main harmful practices, among which was female genital cutting, early marriage, early pregnancy, and female infanticide. So for the first time, there was this idea that you should do something, by the way of law, to change these harmful practices. And of course, this list is not exhaustive. What we try to do in this paper is really to provide an analytical guide about how do you go about social engineering and the impact of the law. So in a sense, you will see an echo to several things that I've been saying in several papers yesterday. And so in a sense, you can consider the paper as a reasoned survey of the literature, even though it doesn't claim to be completely exhaustive in terms of the literature. But it tries to say, OK, what kind of evidence have we to illustrate some ways of looking at the mechanism that is behind this attempt to change harmful norms and practices? Now, the first basic distinction that we have in the paper is the distinction between the expressive and the deterrent effect of the law. The expressive effect of the law is the design that the simple fact that the law is there should change the practices, even if the law doesn't have punishment mechanisms or enforcement mechanisms that is disposed of. The simple fact that it is there should have an impact. Now, the analytics of it is essentially that what you have in mind is a coordination game in which there are several Nash equilibria. One is social inferior to the other. And the custom which, in fact, involves the harmful practices is conceived as the socially inefficient equilibrium. And the whole thing is, how do you shift to the socially superior equilibrium, which is, for instance, you reject the harmful practice, like female genital mutilation, for instance. How do you shift from there? And since you adopt the idea in this literature that you have a kind of focal point story, then you say it's just a matter of changing the focal point of the people so that they believe are converging on the new equilibrium or the good equilibrium. So the idea is essentially, in the way of that Akarov initiated thinking about the caste system of India, is that people don't like the custom. In fact, they hate it. But they all believe that the others are going to follow it. And so because this pessimistic belief about the other thing and the way they will behave, people behave in a way that they themselves dislike. So they say, it's just if you can change the focal point and you bring the law, then this would be sufficient to improve the matter. So the best way to figure that out is to look at this simple game. It's a pure coordination game with symmetric payoff. And I take the example of female genital mutilation. So there are two actions that are possible, your culture daughter or your rejected practice. And the idea is that mu is a parameter that measures the benefit of coordinating. So if you coordinate on the rejection of the practice, you've got the full benefit. If you coordinate on the circumcision equilibrium, then you get minus, but you have to subtract the cost of it. And we imagine that the cost can be measured by a parameter theta. Theta is a psychological or health cost for the girls of benefits. And v is something that I will elaborate on later, which is, let's call it, an amplifying factor, which is determined by the environment, policy environment, political environment, NGO actions or interventions, whatever it is, which may increase the cost theta. If you look at that game, it's easy to see that there are two pure equilibria. There is also a mixed equilibrium, by the way. But if you focus on the two equilibrium and pure strategies, you find that both equilibrium are possible. And that's exactly the story that has been followed by a number of international organizations of NGO, is that if you can bring into the limelight a new law that emphasize the rights of the woman to reject this practice that is bad for women, you should shift from here to there. Now, also I would like to bring another consideration, which is a consideration related to the issue of equilibrium selection, is that you can say, OK, but there is a literature that deals with how you can select one equilibrium rather than the other when there are multiple equilibria available that does not rely on the focal point story. And of course, one of the most appealing concepts in that regard is the concept of risk dominance. No risk dominance is essentially saying that when you select a strategy or an action in this case, you're very concerned by the fact of what you can lose if the other party does not coordinate in the way you thought it would coordinate or you would like it to coordinate. So you take into account these losses. Because you are risk averse and there is uncertainty about what the other players are doing. Now, in the case of a 2 times 2 game with symmetric payoff, the way you compute the risk dominant equilibria is quite easy. It has been shown by Hassani and Zelten that in fact, what you have to do is not just compare what you earn under both coordinating equilibria, but also how much you would lose if the others deviate from the coordination equilibria. And you send that with symmetric payoff is very easy. Then what you see here, and that's what I want to draw your attention for, is that the risk dominant equilibrium in fact coincides with the payoff dominant equilibrium. So we don't have the celebrated stack-hand game story that the risk dominant literature has emphasized in which the risk dominant equilibrium does not coincide with the payoff dominant equilibrium. The proofs of that risk dominant equilibrium is often done in evolutionary theory. It's always in fact proven in evolutionary terms. So it's a kind of replicator of dynamics and people need pairwise in a repeated manner and they change the game. And then you see that they really in fact select the risk dominant equilibrium. That phrase, the first question, is to say, OK, if you have got the focal point story, maybe you're right. But if you have an equilibrium selection process through the risk dominant concept, then in fact you should not upset the shared decision story. It should disappear. OK, so that's the first point I want to make. No, coming to the empirics of it, this is a very important point because in fact UNICEF and so many UN agencies and NGO have in fact adopted the focal point story. And one of the biggest NGO that has worked on female genital mutilation in Africa called Tostan, to which I shall return, has explicitly adopted it and referred to the work of a political scientist in the US called MacTee who explicitly referred to the focal point story. So it has been extremely influential as a kind of story behind the way you do social engineering. And in fact, what is interesting in this Tostan thing is that the ultimate aim of the action is what they call public declaration, which means that the village come together, or about 50, 100 villages come together. And the chiefs of the village made a public declaration according to that, abandoning female genital mutilation. And they have gotten like something like 102 public declaration. And so that's why they declare that thanks to the effort they have obtained, they've got the vanishing of the practice of female genital mutilation, for which they got the Erick Newton Prize in 2002. They got a first space article in New York Times, et cetera. Now we know how to combat and fight against female genital mutilation is just a matter of having a commitment and coordination device, for instance, in the form of public declaration, that can solve the problem. Now what does the empirical evidence tell us that can help us make our mind of how to act in practice? One study is by Belmar endographers, who essentially say, looking at the cross-country evidence about West Africa, that in fact, most of the variation in the support of female genital mutilation is explained by individual and household level characteristics. So you don't see the social norm approach there. Because social norm approach, what I'm doing is influenced by how many other people, in fact, are doing. So in fact, what they show is the opposite. Where the practice is most supported, then you find that individual level and household level characteristics play even more role. So no support there for the theory. Eferson is a study that has been done in the context of Sudan. And Eferson has several findings. In fact, her starting point is interesting. She said, look, if the story is correct, we should observe that in some villages, the female genital mutilation, as the scholars say, practices disappear. In some other, it is fully implemented. Or maybe this is too extreme. But at least I should observe a discontinuity, a need discontinuity when I observe the prevalence of the practice across several communities. And what she finds is that there absolutely no discontinuity. You know, when you have the continuous, sorry, when you have a distribution of the support of the practice, what you find across communities, of course, what you find is a continuous distribution, which has no discontinuity in it. The second thing is say, OK, but maybe, since usually the argument is that you're too excited about your daughter because otherwise she won't get a proper husband on the marriage market, maybe the marriage pool does not coincide with the community. And what I'm looking at, which is the community, is not the right observation unit. But then she brings evidence that, in fact, marriage are mostly endogenous in those villages. And so there is no real evidence in Sudan that you can argue that the marriage pool does not coincide with the village community that she used as a unit observation. Moreover, when she interviewed the women and the father, the father and the mother, they all said that they don't care about marrying their daughter to a man who is not cut or a boy having a non-cut girl as a wife. OK, lastly, she has a last piece of evidence. She used an implicit association test to try to see what are the values of the people who, in fact, support or do not support cutting practices. And she expected them so to see within a community a clear divided that would reflect the association of a positive value with the cutting practices for those who support it and the opposite for those who do not support and she found nothing. Absolutely no coordination. OK, so you have just a bimodal distribution centered on zero. OK, Vogue and all have a different approach. They just have an experimental approach. They have a treatment in some communities where they dramatize the conflict between cutting and not cutting. So they bring it into the limelight and dramatize the pro and cons. And they measure the treatment outcome by the way of an implicit association test. And what they show essentially is that the treatment has an effect on the way people view female genital mutilation. And essentially, where you dramatize, people start having a more positive view of non-cut girls. The argument and the conclusion that this group of other draw is that, in fact, when you act on individual level perception, you have an effect which doesn't go well with the social norm approach. No, Camiotti has worked precisely on evaluating the impact of Tostan. And she has three main conclusions. First, that unlike what was publicized and said by Tostan itself, the impact of five-year action of Tostan in the villages and communities had practically no impact on female genital mutilation. A slight impact, yes, but absolutely nothing very massive of the difference. Marginal effect. Second conclusion is that if it had an effect, it had also, and most important of it, a perverse effect. This perverse effect is that the age at which the daughters are cut had been lowered compared to what they were before in the direction of Tostan. And she discussed various interpretations of that finding, but we know that from the health point of view of the girls, this is a bad outcome. And third, she saw that the impact and the effect in various ways, she saw that the impact of the public commitment, the so-called public declaration, isn't it? It's yours. So people just didn't take them seriously. That was just a perfunctory, you know, performance achievement of duties. OK, this has been recently confirmed by another study in Novak that essentially showed that there is a lot of heterogeneity in support for excision within communities. So it seems that the story is not credible in terms of the evidence that we have. And so the real question that we can ask analytically is, does that mean that the social norm approach, as it is reflected in this coordination game that I presented, is just wrong? And here, this is where we have to be careful because, in fact, when you look carefully at the evidence of some of this study, you see that, in fact, they insist on preference heterogeneity. That's really what come out of it. And in fact, what is really the limitation of the game that I presented is that we assume that people have homogenous preference. They all dislike social norm. The social norm, in this case, female excision. So now what I want to do with you is let's drop that. Still focus on the expressive function of the law. Following that story, but introducing preference heterogeneity into the analytical apparatus. OK, so when you can do that very easily, you just assume that the utility that the parents, let's say, I, denoted by I, have is a function, the positive component of the utility function, is a function of the proportion of the people who follow the custom. So this is the utility from following the custom for the parent. It is positively influenced by the proportion of the people in the community. That's the social norm approach. And this is the cost element. So you see this data that I was mentioning before is no idiosyncratic. So it's no more a common feature. And these is the amplifying factor. No explicitly brought this capital E variable, which means the environment, political, social, whatever you want, which can, in this context, bring either environment more or less hostile to the practice. You would say that if there is a law that is hostile to the practice, it's included in E. And the greater is E, the greater is the hostility to the practice. That's essentially the idea. No, to make things simple, we just assume that the positive element of the utility is linear, it's mu p. And so the maximum value of that positive element is mu. And when nobody follows the custom, it's, of course, 0. No, theta belongs, of course, to the range 0 to 1. But then what can you show when you follow this? How can you work that out? You define an indifferent agent for which the net benefit of following the custom is 0. And you have a critical value of theta. This is the individual form. It doesn't care between following the practice or not. Then you can compute the proportion of the people who are going to follow the custom. All those who have a theta value below this critical value. They're less aversion to the custom. That's all you should think of. Theta is the coefficient of aversion to the custom. All those people who have an aversion, which is lower than the critical value, will follow the custom. That's what this equation is telling you. And you can compute the critical value theta from the equality that ensures that there's 0 net benefit of following the custom. Now once you have done that, then you can see that there are many Nash equilibria. And the equilibrium with nobody follows the practice is always an equilibrium. And in order to bring some order into that, we restrict the range of possible equilibria. And we do that by a simple rule of dynamics that is rather standard in the literature. We assume that there is a probability that someone enters the pool of the people. And he enters the pool with a given theta, which is theta of a script A. And this theta may be above or below theta star. And we assume that he will follow the custom if the benefit of following the custom, which is a function of the equilibrium value theta star, is greater than the cost for him. And for him, it is V time. It's only just in practice aversion to the custom. Now this means that we are looking for a stability condition and that we want to retain only the stable equilibria. And the stability condition is, in fact, written here. And I will return to it because it pays off. Now let me go straight away to the important lesson that we can draw from this new way of approaching the problem of social norm approach to harmful custom. Now everything will depend on what is the shape of the distribution function of theta. You can feel that. This is where heterogeneity is waving. So this is one possibility in which your density function is, in fact, has a new shape, which means that a lot of people have a lower version to the custom. You could say they are traditionalists. And quite a number of people are averse to it. So they are progressive, if you want to say. And in the middle, we don't have many people. If that is the case, the cumulative distribution function will, in fact, have this shape. So theta is measured along the horizontal axis, along the vertical axis, and measured this curve. In fact, new f theta divided by v, which is the benefit divided by the cost, in fact, which plays a critical role because it is when this is equal to theta that you have the indifferent guide. So theta star is, in fact, corresponding to an intersection point. And now the whole question is, yes, but is that point stable? And the answer is, if you apply the stability condition, yes, it is stable if the curve crosses the identity function of the 45 degree line from above, then goes under. So if you follow that, you see that, in fact, in this case, there is a unique stable nasekelebrille that, in fact, define a positive fraction of people who are going to follow the custom and a positive fraction, the complementary function, which will not do it. And there is no other equilibrium. Now, of course, the immediate consequence, our implication of that is, here, you can bring any focal point story. There's nothing you can do because it's a unique equilibrium. There's no vertical equilibrium. No, assume that you have a situation where people, most of the people, have a kind of medium aversion to the custom. So they are concentrated around medium value of theta. Then you would have a convex function, cumulative distribution function that will cross the 45 degree line from below. Then this intersection point, which is an equilibrium, is not stable. And in fact, in this case, you have two equilibria. Nobody follows the custom or a large number of people follow the custom. And of course, you cannot rule out the possibility that you have a corner solution in the case where this expression is greater than 1. Then in this case, everybody follows the custom. So this is a case where you really have two equilibria. Either everybody follows the custom or nobody follows the custom. Then you could say, oh, then my focal point story will be two. Yes, but depending on this kind of distribution. No, you can have quite a number of other possibilities. Like for instance, this one, in which the benefit of coordination is rather low, then nobody will follow the custom. But you could have a function that is strongly OK, but starting with a very steep kind of slope. And then you will have everybody follows the custom. You can have uniform distribution, in which case the cumulative distribution function will be linear. It can be either above or below the 45 degree line. In one case, everybody follows the custom. In the other one, nobody follows the custom. So what does that show is that we don't know what is going to happen unless we know the distribution of the preferences with respect to this parameter theta, which is the aversion coefficient. OK, now because we are skeptical, in a sense, due to this possibility of having a focal point story, we want to look at another possibility that that is the deterrent effect of the law. So you look, in many cases, the law has a punishment mechanism. It intends to punish deviant behavior. And that is its role. So you cannot just use, in fact, the framework that we have to bring that aspect into the limelight, into focus. Now, what we do first is to say, OK, but then you can determine what the government will do if the government is keen on removing the norm. And then we assume, sorry, something is missing there. Yes, the government is, in fact, having a utility function that has this shape, very simple shape. This is, in fact, a positive component of this utility that is a function of the number of the proportion of people who reject the practice of, let's say, female excision. And the cost of hostility, the cost of having the law enforced, et cetera, we assume is linear. So there is a unit that I see which is constant. And what the government does is to select the value of theta star and the value of the environment, the degree of hostility of the environment, under the participation constraint that I have mentioned earlier that defines, in fact, the critical value of theta star. And when we do that, we find, it's rather easy to show, that the value of theta star is going to be smaller when you increase E. Which means since f theta star varies in the same direction that theta star, you have the intuitive result that, in fact, if the government is becoming more styled to the practice, in fact, the proportion of people implementing it, if you have an interior equilibrium, should be lower. That's a simple idea. The only thing that's important is that in order to get that result, you need to have this expression to be negative. And this is exactly we can show the same condition as the stability condition that was highlighted. So under the stability condition for the participation problem, then we have this rather intuitive result that the government can have a deterrent impact. But it's not the only way to look at the problem of the deterrent effect on the book. You can look at it from a bargaining angle. And here I get inspiration from a web that we did with Zaki, where we define a bargaining solution that has the following form that I quickly explained. So we have two parties. One is the husband, the other is the wife. No, v alpha is the extent to which the custom is being implemented. And I will consider the custom as the amount of inheritance. So to make things clear, I have a certain amount of inheritance to share. And if the custom say that the boy should inherit everything, it means alpha would be 1. And 1 minus alpha, what goes to the girl? In this sense, it will be the boy and the girl would get 0. No, f is what the law prescribes should be the share going to the boy. So here the framework is the following. We have a legislator that has the law. And the law provides an exit opportunity for the women or for the girls in this case. And the idea is that by having this exit opportunity, you enhance the bargaining power of the girl. And so you try to bring the Nash equilibrium that it is close to her preferred choice. So no, what we have, and this is important, and I will return to that here. So you have a bargaining Nash product here in which for the boy, he looks at his utility from getting the custom implemented, compared to what he would get if you would go to the law. In fact, he would appeal to the law. And then he would get f. And f is lower than alpha. By definition, we assume that the state wants to improve the lot of the women. So we are exactly the opposite situation that is the one highlighted by Jocelyn, where the state were in fact legislating against the women, or the case of Pakistan or Saudi Arabia that we were just mentioning before. And C1 is the cost of appealing to the law. And we assume that C1 and C2 can be different. And in fact, we assume that C2 is greater than C1. That it's harder for a girl to go to the court. Why? Because she's by this very act kind of antagonizing community and doing a step that is very hard for her. We can come back on that, and I will. So this is what you get. And you can easily show with appropriate satisfaction of the second order condition that you have this result. But if the state decreases the share by law going to the boy, then the custom is going to adjust to that in the light of the law. So that's a simple idea that is there. Now this threat point, this possibility to appeal of the law, act as a threat point. In fact, the cost of appealing to the law is the threat point. And the way the custom practice is being implemented is going to be influenced by what the law is. So it's because, in fact, you enhance the bargaining power of the women. OK, now if you look at this possibility, so no. Sorry, I want to add one more point. Yeah, so far I have completely left aside a corner solution, but I don't see we should go into too many details. But what I want to add is the following. Is that if you can refine this framework by doing what we have done with Zaki and Gagnal Dashev and Iman Chaharain in a series of other papers, is to say, look, let's complicate the matter a bit and assume that there is a customary authority. In this thing, we don't see there is only a formal judge. Let's bring an informal judge. So to make it look, the woman, if she wants to go to the formal court, first she has to go to the customary court. And she's dissatisfied with the fabric of the customary court. She goes to the formal court. So that's the subject. And we, like in the bargaining framework, but not like in the sort of norm approach, we have underlying the theory we have arrived at Clemens' game. So what you win, I lose. What the girl wins, or the wife wins, the husband wins, which is obvious in the case of inheritance. You have fixed pie, and then you have to decide who is going to get a higher share. So this will be the disadvantage of the other party. Now, of course, we will replicate this kind of expression. But we have one interesting result that you cannot achieve to the bargaining story that is there, which is that you can show that two radical laws may, in fact, harm the women. So the extent to which you make F lower to meet the needs of the women, or to protect the women, may have a backlash effect about a certain point. I think the strongest framework in which you can show that, and in fact it's interesting compared to the discussion of John, is that assume that you don't only have customary authority, which is facing a trade-off between, yes, this is the custom, I like the custom, my preferred outcome is the custom. But if I apply the custom to city, I lose a lot of people who are going to deny my jurisdiction and go to the formal court, which is an insult to the social prestige and power that they have. No, we have a formal judge that is facing a similar trade. Instead of saying the formal judge implements the law automatically and mechanically, we say, no, no, the story is more complicated. The formal judge has also a trade-off. He has also a preferred outcome, what he likes or doesn't like. And in fact, he is kind of balancing his preference for a certain outcome against what the law press is going to do. No, he has a positive utility of applying the law because it is his duty to strictly apply the law. But he doesn't like, or his utility decreases, if the law is going too far away from his own preferred judgment. Then if you assume that judges have idiosyncratic preference, then you have many specials above which, if the law goes too far, they just stop implementing the law and prefer to implement his preferred outcome. Then you can show very simply that in that case, in fact, the expectation of the wee members in this case, of the value of the judgment that they will get, may be decreased by a law that is too radical. So this kind of initiates a whole discussion about, OK, it's good to talk about the law. But how radical do you want the law to be? Is really a question that has been largely ignored. I've seen a recent paper, by the way, by Asemolouet Jackson, who exactly reached the same conclusion, but with a completely different setup because in this setup, you have decentralized matching, pairwise matching of people with the possibility of whistleblowing. So the idea is there is not only the law with a positive, but less than one probability of detection of whether you are actually following the law in your actual behavior, let's say in inheritance matters. But your partner, if you behave too differently from you, may in fact denounce you to the state for punishment. Then you will be put. So what they show is that, again, if the law is too radical, then the extent of the incidence of whistleblowing decline. And in fact, you may have a situation where the interest of the people whom the law wants to defend is, in fact, decreased. No, what about the evidence? OK, so the evidence, in fact, is not an ambiguous or complete. And here, I guess I have to go rather quickly. So let me just mention the fact that most of the evidence is on India, on the impact of the reform of the Hindu succession act. And people try to see whether the better protection embedded into this reform have got an impact on the way inheritance is being shared. There is the first paper by Dininger who is a very optimistic thing. The law has really increased the probability of women to inherit. And even more than that, the education level, this paper has been severely contested by Roir, and I think on firm ground, on, in fact, identification ground, that shows something very interesting. You see, the impact of the law has not been, in fact, to make women more likely to inherit. But as increased, the compensations made to women for the fact that they do not inherit. And this is in the form of Doherty's or in the form of better education, higher levels of education. I like this because it's very much, in fact, connected to the theory called the magnet effect theory that I was illustrating with the customary authority, where F is decreasing, alpha is decreasing, but never to the same extent. So you know, it's a kind of compromise solution that you find. The law is saying something. It is not implemented strictly. But still, it has an impact, and the impact is positive. So that's what the papers say. Then a series of papers were more saying that the law can have perverse effect, like increasing female sign mortality, rosin bloom, increasing barotry at all, female fetishite, and, understand, genical paper, which has been quoted yesterday, so that, in fact, the suicide rates of both husband and wife can be increased as a result or have been increased as a result of the law, which, I argue, is a result of the fact that, indeed, there has been enhanced bargaining power of women, but this enhanced bargaining power has, in fact, created more conflict within the household. And so, for instance, can bring more domestic violence. There are series of papers on Africa. I won't go into the detail. I think I have no time. But essentially, we can conclude, like we do in a paper with a survey paper that I wrote with Katin, that, in fact, the evidence is still very small, that it concerns a very limited number of countries, essentially Kenya, Ethiopia, and India, massively India, no other country, that the identification problem, in their hands, in measuring the impact of the law, are far from being satisfactorily solved in many of those studies, and that the time frame use is excessively short. So in fact, we still don't know very much what is happening. And one of the conclusion is that maybe we need something else than the law to, in fact, change the custom. OK, these are all the more so, because here we have assumed that the law, people know the law. The law is credible. But there is absolutely no evidence of that. Maybe you can improve the knowledge of the law by the people that's feasible. But still, you have two problems. You have the problem that people need to be convinced that the law is considered as credible by the other people, not only by myself. So again, the issue of credibility of the law is a big issue. OK, now, I have to go to the most important thing. And just say that it's very easy to see that one of the way of doing is by changing preferences. If you now make an assumption about the value of E, which you consider as given, so that this function measured along the vertical axis is transformed under the cumulative distribution function, then the reasoning that you make about the equilibrium is unchanged. And you have a departing value of 0, a value of 1 here, which is the assumed value of the cumulative distribution function. And you can, in fact, imagine all kinds of way you change preference and see how the equilibrium is evolving. So you see that if you start from a unique equilibrium where everybody follows the custom and you succeed of changing the preference of the people so that this curve is transformed into the dotted line, then you come to the very, very equilibrium where nobody follows the custom. But in fact, if you look at what it means, you are changing the concave into a convex shape here in the beginning portion, which means that you are working on a people who are very adverse to the law. That's all you're going to say, yeah, that's unrealistic. This is the people who are the most traditionalist. So you better work on the people who are individual. Then you get a marginal increase in the people who reject the practice. But you cannot have all or nothing results. So what we want to say is that if you just follow this kind of framework, you can explain the kind of empirics where you see that you can have a slight improvement, no massive improvement with many distribution functions. This is what you would get. OK, after the last point I want to make here is that the density function can be interpreted in a varying way, which is the following. You can say, yeah, but people do not have all the same influences. Then your density function measure the influence that each individual has. So you continue to have a continuous mass of people where they're varying influence. And the f theta function, and therefore the community function, is translating this important. OK, but still then you know, of course, the result is intuitive. You don't need much elaboration to know that work on the influential people. The whole question is, what do you mean influential people? And here I think the theory of social network is very useful. Because when you say, what do you mean by important people, then you have to see whether do you mean the people with the highest number of directing for the people, or the people with the highest direct plus in directing to other people, or the people who have links with the most important people that are around them. So in the literature of network theory, you may have a big concept of degree centrality, cut centrality, and eigenvector centrality. OK, so the last point, and here, of course, I have to be very short, but it's still an important part of the paper, is the role of the economic environment and economic opportunities. Of course, if you stick to the social norm approach, the impact of increasing economic opportunities, in many cases, may be to in fact increase the hostility to custom, which are going against progress, economic progress, like chasing economic opportunities. Then we know what is the effect, even in the context of social norm. But you can use the bargaining framework or the magnet effect theory framework and then say the exit opportunities are better, so the threat points are increasing, becoming stronger for the victimized party. OK, so that's easy to do. No, no, what does the evidence show? Oh, yeah, that's a game that I will not go into the detail, but which is in the paper, which is a game about the rule, the traditional rule, according to women, do not inherit. And you want to know the watch condition. You can have a shift to an equilibrium in which, in fact, both boys and girls inherit. And let's assume that in the initial situation, women do not inherit, but divorce is very low. So stability of marriage is guaranteed, so she can safely rely on the land of the husband. And now you move that and say, but assume that for a certain reason, the divorce rate is increasing. That security is no more there, so the payoffs are changed. And in fact, the paper details the utility for instance that are behind this payoff structure. But you start with an equilibrium in which parents will select this equilibrium, 55, which in fact is the unique equilibrium in these days. And if you change the payoff of coordinating or sharing the inheritance wealth among boys and girls, you have 77, then you see that now you start having two equilibrium. They can still stick to the old custom, but they can move to the new room. But what is interesting is that in this case, then we have the stuck-hand dilemma. If you compute the risk-dominant strategy, it is different from the payoff-dominant strategy. So the risk-dominant strategy is this one. So what do we want to say about that? That theory doesn't tell you that, as the social norm approach was kind of a hinting, that just a continuous increase in A will have an impact. No, it may have no impact at all. People can stick to the old equilibrium. Because this is a question. In many cases, you see all these customers persisted for such a long time, food binding in China, 1,000 years. And suddenly, it disappeared in two years, very abruptly. And so how do you explain that? You know, we had this kind of situation where it may be the case that the payoff research that people still selected the tradition, even so. You know, the superior, the efficient equilibrium is no more that one. And it may happen, you can have another pay-off structure where the opposite happens, where the unique equilibrium will be coordinating on the new one. In the case of food binding, and there is a whole discussion about that for early pregnancy, early mariage, domestic violence, we have a look at the empirical evidence that shows that in many cases, in fact, the economic incentives and the economic environment changes a lot. To last point, because I've exceeded my time, is that, in fact, you know, it is always the case that there is a resistance that has to do with the power structure of the household. OK, I was looking recently at a paper about the granting of property rights to women in movable assets in the US and in England in the second half of the 19th century, where both countries simultaneously abandoned the coverage law based on the common law to allow women to hold and to inherit movable assets. And it's interesting that in the debate before the English parliament, they were saying that we should oppose this reform of the law because this would reduce women domestic authority and, I quote, create discord in their homes. Expanding the economic autonomy of whites would wantonly interfere with the relations of married life and revolutionize the whole household system. So you see, it's really what we are used to when we look at developing countries today. But the interesting thing is that in spite of those resistance, the new law went through. And why? Because the economics incentives were so strong because it was really a wish to diversify out of agriculture and to its industry that, in fact, the economic proponents of the new law in fact won it. And my second and last remark is the role of culture that is a whole section in the paper that essentially saying this, so far we have talked about social norm, but not really about culture. And this last quote is hinting that culture is very important. And culture, in fact, refers to identity. And a key problem is the following, is that people may have their custom norm, not because they like them or don't like them, because they feel this expression of my identity. It's my culture. And this is very important. We bring evidence from Egypt and we bring evidence from Senegal, where the people use the following kind of words to describe the attempt to criminalize the traditional practice. They say it's cultural aggression. It's a manifestation of Western imperialism, whereby the West want to destroy our culture. If you're touching on those custom, it's our whole culture which is on the thread. And that's what the Westerners want, which shows that even when the government, like the Egyptian government, is enacting a new law that burns female genital systems, which is too immediate. It doesn't mean that it was not done under international pressure, but in fact, it doesn't coincide with the forces of enforcement that John was highlighting. So this expression of cultural identity is important. And in a sense, we discussed a number of references to some paper, even for US marriage and law. That shows that it is important. I think someone pronounced the word masculinity yesterday. This is part of it. You are touching on things that are much deeper than the practice itself. It is a symbol of something else that may be had to do with identity, but also distribution of gender growth. So this is where I end up here. I'm sorry I've written a bit long.