 Thank you very much for the invitation. It's an honor to be here. My relationship with WIDER goes back to the 2002 when I came here to present in a similar conference on the topic of violence and human capital accumulation. Since then, I have been following the work of the WIDER Institute with great interest. I think that the Institute is at the center of very important policy debates. I think that the level of discussion is with very high quality. And as Finwas mentioned, last year I came as a visiting scholar during my sabbatical. And during those three months that I spent here, I had several experience that makes my time here very interesting. The third year conference anniversary of WIDER was a very important point of discussion of development. And something that struck me about that conference was the level of discussion and the openness of the discussion. There were a lot of different perspectives and different scholars debating very critical issues in development. And in that sense, in that spirit, I want to present here three ideas that are going to shape my research in the next years. And I think that there are three ideas that are at the debate of policy in education. My perspective about education in developing countries is somehow a little bit different from the perspective that some people are putting forward, mainly the idea that the question of where does the investment in education went, the economies are not growing as they should be. And I always think, OK, what is the counterfactual? So that is something that I think is that I am more positively, my thoughts are more positive with regard to education and what has happened in developing countries. I'm going to discuss that more in detail. But the three ideas that I want to discuss with you are around the following three issues. The first one is the production of those emotional skills. I think that a lot of people are thinking social emotional skills is one part of the production of the school that is important. And I think that I totally agree with that. I want to give you some ideas of how to think about that. The second issue probably is more important for developing countries right now is the issue of heterogeneity. And the third issue is that this debate about the misalignment of incentives in education systems. In order to give you a preview of what I am going to try to convey to you, I'm going to present the ideas in a very broad sense. And then I am going to go type deep in these three ideas. So the first idea about the production of those emotional skills. So what I am trying to claim here is that we need a more clear framework in the formation of social emotional skills at the school. So I believe that it's important to differentiate different type of those emotional skills that are being produced. And I think that it's important to define different periods of time in the production of social emotional skills. As it is right now, a lot of researchers are thinking in a very broad way social emotional skills. And let me say some researchers are saying, OK, we try to understand what is cognitive skills and the rest of the skills. And they pull the other skills in the rest. And I think that is not helpful. And we need to start thinking more about a more organized framework. And I think that there is a dynamic aspect in the formation of social emotional skills that is important. The second point in social emotional skills that I want to make is that there is an important complementarity or sustainability situation between the household and the school. And we need to understand what is the relationship between household production and school production in these skills. The third idea, the second idea, is the idea of heterogeneity of the skills. So a main challenge of teachers is receiving a highly heterogeneous class. And something that is happening right now in developing countries is that the countries are increasing enrollment in an important way. There is no debate about that. And some people are claiming that the quality of education is flat or they are not improving. And I think that at the root of that problem is the following. We are increasing heterogeneities in classrooms. And the teachers do not have the tools to cope with that heterogeneity. And I think that the systems are not conducive for such heterogeneity to teach that heterogeneity. So that's the second idea that I want to discuss. The third idea is the one of incentives. Some researchers point out that the lack of proper incentives are at the root of the quality problem. And other type of researchers are arguing that the problem lies at the lack of proper tools. And I think that those two views has to be complemented. And I think that it is very important to understand that, for instance, changing the incentive structures without changing in a fundamental way, pedagogy and content of teachers are going to not give what we want. And the second point about incentives is that even if we will be able to do the right incentive programs, I think that the data that right now developing countries has are not suitable for the right incentive mechanisms that we want to induce. The third point in this incentive debate program is the point that the bulk of these policies, of incentive policies, are going to the stock of current teachers. And I think the most important investment right now is in the future of the flow of the new teachers. So I think that we need to start also thinking very heavily on how we need to change the new people that are getting into the system and how we need to change the training that they are receiving. And finally, given that we have some limitations in interventions in the supply side, what do we need to do? I try to argue that a promising venue is to change the families and students and to use policies that change families and students as a leverage to change the education system. Around these three big ideas is the discussion of my presentation is going to be. So social emotional skills. So we know that schools produce a large range of skills, both cognitive and social emotional. And research has been showing that social emotional skills can explain an important amount of labor outcomes and labor and other outcomes later in the life of individuals. This work by Hegman and other researchers has been very important and has been very active in the last 15 days. Moreover, we think that most people can acquire these skills. However, it is very clear that child's environment determines both cognitive and social emotional skills in a fundamental way. And difference early on can result in a striking difference in the endowment of individuals. And that is so far what we have been learning. But I think that there is a fundamental problem, which is the lack of a unified framework in organizing and defining these skills. What do I mean by that? What I mean is the following. So let's say that in a framework of social emotional and cognitive skills, you have two main outcomes. You have the household, and you have the institution of the school. And you have different periods in the life of the individuals that go from minus nine months to higher than 11 years. And then a fundamental part of this period of time between the nine months and five years happened either at the household level or at the childcare centers. In this period of time, the creation of executive functions are critical. And we know that executive functions basically are mental flexibility, self-control, and working memory. Those functions are critical. And those functions have both components of cognitive and social emotional skills. The psychologists have been researching executive functions very, very deeply. And there are very clear guidance of how do you produce this? You produce this in households that are safe, in households that have routines, and households that promote creativity, and that provide affection. Those executive functions are produced mainly at the household level. Once they are produced, the kids are right to the schools. And these work as an input of the school system. And if there is huge difference in kids in these executive functions, those differences are going to be reflected in the work of the school. These executive functions work as an input. And then the schools start trying to do cognitive skills. And in the first stage, they try to do reading math. And in the production of social emotional skills, they are trying to do attentional to regulation, great discipline planning. But the first point that I want to make about this is that these executive functions work as an input. And there is a huge leverage if we can change early on the trajectory of the kids such that they enter with a more homogeneous skills in the education system. Then the education system has to produce more skills. And basically, there is the production of higher cognitive skills. And there is the production of other social emotional skills, like locus of control, state of mind. And then in 11 years old, there is another opening in the brain. The brain is very flexible and very vulnerable in the first stage, minus nine to five years. And then at 11 years old, when people start getting into adolescence, there is a new transformation of the brain. And in those period of time, the relationship with other individuals is critical. And the creation of skills like empathy, cooperation, and sense of belonging is critical. So I want to make three points. First one is that an intervention that shape executive function early on will have large returns. And it's not the typical argument that you have to shape the executive function because later on, this can have the high returns. The argument is that you have to change the executive functions because that is one of the main drivers of change in the education system. If you are able to homogenize and to increase the endowment of certain skills before the kids enter the school, that's an important leverage in the quality of the school. However, there is no clear way that we know how schools are producing these social emotional skills. Thus, it depends on the climate or the culture that foster social emotional skills. It is produced at the school level or it's produced at the classroom level. And I think that that's another point in which we need to have better understanding of how these skills are being produced. There is some research that shows something that is very interesting is that teachers that produce the cognitive skills not necessarily at the same ones as the teacher producing social emotional skills. And that, from the policy point of view, implies something that is quite interesting. We don't know if we need to couple certain type of students to certain type of teachers or we need to rotate kids during the school life with different type of teachers. And another point is even if schools are not producing cognitive skills, they may be producing relevant social emotional skills. This is something that is very interesting because in this debate, there is a debate about conditional cash transfers, which is we are bringing kids to the schools but they are not learning in terms of cognitive skills. And the argument is here, the argument is they may be not learning in cognitive skills but they may be learning in non-cognitive skills. There are some evidence pointing towards that. My colleague David Deming have some paper in which he showed that during the school trajectory there is no difference of cognitive skills between one person intervening and the other one not. And later in life, he showed some important effects and one mechanism can be to the production of social emotional skills. And I have some work in conditional cash transfer in Colombia that shows similar results. Later in life, people will have some effects even though we were not able to detect effects on cognitive skills earlier on. So that is another point is that the schools are producing a wide range of skills and we need to start thinking in a broad sense not only in cognitive skills. An area of exploration is of course interaction between household and institutions in the production of social emotional skills. And here there is this idea that in a lot of developing countries, families tends to believe that once they give the kid to the school their work stops because the school will take the load of the work. However, there is a clear relationship in the production of social emotional skills between what the household is doing and the institution is doing. A heterogeneity. The problem of heterogeneity is very straightforward. Class rules varies with heterogeneity but in cognitive and social emotional skills. And it is more costly in terms of actions and time to reach students with low levels of these skills with low level of endowment in both cognitive and social emotional skills. And moreover, the curriculum and the materials at the national and subnational level sometimes are better suited for students with higher skills endowment. For instance, the textbooks that the systems are producing are more suitable for students that have higher endowment. And there is a correlation between these skills and the home environments. There are some home environments that do not have the right estimulations that don't provide the right opportunities to kids and basically kids that goes to those home environments have lower endowment in these skills. In that regard, several education systems are leaving behind students for low income and vulnerable households. This is an idea that Bernardi and Duflo are insisting and I think is important. There is some indirect evidence of this idea. The indirect evidence comes from, for instance, this is a clear example in the provision of textbooks. There are some experiments that provide textbooks to one school, not to the other one. And basically the researchers don't find any difference in the performance of students going to school that receive the textbooks versus the other one. And one of the hypotheses that they present is that teachers concentrate effort in high achievers at baseline students. So the textbooks are right to the classroom and the teachers start working with the kids that were able to read the textbooks. And in that sense, the teachers adopt the pedagogy and adopt the input to help a certain type of kids. And that is one indirect evidence of this type of heterogeneity. The other evidence comes from explicit policy that targeted students by a skill level. So we have experiments in remedial education. I am just giving examples, for instance, Bernardi computers that adapt to skills or tracking. All these policies, what they are trying to do is they create exogenous variations in the homogeneity of the classroom. And one potential mechanism that a lot of these researchers are having in mind of the effects of these policies is that teachers can adapt better the pedagogy when they have homogenous classrooms. Of course, there is another perspective in which you change the tools that the teachers have. So the idea here is the following. There is one idea that if you change the incentives, the teacher will respond. The idea, another way to look at the problem is to try to change the teacher so that the teachers can respond to heterogeneity. And those are two different approaches. So why is this program is important right now in developing countries? This problem is important because a lot of developing countries had increased enrollment rate in a substantial way in the last decades. There are several policies that increase enrollment for instance conditional transfer and these policies are targeted usually to low income and vulnerable populations. And basically there are new enrollment in the system that are coming from these households. And a lot of people are pointing out these two facts that are happening in developing countries. The first fact is that enrollment is increasing but quality is either flat or not good. And they are saying that one hypothesis that explain that is that, well, you are bringing to the system kids that are more difficult to teach. This hypothesis that I am trying to put forward is different. Is that yes, you are bringing kids to the system that have different endowment of skills and those kids are very difficult to teach in a classroom. And the teachers do not have the right incentive to teach them. And we need to change that. In the macro perspective it's important to bring the private sector because the schools that are receiving this population may be private or public. There are two type of systems one in which you have a system in which there is complete segregation, private schools attending people with high income and public schools attending only people with low income. This is a case of complete segregation. However, there are some countries right now in which we have private schools attending low income populations and high income populations. The cases, for instance, India and Pakistan, this is happening. In any of these two cases, the heterogeneity in the school is increasing and that is the challenge that we have at the micro level. If in the classroom heterogeneity is increased, the teacher has the following decision. Either teach to the mean, teach to the high performance students, teach to the low performance students and teach differentiated pedagogy. And those actions have different costs attached to them. It is more easy, less costly, for instance, to teach to the high performance students than to teach differentiated pedagogy. It takes more effort and takes more skills to do differentiated pedagogy. This is data from PISA that tried to provide some of the flavor of what I think is happening. These are data. Each of these dots are schools and we separate between public schools and private schools. We have two years, 2006 and 2012 and we are plotting here index of wealth versus an index of diversity of dispersion. The higher this access, the more heterogeneous the school is. And this, again, each point is a school. So we see in this graph three facts that are happening. First of all, this is the case of Colombia and in Colombia there is segregation that is important. Private schools are attending high income individuals and the public system is attending middle and low income families. The second fact that this graph is showing us is there is a down slope regression in which basically the dispersion decrease with wealth. These classrooms, these schools are more dispersed, more heterogeneous than the public ones. And the third one we believe, and we are trying to test this, is that heterogeneity is increasing in the system. And this is not only the case of Colombia, there is similar graphs of other countries. I just put Colombia, I replicate Uruguay and Turkey but the same pattern is happening. The same pattern in which we believe heterogeneity is increasing at the classroom level. So what is the conclusion of the second idea is the following. I think in developing countries we need to investigate the level of heterogeneity in classrooms for both public and private schools. Secondly and more important, I think that we need to investigate, differentiate endowment of skills at a school entry. That is one of the points that is very important again is that once the kids arrive to the school, the kids, the endowments of the kids work as an input of the school. And if the endowment is very dissimilar, the teachers will have a very difficult task of trying to, for instance, increase the education of low income individuals. And what does that mean for terms of research going more inside the classroom? It means that I believe that we need to start looking inside the classroom with instruments that can capture what are the pedagogy that teachers are doing in classrooms. For instance, there is this class instrument that Pianta at the University of Virginia is using. And we can go inside the classroom, start observing what are the actions that teachers are doing when they are facing more heterogeneous population. And finally, I believe that there are some pedagogy methods that allows a better approximation of the problem of differentiated pedagogy. There is a model called Escuela Nueva in Colombia. The model is the following. This is a model that is multi-grading in the same classroom. And the idea was that the density of schools in rural areas were very low. They tried to bring more kids to the school of different grades in the same classroom. And basically the teacher organized the kids by desk and the kids work in each problem at their own rhythm. That type of model allows a more flexible and more differentiated approach to pedagogy. The model that is happening right now in the typical school in developing country is not like that. The typical problem in the typical school in developing country is like this room. You have kids sitting in lines and there is one person talking to them. And that will not be conducive to differentiate pedagogy. I am doing an evaluation of a model in Vietnam with the World Bank on the model. And the results are very promising. Third idea about incentives. So the debate here is the following. There is one perspective that says that system are failing in the linear quality of education because of the misalignment between the incentive structure and the final objective of education. And there is the example that teachers, the promotion of teachers is automatic and is not linked to the result of the kids. And in that case, you, the teacher will just receive remuneration and promotion regardless of what happened with the kids. Another perspective is that the system do not really require the education because of the lack of proper pedagogy and content on teachers. And if that perspective is true, what is key is teacher training that provide content on pedagogy. And something that is very striking is in a lot of our system, the training of teachers is not about content and pedagogy. The training of the teacher is about philosophy of education. And these people know a lot about, let's say Piaget and a lot of other people who know a lot about education. But when you put those people in front of a classroom, they don't have the tools to teach the kids. And moreover, they don't have the content to teach the kids. So there is some evidence in Latin America very clear that this is happening. So fine, there is a problem of incentives and a lot of people are trying to do these programs in which they attach incentives to the production of certain outcomes in the kids. And there are some emerging evidence coming from a pilot program from by researchers that test the idea of attaching a payment to teachers. And my reading of this literature is that actual incentives can change behavior and may induce better student outcomes, at least in the short run. For instance, the evidence coming from Kenya, from India, from China, but clearly incentives may change behavior in an unintended way. For instance, the case of Kenya, they detect that there were a lot of teaching to the test. And in the case of Mexico, they discover dishonest behavior. One, almost like a food note, teaching to the test not necessarily is a bad idea. When you have high power test, teaching to the test is a good idea. What do I mean by high power test? When you have test that actually measure analytical thinking that measure actually is how people solve problems, teaching to the test is not necessarily a bad idea. But anyway, so there is, basically there are these programs that try to attach the incentives to certain outcome of the kids. And I think that there are some promising results there. However, the response of those programs depends on the nature of the problem and in the capacity to respond. If the problem is teacher after season, and we know that there is a lot of teachers and students in developing countries, then there is a large margin to respond in absent season. We will mainly use the teacher to come more to the school, but we don't know if the kids are going to learn more. If the problem is content and pedagogy, teacher incentives will have less capacity to do what we want to do, which is more learning of the kids. So that concept was proven in different context under pilots and some researchers. And we did something in Pakistan, which was the following. We convinced the government of Pakistan to run such a program. And we convinced the government to try to do the best program possible given the context. And something that is very important about the pilots of the researchers that happened in Kenya and India and China was the following. Usually the researcher do the following. They test the kid at the beginning of the year. They follow the kid after some period and they test the kid again. So you have tests that are applied by the researcher and the test follows the same kid at some time and another time. And that is done because of the following reason. You want to have incentives attached to change in test scores. You will measure what happened at baseline with the kid and then you measure sometime later because when you have incentives attached to changes, you have, you inputting incentives into the teacher to concentrate effort in all the kids. And moreover, they have more incentive to concentrate effort in the kids that have lower test scores at baseline. That data is not the data that governments have. Governments have data that are very different. Governments in developing countries usually have data that are great specific. And in Pakistan, for instance, they have these grade five exam in Punjab. And when you have data that is great specific, what happens is that you measure every year a different cohort of individuals. And when you try to attach teacher incentives to that data, what happens is that the teacher do not have the incentive to produce any effort because you have shocks in each cohort that are exogenous that the teachers cannot control. So one year you have a good cohort and the teacher says, okay, I have a good group of students, I will do less effort. If you have a bad cohort because of a shock of the cohort shock, the teacher says, okay, I have a bad cohort, I will not do the incentives. Moreover, there is a dynamic problem in these type of programs when you have cohort of individuals, which is once the school is able to increase the scores one year, the next year is more difficult to increase it again. So you have these problems that are attached to the nature of the data. And the vast majority of data in developing countries are not suitable for teacher incentives. So a point that I want to make here is, yes, teacher incentives may be very important, but in developing countries, there are some limitations that are critical. The limitation of the teachers cannot respond because they don't have the right tools. And secondly, the nature of the data that developing countries have is different. So one of the most important ideas of our teacher incentives is that they try to change behavior of current teachers. They, the margin of actions for the flow of teachers is very limited of these type of programs. And the most important investment in the future of quality of education in any system is the flow of teachers. It's how can we change the people who are entering into the profession? How can we change the type of training they are receiving? How can we retain the best ones and let go the ones that we don't want? And how do we do professional development? And again, one point of policy that is very relevant, very important is the second one, the training that the future teachers are receiving and there is little investment in research there and the policies usually are not about that. It is very difficult that policy will concentrate that. What we need is we need teachers that receive right content and that receive better pedagogy. So if we don't have that much room for action, what should we do? And one solution is that we do demand interventions. And demand interventions are important for changes in this education system for several reasons. But one of the clear reasons is that there are some policies that have been proven that actually are changing the way families behave. I have a hypothesis of why it's happening is because probably we know and we understand better how family act and we know family better as an institution and schools and institutions are very difficult to understand and they vary a lot among contexts. But demand interventions have been proven to be very successful in different contexts. For instance, we have policies that bring kids to school which are for instance, conditional cash transfer and information policies and those policies have been very successful. One area of research for me that is interesting is that once you have a conditional cash transfer policy, you are tackling another problem which is a fundamental problem in education system which is the following. Kids that are coming from certain backgrounds do not have the mental space to get the knowledge that they should receive. For instance, if a kid that come from a low income family arrived to the school and the only question that she has in her mind is I am hungry and what I am going to eat today, that kid cannot learn. So conditional cash transfer are interesting to me because conditional cash transfer can be another mechanism to create mental space for these individuals. But in any case, policies that bring students to school like conditional cash transfer has been very successful. There are some demand interventions that also change student achievement. For instance, scholarships. And here I am mentioning just two cases, the case of Kenya, the case of Cambodia and my paper in Cambodia shows something that I think is relevant and it's also a point of interest in my research which is the following. If you target programs using poverty, you will leave certain behavior different from a scholarship that target using merit. So when you target using merit targeting mechanisms, you will leave certain responses that are more schooling and higher achievement in test scores in contrast to poverty-based mechanisms, targetings in which you only induce higher progression in the schools but you don't induce changing achievement in tests. So in this point, what I want to say is that targeting is very important not because it's a way to reach certain populations but because targeting by itself can produce different behaviors in individuals. It's very different if you say to the kid, we are going to support you because you are poor or we are going to support you because we believe that you can do it. Those two approaches are totally different and in targeting, we are creating certain behaviors that are differentiated. And the last point about the money interventions, we know that early child interventions that goes to the family or that provide a high quality child care centers can make a huge change in the system. So all in all, the money interventions can be very affecting bringing kids inducing higher achievement and changing social emotional skills. And what we are trying to do right now with a group of people is to say, okay, we are going to intervene the demand and we are trying to see if that change in the demand can trigger changes in the education system. And how do you do that? We are doing an experiment in Manizales in Colombia which basically is trying to first to intervene the demand side, providing information. And secondly, we are trying to go to the school and we are saying to the school, look, it is fundamental that you have a good interaction with the families. And you have to start having a relationship with the families because the family is your best partner in the education of the kid. So we are trying to make changes in the demand side that try to trigger changes in the education system. So let me conclude here to have some time for questions. A long time ago, I was in a conference in which Benarji says something that was for me quite illuminated. He said, we need to go inside the black box of education. And believe it or not, I was thinking, okay, do I need to go into the black box of education? I am an economist and economy is thinking the following way. Economists think you have a leverage, you can change the incentive structure and you let people do what they know best. And I think that that is correct. I am an economist. However, we need to go inside the black box in a fundamental way. We need to start understanding how teachers react to incentives. How schools react to incentives? How schools organize an institution? And that means go inside of the classroom, go inside of the school. And we need to understand how teachers are or not being able to do heterogeneity differentiated pedagogy. And that is a critical point. Why? Because in developing countries, the challenge is the following. Yes, we are increasing the enrollment rate and a quality is flat. If we want to start moving the quality, we need to start thinking about the following problem. We increasingly have classrooms with more heterogeneity. And we need to have pedagogy that can tackle that problem. And that, we need to provide tools to tackle the problem of pedagogy. Or we change the type of models of education that is happening in the school. If we keep with the model in which there is a teacher in front and kids sitting in rows, we are not going to be able to tackle the problem of heterogeneity. The last point is my reading of the literature what had happened in the last 20 years, which by the way, I have a huge admiration for the work of the flow of energy, Kramer, in the sense that I think that they are seriously making a revolution in the way we think about development. My reading of the last 20 years of this agenda is that we have been very successful doing demand interventions, and we haven't been very successful doing supply interventions. And part of the problem is we understand better what happened in the family. We don't understand that much what happened in the school or the school as an institution. And we need to start thinking about the school as an institution. But in any case, demand side interventions can be very powerful. You can start changing the system if you start changing the demand and if you start changing the families. And what I am saying is that through demand changes, you can start triggering changes in the system that has profound consequence for the development of education. Thank you very much.