 I share my screen now. Yes, please. Yeah, okay. Okay, thank you very much. And Jaya's gonna talk about whether do public policies promote equality of opportunity for well-being. Over to you, thank you. Okay, just a minute, I should go back. Okay, thank you very much. So this is going to be a big change from the previous one in terms of content. It's a paper which is a very, it's very much empirically oriented and coming from a quantitative perspective of looking at how can we kind of evaluate or assess equality. So this is a joint paper with Ricardo Nogales, who was my ex PhD student, who's at the University of Oxford now. So this is about combination. And in fact, earlier I was talking about friendships. This is another friendship, a combination of different approaches to equality and well-being. So there are two big words here, equality. So we're going to ask what equality are we going to talk about? And well-being, what is the concept of well-being that we're going to use for defining equality or not? And then there will be an example on using Bolivian data. So the overview of the presentation is this. First, a little bit of background. Then I go to the theoretical framework. As I said earlier, this one is based on an econometric modeling. Then the econometric strategy, implementation, and then application and with some concluding remarks. So we start with the idea that the recent theories, more modern theories of development and well-being acknowledge the fact that the policies should be aiming at not necessarily equalizing outcomes, but equalizing opportunities for achieving outcomes. So equalization of what is the equalization of opportunities for well-being? And so with this idea, we combine two different approaches. One is saying what is well-being? The other is saying what is a policy that aims to equalize opportunities for well-being? So coming from the well-being perspective, a point of view, again, there's a wide consensus that well-being is a multi-recently in the recent years, that well-being is a multi-dimensional concept. So one should be going beyond economic dimension to cover all aspects of life, all aspects that people have reason to value and people can consider as important in the definition of what a good life is all about. So we say that well-being should be not just economic, not just income, but also social, political, everything. And we're gonna come back to this. So this is called the capability approach. And it was one of the pioneers of this approach is Amartya Sen, but also there are other philosophers who have contributed to this approach. So this says it's not only it's multi-dimensional, but it is all about opportunities. It's about the ability of people. This has its own jargon of terms, but it's, I mean, I'm going to minimize the jargon as much as possible. So it's about well-being. It's not just well-being, but potential well-being. What the ability of people, what could people do and be? Like in terms of be educated, be healthy, have be employed and things like that. So what the ability, the opportunities for people to do and be things. And these are what are called capabilities in this approach. And they are kind of a counterfactual in nature because obviously you don't observe all that people could do. We only observe what they have actually achieved in their life. So other things they could have done, for example, I mean, even if you take the example of this lecture, you could have not been, you could have been somewhere else during this lecture. So you had other possibilities and you chose the possibilities of being at the workshop. So you don't observe all the possibilities that you didn't choose. And so this is by kind of its nature, it's a counterfactual in nature, this concept of capabilities, what people could achieve. And then the idea is to say that this is what we want to equalize. Equalize people's opportunities to do things and people's capability sets. Capability sets in all dimensions. So not just in the economic dimension, but also in the social dimension, the political dimension, in the environmental dimension, to be in a clean environment, to have good social relations, to be emotionally stable or healthy and physically healthy. So all dimensions to equalize capability sets of individuals in all dimensions. That's coming from this approach that would be the equality concept. And then what is a good policy that will aim at least enable equality in this way? And that we take what another very important approach in the inequality literature, which is the equality of opportunity approach, which was developed by John Romer, again one of the pioneers is John Romer, who said that any outcome, any wellbeing outcome, and this approach is more used in the unidimensional literature more, but we are actually going to apply this combined. That's why we are combining capability approach with this. So we're gonna combine this, take this approach and apply it in the multidimensional setting. This approach, it says that normally any outcome is a result of many things, but it should not be like effort, the effort that people put in, but it should not be if there is equality of opportunity that people's circumstances should not matter for what they achieve as their wellbeing. So to be equal to the circumstances are some things, are defined as things that are beyond an individual control, like gender, for example. So it should not matter what gender one person is, provided both men and women put the same effort or have the same, you know, same capabilities, they should be able to achieve the same things, irrespective of what their circumstances are. So effort equal effort should lead to equal rewards and circumstances should not determine the outcomes. And if circumstances do matter, then there is inequality of opportunity. So these are the two concepts, the major concepts that we are combining. And as I explained the first approach is capabilities, capabilities are defined as opportunities and choices that people have to lead the life so they have reason to value. This is the kind of the formal definition of it, but as I was explaining it just to say that, you know, it's what people can achieve, potential outcomes, let's say potential wellbeing. And the actual wellbeing achieved because of the particular choices made is called wellbeing. And in this jargon, it's called functionings, that is they say beings are doing, you're healthy, you're educated to a certain level and so on. And then capability set is the set of potential achievements in all dimensions. So it's a multidimensional concept. Now go to the equality of opportunity or inequality of opportunity. According to John Romer, this any outcome, wellbeing outcome here is a result of three things circumstances. As I said, these are factors that are beyond individual control like gender, race, color in a family background, you don't control the family that you are born into and things like that. So these are circumstances, efforts, of course, these are deliberate decisions taken by the individual, how much effort you wanna put in and things like that. And then of course, there's always a luck part, a random shocks. So when there, we say that there is equality of opportunity when, as I said, circumstances do not play any role in the determination of outcome. Either directly, so it should not matter whether one is of a particular race, particular color, particular gender and or indirectly even via the efforts because of the fact that sometimes if you were a person of a certain color knows that how much ever effort that she puts in, she might end up with a lower outcome. Then it might, in fact, discourage her from putting adequate effort. So that's an indirect effect of circumstances on the effort, which ultimately leads to a lower outcome. So either circumstances should not matter either directly or indirectly by discouraging efforts because knowing anticipating a lower outcome. So that means what, so if circumstances don't matter, so that means if two people, persons for equal effort, then they should be getting an equal reward independent of the circumstances. And that's the concept of leveling the playing field. So how to verify this in practice or how to go about operationalizing these things in practice. So in the equality of opportunity literature, there are two ways that it is done. So people are, a population is divided into different groups of individuals characterized by different circumstances. So a group of individual is characterized by the same set of circumstances. They're called types, this literature. So by the same age group, gender and other things. And then we examine if the equal effort in these different types gives rise to the same reward or not across these types. If they're not, if they don't give rise to the same reward then we say there is inequality of opportunity. That's one way of looking at it, dividing the population into homogenous circumstance sets. The other way is to just do a regression of wellbeing outcomes on circumstance variables as well as efforts and check if the coefficients of the circumstances are significant or not. If they are significant, then there is inequality of opportunity. These are the two major ways in which the quality of opportunity concept is applied in practice. And then normally in this literature, as I said, people look at a single outcome which is mostly income. So they look at an outcome such as income and see if that depends on circumstance variables as well as efforts. And sometimes even efforts is hard to observe. So that's merged into the random, random shock term. So now having, and most of the, most of the time we do observe that there is inequality of opportunity that the circumstances do matter. So then how will you look at a public policy? What should a public policy do to reduce inequality of opportunity? It should aim to reduce the effect of circumstance on outcome. And so I'm not going to say a public policy is optimal, is something that completely eliminates the impact of circumstance on outcome. And we'll see that in an example. So just a picture saying that if you say equal resources, if you give equal resources to people who are differently, who have different circumstances, hear the circumstances, simple saying that the height of the individual, then if you give them equal, if they are equally, if you distribute resources equally, then obviously people, all people will not be able to achieve the same outcome. So that's equality in resources or equality in resources. And so what we are saying is that's what we want. So in order to have social justice or fairness, you should be offsetting the impact of circumstances. Policy should be offsetting the impact of circumstances. Here being a shorter version should be given a bigger stool so that all of them can watch the match. And so in this paper, we are combining the two in the sense that we keep the concept of equality of opportunity, but we say opportunity for what? Opportunities for wellbeing in a multi-dimensional concept. And then we kind of say why there is compatibility between these two approaches and they can be combined and operationalized to check if whether public policies are reducing inequality. So now, so in this, in the theoretical framework, so we set up a model. And as I said, it's also going to be a simultaneous equation. The expected outcome is a function of circumstances, efforts. F is the efforts and public policies. Of course, circumstances and public policies can be merged, but we are looking at policies and see if they are optimal or not. So we are going to give a separate name to that, symbol to that. And as I said, efforts can themselves be dependent on circumstances. So there may be an endogenous endogeneity here. So we take that into account. And then this is expected, there's all possibilities. And the outcome is one possible outcome among different things that people could have had in terms of capabilities. So we say that this Y is a particular choice, a particular vector, it's multi-dimensional. So you have outcomes in the different dimensions of lifestyle dimensions, like material income, wealth education, health, other things, social relations. So this is a vector. And this is a particular vector out of all possible, other possible vectors that they could have had, but we may not observe it. So we're going to take it as latent variables. And these are determined by the resources that people have, their own resources, efforts, as well as what they have as a circumstances. So the model is this. So they achieved what we observed as achieved outcomes are a product of efforts as well as what they could, the choice said they had. And this is determined by circumstances among which we are going to isolate public policies to see whether, see circumstances, the equality of opportunity means that there should not be any arrow between circumstance. Well, there's social circumstances, family circumstances and individual, as I said, color, race and other things. So normally these arrows should be non-existent if there is equality of opportunities. They should not be influencing and we do observe, they do influence. So what we are trying to see is whether these public policies, how much they reduce the strength of this influence between the social circumstance and efforts or capabilities or opportunity sets. So therefore the Y is a particular choice of the Q. Q is the capability set, which is this. So Y that we observed here is a particular choice, is a particular outcome out of the choice set. And then of course that has been achieved through efforts. The choice set is dependent on each other, the capabilities influence each other. They are dependent on the circumstances, they're dependent on the efforts and they're dependent on public policies because public policies do influence the surrounding factors and then the efforts themselves, as I said, could be dependent on them. So this is the framework and then we're gonna put it into an econometric model. I may be skipping a few of these. So this is just to say that because of the counterfactual nature of capabilities as well as efforts, because we may not be able to observe all the efforts that people put in, we're gonna assume them to be latent, that is not observed, directly observed, but we may have certain indirect measurements of them, like efforts, a number of years of schooling and the number of your hours of study and things like that. If you're looking at an educational outcome, capabilities that what are the choice sets that they had while the choice sets will be dependent on the resources they had. So we may be having measurements of these things but not fully, they're not directly observable. They are mutually influencing each other. That comes from the simultaneity of the model and they are obviously influenced by circumstances. The X is the vector of circumstances and the T, the key parameter to be examined is the T, which is the impact of circumstances on these capabilities, efforts and other things. So the T, how much, if the T ideally should be zero, they should not matter, but of course, the X does matter and so we have some significant T values, but then do public policies help in reducing this impact at least and that's what we are looking at. So if public policies are achieved at least to reduce the impact of T, then that means they are at least working towards equality of opportunity. So the T is a function of the policy variables and that's what we want to find out whether how much the policy variable is the counters the effect of the circumstances. Then if you look at the effect of the circumstances, we can calculate the effect of the circumstances as a function of the public policies and see that if the public policy has to counter it, then these two have to be of opposite signs. And that's also if the, by the, we define the policy to be contributing to an attenuation of unfair inequalities of opportunities if the beta parameter and the delta parameter have opposite signs. So if beta is positive, delta is negative, that means that it's going to reduce its impact, right? And then an ideal policy is something which makes it zero. The excess should not impact on the capabilities or the effort so that an ideal up, we will define an optimal policy to be one with completely annihilates or offsets the effect of circumstances. Okay, now we go to the empirical example. We use the household survey, mind you, most of these surveys are secondary data. We don't have primary data in terms of we didn't, we don't do the surveys ourselves. And most of the time our applications or my applications have always been in the context developing countries to say that although the model looks a little bit sophisticated and developing countries may not have. And of course, in the case of capabilities, it's developed or developing countries, there are very few countries which have data on capabilities. So most of the countries don't have data on capabilities. So that's why we take them to be latent variables, but we want to show that in spite of the kind of complexity of the model, we can use the information that is already available and derive useful policy insights. Okay, the public policy variables are the public expenditure on infrastructure like sanitation, water, electricity, roads and communication and the public expenditure on social services like education and health. These are the policy variables that you were looking at. And the circumstances we are looking at are gender, ethnicity, ethnicity is basically indigenous, non-indigenous, age and family background. That is whether the household head is educated or not. To a certain extent, rural, although it's sometimes not considered to be a circumstance because people can choose to be in a rural urban. But the bane for circumstances are gender, ethnicity, age and family background. And normally they shouldn't matter, provided ethnicity, any non-indigenous and an indigenous person putting in the same effort should have the same outcome. That's the idea. So what are the outcome variables that we have? One is the material conditions, but since we say we don't want to adjust the material, we also have what is called a life satisfaction measure. So material conditions of living are basically the residence quality, ceilings, walls, floors, basic living conditions, drinking water, do they have access to drinking water, electricity, cooking, what kind of, do they have access to clean cooking fuel, residency equipment, telephone, mobile internet, crowding, the number of bedrooms per members. And then there's the subjective well-being, which is are they satisfied with their wealth, with their community relations, with the environment that they live in and with their affection within the family. So this is the similar, I mean the Bolivian case, which is exactly similar to what we saw before. And so just some illustrations of examples. And as I said, we do always observe that the circumstance matters. So if you take the circumstance of belonging to the indigenous community, you see that this has a negative impact on the material conditions of living. So indigenous communities just, I mean for equal effort, because effort variable is in there. So for equal effort, indigenous communities tend to achieve a lower outcome than non-indigenous communities. And you can even see that they also influences the effort they put in. Indigenous communities tend to put in lower effort because of the fact, as I said, an anticipated lower outcome, irrespect, I mean, whatever be the effort they put in. So it also affects indirectly. So there is, the first observation is there is inequality of opportunity. Do these expenditures, public policies, do they offset that inequality of opportunity? Well, in the case of material conditions, the infrastructure expenditure doesn't do very much, but the social expenditure seems to help in the sense of the negative is offset by this positive interaction coefficient. So that means the net effect will be reduced by the net effect of being indigenous is reduced if there is some social expenditure by the state. So the reading of the stable is that being indigenous generates inequality of opportunity. In fact, most of the inequality of opportunity is direct. There is some which is indirect going through the effort and then the effort on conditions, 84% direct and 16%, only 16% is indirect. Social expenditure seems to contribute to reduce this unfair inequality because the fact that being indigenous shouldn't matter. Well, I'm gonna skip this because of the time. And we also show the individual heterogeneity is there. So there's a whole distribution of the impact. On that impact, if you reduce, we simulated scenarios to say if you reduce expenditures, obviously the impact is going to be much higher. So the reduction of the expenditure, social expenditure by the state is going to aggravate the inequality of opportunity. So it is clear that there is inequality of opportunity. And if you look at the another circumstance variable, which is a family background, again, something beyond an individual control. And that we have, for example, an illustration, parent schooling, parents' education level. Again, that does seem to matter. So normally a child born into an educated household seems to achieve irrespective, again, putting in the same effort seems to achieve a better outcome than a child born into an uneducated household with parents uneducated, which should not be the case. Now here, both the social expenditures, public policy variables seem to have a good impact or a significant impact. They seem to reduce the effect of the schooling variable, the parent schooling variable. So they have the opposite signs. So they seem to eliminate, now there's not eliminate. They seem to attenuate the inequality of opportunity due to this effect of the family background, but they don't necessarily completely eliminate it because they don't completely offset this coefficient. Here, the contribution of indirect, direct seems to be a very different configuration. Only 45% of the effect is direct. 50% comes through indirect. So people, children born into families with a low level of education of the parents seems to put in lower efforts and that also obviously contributes to a lower outcome. So both the policy variables seem to reduce in the case of Bolivia here. Again, some graphics to show the same thing. Again, some simulations to show that if you did reduce the infrastructure or social expenditure, you'll end up in a much worse situation. To say that they do matter, the whole distribution itself is tilted. So not just average because the regressions give us an average outcome, average impact, but even if you look at the distribution of impact for all individuals, that seems to be the case. The whole distribution shifts like. Now, do they actually completely eliminate? As we said, no. How much of expenditure is needed to eliminate? The current expenditure is $5 per year per person. You need $31, for example, to completely offset the disadvantage due to being indigenous. So it's almost five, six times the expenditure that is needed to completely offset the unfairness due to the circumstance variables. So what we conclude, the capability approach as well as the equality of opportunity approach can be combined to be an operationalized to look at whether policies are contributing to reducing the equality of opportunity. And we have also explicitly modeled the role of efforts because most of the time the literature puts efforts with the random shock variable because effort is very rarely observed. And in this, using it on micro data from Bolivia, we show that being indigenous is one of the main sources of inequality and social expenditure contributes to the reduction of inequality, although much more expenditure is needed to completely offset it. Similarly, having being born into a privileged family background is another important source of unfair inequality. And that is also the policies seem to go in the right direction in terms of countering the effect of these circumstances. Thank you very much. I think with that I stop. Okay. Thank you very much for the presentations. Yeah, I already see some questions. Thank you very much. What's very interesting. I was thinking about the last result that you showed with regards to family background. Yeah. And the impact, the positive impact it has and if it's included with infrastructure and social components actually has a slightly negative impact. And that reminds me of these social scientists, conservative scientists that I have recently started reading his work at Thomas Sowell, who argues that for example, the welfare state is not helping people. I mean, it's important to have education in family and a stable family structure, but actually giving equal minimum wage to everyone is in fact has a backfire effect and does not help especially marginalized community to thrive in the community. I was wondering, what do you think about this? I don't know whether, we haven't done empirical work on that, but one point I would like to kind of highlight here is that this infrastructure expenditure and the social, the two policy variables are infrastructure expenditures and social expenditures, but they are not targeted towards a particular, these families with low level of education of parents or indigenous communities. These expenditures are targeted towards the whole population. So it's not, it's not a targeted policy variable. So even if the social expenditure is beneficial, is available for all the population, it seems to be helping the unfair, the people with an unfair circumstance situation more than the people with the other, than the other. So here, what we are showing is that even if you have a universal policies because in the policy variables, people in the policy literature or policy field, people talk about conditional or targeted policies as versus universal programs, universal education for all in health, universal health, free health access for all. So the universal, these are all the effect of universal programs that we are looking at, not targeted towards a particular communities. Even if you have a universal program, that does go to the right people or right side of people are able to benefit from it in terms of that seems to, that means these programs are availed off or seem to be beneficial to the ones that need them the most. So that's what here it's shown, empirically shown, and that's what seems to work. Now, whether the universal program of just giving out free resources, whether how much that will compare that with a situation like I have here in terms of not giving them free things, but also improving their surrounding factors like better schools, better hospitals, and better sanitation, better water access, and things like that. So these seem to help, whether just distributing income to everybody, whether that will help or not compared to this, I wouldn't be able to comment on that because we haven't done that comparison. So it might well be true that they may not help as much, it's clear that, and then this is a really in the spirit of helping people to help themselves. Providing education, providing good healthcare, and et cetera are giving opportunities for wellbeing so that if you give these opportunities, then people are willing to put the effort, I mean, people put the effort because the effort variable is there. So the effort is put by everybody, but the equal effort will give equal reward, provided you give them the opportunities for flourishing. Hi there, thank you so much for your presentation. My name is Ruth. I'm doing a PhD in Urban and Inqualities at TU Delft. I'm interested in your infrastructure and social variables. From my understanding, they are... Oh, yeah, sure. Yeah, your expenditures. Sure. So from my understanding, they're aggregate variables, and I'm just wondering if you have any insights into if there's specific kind of infrastructure... If you spend on specific infrastructure, it has more of an impact, or specific social variables, it has more of an impact that be interested in the decomposition of those effects. Yeah, you're very right there, aggregate variables, and I forgot to mention that all the outcome variables are at the individual level or household level. The circumstances variables are also at the household or the individual level, but the public policy variables are at the municipality level, because for the public expenditure and social services, we don't know how much of it, as I said, goes to every individual. This is how much they spend per municipality, so that there is still variation across... A huge variation across municipalities, but they are at the municipality level. Now, we only have this disaggregation. Definitely, I mean, my conjecture is, if you had a finer disaggregation of these expenditures on particular and specific things, you would be able to even... You saw that some of the variables were not significant for certain things, and I think there will be much more significant if you had a finer classification of what the actual expenditure goes into. So we had a general category, education and health, and that's all we had in the survey, so we couldn't get a finer classification in the survey as well as from the municipalities. So we couldn't get a finer classification, but my, as I said, hypothesis is that they will definitely improve the significance of the coefficients that we have, and we'll be able to pinpoint to what particular expenditure is responsible for offsetting the effect of particular circumstances. But then the reason, as I said, the minute you want to go into these finer classifications, then you have very few countries which have these, especially in the developing world, and then it becomes, you know, we take the same old US data or other things, and then it becomes the same old story. So the reason for one of the reasons that we took a very crude data set because this data set was not meant for such an analysis, very crude data set with very high level variables is to say that even with such, even if you're able to get some good outcomes, I mean, good results even with such crude data sets, then improving a data set of our data collection should be able to give much finer insights, yeah, definitely. Thank you very much. Can you please come here? Yeah. Okay. Yeah, so hi, thank you. I'm Marta, I'm one of these, I guess, physicists running around the workshop. Thank you very much, especially thank you for the effort of explaining to people who are not used to these terms in a sort of more accessible vocabulary. So I was wondering, so if I understood correctly, effort is not something that you can measure, you don't have data for, you model it, is that correct? Yes, well, I don't know whether I skipped that slide, there is the effort slide, I think I did skip it maybe. So we don't have one single effort variable. So as I said, we postulate that effort and capabilities are by definition not directly measurable. So for the capability variables, as you saw, we had two different dimensions, and then we have indicators of what a good material conditions capability. So capability in the material conditions was, there are many, many indicators here, unscapabilities in the subjective wellbeing dimension was this similarly effort, we said, we don't directly measure, but we have some partial indicators of that, what these are the four indicators that we chose. So the age at first job, whether the person is working in a formal setup or an informal setup, whether what kind of a position the person has in the occupation category and the years of schooling. Of course, again, there may be arguments in terms of whether all these are personal decisions or they could be also been influenced. We do allow for there being influenced by circumstances. Well, we do believe that there is a certain decision making in these variables, whether how many years of schooling you wanna do and whether you wanna go to of course age at first job in a developing country context is maybe it's one of that. And then one of the referees in fact said, this came out in the world development. One of the referees said that maybe age at first job is may not be completely under the individual control because the situation may be demanding that they go take up an employment for family sustenance and all that. But then there are still some personal decision making. So these are some variables where them personal decision making is involved. And so these are our partial and we don't claim that these are the only effort variables. So we account for the fact in the econometric model that these are some partial indicators of effort and then there are some unknown effort variables. Okay, thank you. No, no, no, they look reasonable to me. But the reason why I was asking this is that in your model you have lots of parameters, right? It's a linear model, which is I guess the simpler you can have, but you still have lots of parameters, right? And I wonder, so what advantage do you see at using these kind of models with respect to doing some simpler data analysis in which you don't have to assume that there is a linear model behind it? And just trying to find equal, so this equal opportunity, I mean, if it's all equality of opportunity, sorry, I don't remember exactly the name, just by looking at say the outcome and trying to control from some variables just by randomizing data or any of these other approaches. I mean, I'm thinking that probably the outcome would be the same, but do you see like another, I mean, I know that you're an expert in this so this is why I'm asking. So what's according to you the added value of assuming having to assume that there is a linear model when in reality you do not know what kind of model is behind this data? Yeah, okay, good question. So the first thing is that, okay, we do account for some non-linearities because of the fact that the model is linear, looks linear, but the variables may not be necessarily in the linear form. So this may be, you can put an age square, you can put things like that, right? So the variables may be non-linear, but the impact coefficients are linear, yeah. And here the measurement equations may be non-linear. So there is a little, there is non-linearity within the linear regression setup, within the linear structure. So the other way is what I was talking about earlier where we said the two ways, where we compare outcomes among different groups of individuals with different, each group being identified as a set of homogeneous circumstances, facing homogeneous circumstances. But the problem with these models, I mean problem with this way, which is a completely non-parametric way of doing things, is that the more circumstances you have, on the more classifications you have, like gender, there's only two, but ethnicity, there may be more age, you may be different age groups. So if you combine all that, so then you will have a type will be defined by persons of people belonging to a certain age group, of a certain gender, of a certain ethnicity, of a certain level of education of the parent, and all those, so if you combine all that into one type, then you end up having many, many types of individuals, many, many groups of individuals, then you face what we call, you know, curse of dimensionality, and so then it will becomes hard to compare the outcomes across these different groups and see where they stand at equal levels of effort. So this way of doing things just as descriptive or as you said comparison or data analysis helps when you have fewer groups, fewer circumstance variables, but the minute you want to add more circumstance variables, the regression approaches is useful in detecting or identifying equality of opportunity or inequality of opportunity. Okay, and can I ask one short question? Yeah, and do you typically check how robust it is if you remove data? So from what I understand, you just put all your data, run a regression, and look at the significance of the coefficients, but there is the question of how generalizable is your regression model. So typically people just in other domains say computer science or physics, you remove data, you fit your model and see how well, what's the error that it produces on that data, or how the coefficients change if you want, I mean it depends on what you want to look at. So do you, did you look at this and what are the differences in terms of like, so how much would I have to expend more if I wanted to remove inequality and opportunities or things like that? Yeah, so there are differences, of course we did three or four variants of our model, it's all in the paper, and there are differences, qualitatively the results are similar. So the direction of influence and the fact that these policy variables reduce the impact of circumstances, all the qualitative conclusions hold. There's no big difference. Quantitative of course the numbers are different, but yeah, so we do say that this is for this particular variant, the results I showed is one of the four variants that I have, so we do have other variants where there are qualitative, I mean the quantitative numbers are different, and the graphs are slightly different, but I mean yeah, it is true that there, we don't have a point, we don't claim to have point estimates, it's just to give a general idea of the range of values that you get in terms of your results, right? So we do have, that's why instead of looking at just the coefficients, we plot the distribution, the whole distribution of values that you can get. So you see the range of values that the impact variables, you know the distribution of the impact variables rather than just one one. So we have the whole distribution of impact for the different variables, that they are, I mean they're not very far, but they are different, yes. Okay, thank you. Okay, thank you for this interesting presentation. I think you empirically debunked now this laissez-faire, very individualistic thinking of equal opportunities, so very strong message I think. I'm glad you now showed the efforts, and I'm a bit skeptical if you don't mix up your independent and dependent variables. So the efforts to me look like more outcomes of what else you have shown of these opportunities, capabilities, and I'm sure you've done the statistical tests and checked for variants and so on, but I'd be curious how confident you are that this is really showing effort that is put in, and I know this is a real crucial area of research, and maybe late in the discussion we can also think about how to better and more spot on we can measure these effort variables. Yeah, you're very right. Effort variables is one of the very, as I said, you know, maybe I don't know. I don't know of many studies who explicitly model effort variables. They are all part of the error term, you know in each equation as you can see there are these error terms, so typically the F is outside the model. I mean outside in the sense it's not explicitly put in and there, and so in another paper what we did was we, this was applied to Indian data sets, so what we did was we said we split the efforts into the effort which are, which I mean partial measurement of efforts, but before, yeah, partial measurements, and then unknown, so observed and unobserved components of efforts, and we said the observed components of efforts which is also here are themselves endogenous, are taken as endogenous variables. Now whether you call them as outcomes or efforts, it's endogenous, so if you wanna say that this is what they are endogenous, that particular point is well taken and it is part of our model here as well as the model for the Indian data. Now whether these are, as I said, correct effort variables, we don't say these are the only or correct effort variables, we say that these are possible things that we can think of as effort variables because years of schooling, for example, is something that is typically taken as an effort variable. That is people in all of equality of opportunity literature, people, the only effort variable that keeps coming up is the years of schooling. That is people have, take deliberate decisions of the number of years of schooling beyond the compulsory, of course. The number of years of schooling they wanna do, complete at the compulsory level or secondary level or beyond secondary tertiary, so that is considered as an effort variable. So we said age at first job is another effort variable because what they, you can decide, again, it's almost similar to the years of schooling, you can decide whether you wanna continue your education or you wanna go to take up a job and whether you wanna go to the formal sector in all developing countries, the informal sector is very, very big. Like in Bolivia, it must be more than three quarters or something. So people can do, and we, another paper, we showed that it is a deliberate decision. It is not by lack of choice that people go to the informal sector. They do deliberately decide to go to the informal sector because of the fact that, of course, it has certain disadvantages, it may not have the social security and other things, but it may give some flexibility, they may have two jobs and other things. So it does give other advantages that people value. So the fact whether they're going to a formal employment or an informal employment is a willingly taken decision. I mean, it seems so. So these are the kind of arguments that we put forward to say that these are, efforts is basically, the circumstances efforts are factors that are within the individual control, circumstances are beyond the individual control. What your races, it's beyond your control. What your ages is beyond your control. Whereas how many years of schooling you want to do, whether you want to go to the formal sector or informal sector are within the control, at least partially. But they are influenced by your circumstances, they are influenced by your capabilities, they are endogenous. That is clear. I don't know whether I answered your question. Yep, yep, thanks. Thank you, thank you. So Jair, I wanted to ask a question. So when you're talking about endogenous variables, so you can introduce an instrumental variable and redo the panel regression. Have you done those type of things as well and what type of instrumental variable? And what type of instrumental variables? Sorry, yeah, yeah. So we do have in the circumstances of the instrumental variables in a simultaneous equation model, the advantage is that the instrumental variables are given by the excluded excess from each equation, right? So we do have exclusion restrictions which provide the instruments. Yeah. I don't understand. So certain circumstances do not impact certain variables and so, I mean certain circumstances, I mean in these, the four circumstances are all in all the equations. There are certain other X variables which are not present in all the equations. So the exclusions give us the instruments for the estimation. Right, the other thing, what I understood when you were doing and it's panel regression that you're doing, right? Did you also account for the? Just one, sorry, one cross-section, there's no panel. Oh, no, no, only cross-section, okay, fine. Yeah, because there's one cross-section. Right, thanks, thanks. Other questions? Okay, I have one actually. So I'm also a member of the physics sub-community in this workshop. So I was curious to know whether you have done similar studies or others have done similar studies to in different countries, whether that would make sense and whether it would make sense to compare results across countries. So how you can compare and generalize essentially? I have done myself, as I said, another empirical study in India for India, Indian data. Again, of course, people, other people have done. In the multidimensional setting, I think the studies are rare, but in a unidimensional, in the inequality opportunity literature, there are tons of studies. And most of them look at earnings, wages, some variants, income, earnings, wages, and things like that. And they all show that there is inequality of opportunity, most of the time, all the circumstances, most of the circumstances variables always turn out to be significant. So that is one thing that is comparable in terms of across studies. But I think it's very difficult to compare the different studies across different countries, simply because of the fact that the variables are not the same and the definitions are not the same. So I think what can be said is that it's that across all studies, you do find inequality of opportunities. And most of the time, you can find policy variables being significant in reducing the inequality of opportunity. And conditional, as I said, whether it is universal policies, or whether it is conditional cash transfers, like in the Mexican, we did an impact, we did a study of, yeah, I also did it for Mexico, a study of the Opportunidadas programs, which targets individuals and households, and then there are transfers, cash transfers made to individuals who are considered to be in the poor. So there we did also a study on whether these cash transfers help to reduce the unfair disadvantages that people have. And of course the answer is yes. And none of these studies, so the qualitatively you can compare, there is inequality of opportunities in almost all countries, and this is developed or developing. There is no kind of big difference there. And the policy variables do contribute to reducing this, but they don't obviously completely eliminate it because the resources that needed will be really enormous for that, yeah. Thank you very much. Sorry, can I have a clarification? Suppose that your set of circumstances is enlarged or restricted. You get some result on the effort, on the impact on the effort, because I suppose that your effort, not observable, is a sort of residual variable. Is it like total factor productivity? No, the effort is, sorry. Yeah. Yeah, so the effort is not observable directly, but as I said, we have observable, partial observable indicators of effort. From that, these are, you see, when you assume a certain thing to be a latent variable to a B and then say that you have certain indicators of that, we put in a factor analysis model here. So each of these is a mini factor analysis, the Y1 star to the Y1, and then here there's an effort to the observable indicators, there's another mini factor analysis here, a third factor analysis here. So these are factor analysis variables through which we can get some factor scores on the results that I presented, the effect of effort on material conditions. This is the factor score effect, impact of the factor score, which is not observable, but we can still find estimated coefficient, right? Through the factor analysis. Yeah, but this coefficient is inflated if you forget some circumstances or not. Oh, if you forget some circumstances that they do get on the efforts, yes, they will be, because then if you, for example, if you remove the indigenous one, because of the fact that the indigenous, the impact of indigenous is not there, that's going to be kind of, yeah, it's taken up by either the effort variable or by the other X variables. Yeah, so this is, this makes, if you omit, yeah, yeah, please, sorry. If you omit a variable, the other coefficients will be biased. Yeah, this is making it very difficult to compare the result of, for example, for Bolivia against other countries, because you are not sure that the circumstances are the same, okay? Exactly, exactly, yeah, that is correct. Other questions, comments? Yes, I'm just a little bit not convinced about the measurement of effort, because, well, you mentioned about the years of schooling as the common, like, measurement, I'm surprised, like, it is not heavily criticized, because I think for, I can think, as a sociologist, I can think of millions of cases, individual cases where they actually, the year of schooling actually is reversed, correlated with effort, like, compare someone who was forced to drop out of high school with, like, a get pregnant, her family can't spot her, she's forced to drop off high school and then do, like, three shifts a day, and that's a lot of effort compared with someone who has a wealthy family put them into college, like, they don't need to worry anything, and they actually just party every day throughout college. And I think if you measure these kind of things by years of schooling, these two people actually not, if you are measuring effort, it's actually not really the case, I feel. Okay, so I agree with you, and this is exactly in our model, as you can see, for example, here, the indigenous take less efforts, and as you say, the circumstances, like, for example, being in a different family situation, will impact on the effort. That is perfectly taken, and that is also modeled into the, that is in the model. So people with different situations, different family situations, different family wealth, different resources are going to put in different efforts, and that is in the model, and that's what we are actually, as I said, one of the, and that can be done only if you have some effort variables, because if you put the effort variable as a random error term, then you don't have the possibility of taking this influence, except to say that there is some endogeneity somewhere, but here because of the fact there's explicit effort variables, we can add the influence of the family situation on the effort variable, exactly the negative effect, and you can see the negative coefficient here. So the negative effect that you're talking about can be explicitly modeled if you have explicit effort variables, yeah. Okay, thank you. Thank you. Okay, thank you very much. I also want to go on about this effort variable. So one thing I was wondering about is an additional thing is whether the impact of effort may actually also differ depending on family background, for example. So if I think of years of schooling, well, if I think from a cultural capital perspective, for example, whether your parents have brought you up in a way that you better know how to study, that you better know how to use the terminology that teachers like and so on and so on, may have an impact on how effective actually the years of schooling are. So for the one person, the same effort may be different than for another person. So I was wondering how you think about that. Yeah, that may be true because in fact, yeah, you can interact effort with family background, for example, circumstances variables. So here the effort goes like this. It says that people of different, well, we saw also in the case of education, people of different educational background. So people with, it is normally the case that people, children of educated parents tend to be pushed into making better or more efforts than people of maybe a lower level of education. And so that is seen in this effort coefficient. But then whether the, what we are saying, even the two parents pushing their children to do more effort because of their different backgrounds, but given that the children put the effort that they put in may even also give rise to different outcomes depending on the families that they are in, that is not in the model, in this model. So what we have in the model is that people, children of different family backgrounds may be putting in different efforts because of the influence of the family situation or as you said, of the parents. But then given that the effort themselves will give rise to different outcomes, the effort put in by these two different children, will they give it to different outcomes? Could be, and then it will be a question of adding the interaction on the effort variable also. Interaction of effort to circumstances. Yeah, so I would be curious to know what happens once you do that. Yeah, okay. I don't know. We didn't do that. That would be something which we can explore. I mean, which could be explored. Yeah, we didn't do that. But we said that given the fact that we do take into account that the family situation does impact on the effort and the fact that that impact can be quantified by saying, for example, if you say indigenous people put in less effort because of their family situations and that multiplied by this will be the kind of the interaction of the effort, the indirect effort of the indigenous on the outcome through the effort variables. So if you multiply this and this, that would give you the point minus 0.112 times 0.398 will give you how much the difference between the effort in the non-indigenous will be this and the indigenous will be this times this, right? So that will be, that's how indirectly you can calculate but you can also put it as I said if you wanna try a different variant we can put it directly on the effort variable, yeah. Okay, thank you. Anyone else? No more comments or questions? So I ask the, I see Matteo has one, okay. Actually, I have a question for psychologists here. So essentially going to the other talk, the first talk, I think that was interesting because it was tracing back sources of determinants of inequality to cultural aspects or so how, I mean, as far as I understood how cultural traits can be combined in different ways. So I wanted to ask psychologists or say people who can comment on that talk can give some perspective on this because I think this is an interesting dimension. Of course I'm looking at Mirtha, but I don't know. Tim, you want to, Eric? No, the question is how much this perspective is a prevalent perspective on inequality is a, there is this tension between say hierarchies and egalitarianism. And this is rooted in cultural aspects. I think that is an important way of looking at it, but we can also think about it in wider terms. So why do we have those beliefs about hierarchy and egalitarian ethos? I would like to think about it more in terms of what is best for a collective at a specific point in time and how these different, not only ideas about hierarchy and egalitarianism, it can be other dimensions that leads the collective to be more or less adaptive. In certain circumstances, it might be good to have a hierarchy in other senses, better to be egalitarian. And of course, that and other aspects of being a collective and work together can be better or worse over time. It leads us to the done different paths. We can be stuck in one of these kind of semi-stable states and then maybe the environment change, but then we can switch over them and switch over to another state, another state, either because we are more adaptive or we perhaps have miscalculated or learned something in the past that didn't really work out, so we switch on to another thing back and forth. So that's why I came. So my understanding was that it was thinking about these two different say archetypes, like hierarchies and egalitarianism, as a hardwired in human nature. Like, I mean, maybe not in hardwired, let's say in biology, because say you find hierarchies in animal societies and then, okay, you think, okay, humans have developed this abstraction for egalitarianism or things like that. I mean, how much this resonates with the way people think about the processes and the human, I don't know. I'm usually kind of hesitant to talk about hardwired stuff because that's bringing you down a road that, I don't know. So I think even in a purely kind of experiential or learning situation, we actually learn to see that, well, some things just have to be in hierarchy. I mean, we have democratic societies, but we have hierarchies everywhere, because sometimes it's just better to salt us that way, because we cannot just have everybody decide everything whilst it can be slow, inefficient, and so on and so forth. So even if you go beyond this hardwired thing, any kind of learning system that starts from scratch will certainly learn that all. If we're gonna do this task quickly and efficiently, maybe we need a hierarchy for this one, but this for the other task, we might not need this hierarchy, we need more of a egalitarian way of doing it, the more consensus-based or whatever decision-making process for solving that task. So I think that this talk about hardwired stuff, I'm not so sure about that, even though it can be true. I mean, say one way of thinking about hardwired, say this hard-wiredness is like the way we are built from a neuroscience perspective. So for example, we have neuromodulators that modulate a lot of social interactions and there are five of them, okay? So there are five modes in which we interact and this shape our behavior, but so this is the way I was thinking about Thomas and so. So maybe from my own experience, I'm not an anthropologist or psychologist, but I'm working with those people and I also always thought, okay, are people egalitarian? Are people hierarchically? No, it's discourses. So we're looking at discourses and discourses can be egalitarian, hierarchically and we are forgetting about the individualistic discourse which is actually quite prevalent in global North countries and then we can also see this flexibility. In one discourse, I might be more egalitarian but then you switch more to what's a hierarchical discourse and that's how I understood then to see problems, challenges. Look at those from a discourse but not connect individual persons to this hardwired, you are hierarchical, you're egalitarian. It's more really a fluid thing also. Of course we might have tendencies towards one or another but yeah, that's how I learned it. So I think I hijacked the discussion, sorry. I help you to further hijack it. Yeah, well, maybe one thought from a sociological perspective here would be that let's say in every society we have authority for example, because it's just needed to get things organized. We cannot always discuss everything and so on but a very important question is to what extent is it seen as legitimate? To what extent do people accept it? And I think in at least some theories say that humans, let's say, have a mix of fundamental needs. One of the need is to get approved or for doing the right thing. So to say, get confirmation that you are living up to the norms of your society and you can get that from authorities, that's very nice. But you also have need, for example, just simple basic material needs, needs for recognition by your peers and friends for affection and so on. And I think one important question in society is always to what extent the authorities and the authority and the hierarchy structure we have is consistent and not too much in conflict with the other needs we have? Of course it's not nice to get told what you have to do but if you see that on the whole it helps to have a good life in this society you are willing to accept it. That would be what many people would argue but if people get the feeling that and maybe at the moment we see that that more and more people get the feeling these elites are actually not really solving the problems that we have and on the contrary, things get worse for all of us than at some point that legitimacy of authority declines. And I think then we are at the point that we may see an uprise of the more egalitarian discourse. Yes, yeah, yeah. I follow the directions. I follow the direction. I send two slides. Okay, if there is time then I can... You know this egalitarian versus hierarchy, particularly about the transition from one to the other. It might be that, I mean if one accepts the basic premise first of all that there are these two basic modes, hierarchy versus egalitarian in people's minds and there is a collective sort of transition from one to the other, which is what he was presumably talking about quasi-steady states. And that transition might be because of an overshoot. If you spend a time, spend a certain amount of time in close to one quasi-steady state. Then there is this idea that over a period of time, certain things get entrenched and certain unproductive things or unnecessary things or things that really weigh the system down get entrenched. And that overshoot basically then has a reaction and causes eventually over a period of time a transition away from that sort of... I mean this is one of the ideas. I think broadly that Tainter has about collapse of societies. There's an overshoot of some kind, maybe of complexity or something like that that causes this overshoot. So it could be that spending long enough in a particular equilibrium or not equilibrium but quasi-steady state as he calls it releases, I mean sort of create certain vested interest and create certain inefficiencies. I mean I don't know if there is an economic description of that but it sounds to me that the transition is possibly is happening because of some kind of overshoot that is inherent in being close to one particular quasi-steady states. But at a more basic level, it seems to me to be somewhat arbitrary to make this kind of a classification as a basis for discussing what's going on, right? I mean I don't know whether this classification of egalitarian versus hierarchy is that basic in our minds. Maybe we are concerned, I mean this is in part in response to your psychological question. Is that in our minds, are we really concerned about hierarchy and egalitarianism? Or are we concerned about some other things? I'm not completely convinced that this classification ought to be a basis of our thinking of how things happen. I'm also skeptical about it. It just seems overly general categorization in two things and this is, I mean it really depends on the function of hierarchy. So what do we mean? What do we mean by hierarchy? The example was the speaker had an example of chimps or whatever and the function of hierarchy is there to, is made control, to have control over females. That's very different than many other forms of hierarchy that we have. So one, for example, a hierarchy that we don't even notice is the hierarchy in the way we learn from each other. Students naturally, to an extent, I wish they would more, listen to their professors and learn from people who they think have more expertise. In decision making, we will naturally defer to you to tell us where to go for coffee. And there are many, that's also a hierarchy in a way, right? What is a hierarchy anyway? But then there maybe what we are more against is hierarchy based on some arbitrary, culturally determined norms such as who is more virtuous, who is more moral, who has more right to some kind of benefits based on relatively arbitrary markers. I guess that's what we are revolting against. But not all hierarchies are, as Henrik said, I mean we need some kind of, any information processing account of any kind of collective will get to some kind of hierarchy. Our chat GPT is a big hierarchy, right? Of layers in a neural network. So is that bad? Hey, can I just add? Yeah, I saw at the beginning of the hierarchy and the egalitarian debate, I do found some resonance about Chinese history, like the dynasty beginning as kind of a egalitarian force and at least promised to be and then gradually like people become corrupt and stuff and then there's a hierarchy and the dynasty end and like it has been going on like that for 2000 years. But I do feel like the problem is this kind of simplified relationship is sometimes actually, well, I need to choose my words carefully here. So take our like current president as an example, he actually legitimized his hierarchical centralized power by promoting egalitarianism his claiming to equalize the redistribution from the rich people to like normal people. But he claims to do that, he has to have enough power and each layer of government has to have enough power to force the capitalist to give out their resources. So it's actually kind of a paradoxy development between hierarchy and the egalitarianism. Hey there, so to just add a comment, I was a little uncomfortable with Jim Robinson's set up and I agree with what was said that this strong dichotomy between egalitarianism or hierarchy is just an ideal picture, it doesn't really apply. Every organization will have bits of both, right? And be more on one extreme or the other. So, but what I want to warn about, what I want to say is that in problems of inequality or hierarchy or social structure, the purely structural view which is quite dominant, particularly in some social sciences. So you described a structure just like Jim was doing is creates a situation that's undecidable. This was said by some of you, you don't, okay. You say, oh, it looks like this is this kind of hierarchy. Okay, and then you have to say, is it good or is it bad? Should it be different? Then you have to say, what Mirta was just saying, for what? Right, what is it doing? Is it doing something good or not? Good for the collective, good for the individuals for whom? So you have to basically unpack a sense of process of what is the structure doing and how is it doing it across scales from the individual agents to the collective. And then you can decide if it's doing it well or not, if it's freezing development and creating power in the hands of the few, or on the other hand is organizing people like in the military or something to potentially be effective, which is a hierarchy. So I think without a sense of function and purpose of what the organization is doing, it's just completely undecidable. I'll give you some examples tomorrow in my talk. So I think I would be good for us to keep in mind this overly structural view of inequality, which we often do when we have a lot of cross-sectional data and always be asking, what is it doing? Could it do it better? Could a different structure service it better? Is it arresting dynamics and so on? So to me that helps a lot try to decide what kind of inequality and what kind of quantities we should be looking at. No one, just to say that, this is one of the main question I have. I mean, so society there should be an interplay between say structure and I mean, there should be some hierarchy. But on the other hand, it should not be too strong to have zero mobility. So there should be some, and so I'd like to understand more about this issue. So I think we have a coffee waiting for us upstairs. Maybe when we reconvene, we can start with Vito presentation. And also we decided to have slightly longer coffee break because we can interact more on coffee breaks. So maybe we reconvene at quarter to five. Is it okay? I would like to just thank everybody for their questions and feedback. Thank you very much. Thank you. Thank you.