Investigating The American Dream: The Role Of Neighborhoods

- Everyone for coming. We're happy to have Marta Prato who's visiting here all year at spr, so, so you'll have plenty of opportunities to talk to her if you haven't already. And she's gonna talk about investigating the American dream, the role of neighborhoods with Aand, Foley, and Veronica.

- Yes, thank you very much for having me here. Today. I'll talk about this work is a chapter we are preparing for the handbook in intergenerational mobility. So it's an overview of our research agenda that studies the macroeconomic implications of the neighborhood exposure effects. And please feel free to stop me at any time with questions. Our motivation and the starting point of our research agenda is the observation that over the last 40 years, there has been a large increase in income inequality in the United States. But at the same time we have also seen a simultaneous rise in residential segregation by income. And so what we mean by that is that we see increasingly more higher income families in sitting being concentrated and in certain neighborhoods and more and more separated by lower income from lower income families. And the third observation is that at the same time there has been a growing empirical literature that has documented the causal role of neighborhood exposure effects in shaping the outcome of children in different neighborhoods. And what that means is that children that grow in higher opportunity neighborhoods systematic, have systematically have better outcomes as adults in terms of their educational attainment, in terms of their adult income. So we've all of this in mind. The question that we ask is what is the role of neighborhoods spillovers in shaping inequality and intergenerational mobility? And my plan to answer this question today is to first document some empirical facts in the aggregate data that link inequality segregation in intergenerational mobility. And then I'll move on to our theory, the links residential segregation and inequality based on a key mechanism of neighborhood exposure effects. And as a part of that, I'll show you how our, we use our theory to also think about policy implications. So let me start by documenting some aggregate empirical facts. And the first one here is documenting the very first point I started from, so the increase and the movement over time of inequality and segregation. To do this, we're measuring income inequality in the US with the genie coefficient and as a measure of segregation, instead we're using the dissimilarity index. So this index might be less well known than the genie. So the way this works is that the dissimilarity is a measure that is considering two mutually exclusive groups and understanding how evenly they are distributed across a set of units. So in particular in our context for dissimilarity by index, that's, that means we're taking two groups for us. The high income families throughout this talk will be the families in the top 20 percentile of the income distribution. The other group is everybody else in the bottom 80. And then we're asking how are these higher income families distributed across neighborhoods in a city? And when taking

- Question,

- Yes please,

- Mitchell, how are you measuring income? So

- We have done this with different measures. The baseline is the household income for families with children will be the, the benchmark measure in our data

- What's be included in income.

- So this is

- Cash transfers include any, only cash transfers including

- Transfer. This is before taxes and transfers.

- Okay. So it's, it's some fraction of market income. So you have to impute income donor occupied housing and you have to get retirement income that will include,

- Yeah, so we've done one version that our baseline WA version is wage income. We also have a, a version of other sources of income for the households that we're considering that are households with a young child. It's mostly not retirement income, it's other sources. That is our baseline measure. And that's before, before taxes and transfer

- Using any other measures besides the ginny, 'cause the buddy showed us the main issue is from the 50 10, the 50 90, not the 50 10.

- Yeah, so - Consumption would be very different if you include all the value of government transfer would be very different.

- That's a great point. So self insurance, the rising inequality quality, this, we've done it with several different measures. Some of that reflected more than other, but we've added these with the 90 10, the 80 20, the 50 10, the 50 20, this, this general aggregate back, it is there, but it is really mostly about the top of the income distribution. And that your second point, I agree that if things are much different when you think about transfer and how consumption changes, in fact, I'll show you later pictures. We'll, I'll always put together the inequality and income and inequality and consumption and show how different they are.

- Maybe you Okay.

- No, I was, it was on exactly the Oh, okay. Go, go ahead John.

- Which is depended. Maybe your theory and what you wanna go to is more about market incomes. And so if you got a lot of poor people around and they, we all have the same income, but half of 'em are getting government checks, that's a a different phenomenon. So yeah.

- Yeah. I depends

- What you're gonna use it

- For. Yeah, I want, given what I think you wanna use it for, you wanna do the pre-tax one because you, you want sort of productive potential. So, so if if it's just transfers that get you to a higher income level, that's not a sign of success.

- Yeah, that's exactly right. So, so here we're, what we're trying to capture is the productive potential that these, these people have. And I'll show you some of our, our policies in our model, how they, they change the experts

- Market

- Marketing comes

- Endogenous, but the key thing is top income inequality by all these measures has gone up at the top. Absolutely. And so that's, I think your mechanism is about top income inequality.

- Yes,

- We can debate everything else, but it's really top income inequality I think is gonna be the driving.

- There's not that many top, top people around. There's only one aton.

- You're gonna see the model and it's gonna be top income inequality is gonna be what's gonna be driving. So

- Investment income, just difference of wage versus investment

- Gonna be the, I think top income of the productive

- Yeah, labor income. And let me say this, the, the model is really going to be focused on the role of neighborhoods in shaping the potential income that children will earn as adults. So, so the focus is on the neighborhoods developing these, these potential for children and not very much in explaining all the sources of the income process later on. So, so think about that as potential, potential determinant that's, that's in fact the, the words I'll use. So,

- But if you think of Ian terms, it would be the productive pot distribution of potential wages basically.

- Correct. Okay, great. So returning to this picture, yeah, as I said, so that dissimilarity is capturing how evenly distributed are the higher income families across neighborhoods in a city wherein the data, our measure of neighborhoods is a census tract. Okay. And what this picture is pointing now on the X axis is the time starting in 1980 and I'm planning on the y axis inequality in blue and segregation by income in red. And you see these co movements. So the joint rise over time as I mentioned as my first point, but let me show you. Yes.

- Is the 20% just a number you picked or there's some reason, 'cause I think this is much broader than 20%

- You the 20% to define the higher income families. So that is we need two mutually exclusive groups. So we, we picked that as a benchmark. But you see the dissimilarity raises also if you define, if you move the high income threshold, it is you, you, it is robust to that. So if you go to the top 90, you still see it. It is quite pervasive.

- What about 50,

- Top 50 splitting the median? I don't think we have the picture for that, but my guess based on all the cut we've done of the data is that you would find something similar to

- Some of this is just mechanical. Suppose John makes a billion dollars than Jonathan's neighborhood has. There's more income inequality and Jonathan's and more, more segregated people live near Jonathan live in a rich neighborhood and people who don't live in a poor neighborhood.

- That's a great point. So thank you because that's exactly what I'll show you today is that exogenous differences in neighborhoods already generate some sorting I think along the lines of what you have in mind. And then I'll show you as a second step if you think that the differences in neighborhood exposure are actually endogenous and they depend on the composition of people, that is an amplification factor of the underlying sorting force. So that's exactly an overview of what I'll talk about. So, so before getting there, let me just show you a couple of other pictures. So this is a movement of these two measures over time. Let me show that this holds across phase two. So now what I'm plotting here is on the x axis, the measure of segregation in 1980 and the Y axis inequality in 1980. And every.in this map is a metropolitan area in the US. And so you see here that cities that had higher segregation in 1980 also had higher inequality at one given point in time. And as one more piece of evidence, let me also show you how this correlation across space also holds over time too. So that will be this next picture that looks very similar to the one before, but rather than being one point in time, this is the change from 1980 to 2010. So the X axis is changing segregation from 1980 to 2010. The Y axis is the change in inequality. And again, every.here is a metropolitan area. And these positive correlation is indicating that those cities in the US that had the largest increase in segregation by income also saw a largest increase in inequality. And with this movement in mind, the last aggregates picture, aggregate picture and aggregate data that I want to show is how this connects with intergenerational mobility. And we're doing this year using the GEOCODED NLSY data. So the NLSY is a survey of households that appears as a panel. So thanks to these we can track family household income starting in the eighties and now we can tr follow the children of these families over time and see what is the income of these children once they become adult. With that we're building these intergenerational mobility mattresses, meaning in each one of the panels in this picture, what you see is what I'm ploting on the x axis is the initial quale, meaning the qua of the income distribution for the parents in these households. And then we're tracking over time what is the probability that the child of each of these families ends up being in any portal of the income distribution of once these children become adults. And I'm splitting this in two panels on the left hand side, I'm putting together all the MSAs in the US that have the lowest segregation to start with in 1980, whereas on the right are those MSAs with high income segregation. And the key point to take away here is I'm highlighting in red a particular point of this intergenerational mobility metrics, which is the Q1 to Q1 transition, meaning the probability that if a parent is in the bottom quarter of the income distribution, their child once once becomes an adult will also belong in the bottom quarter. So it's a measure of the persistence of poverty. And the key point to see there comparing those two red squares is that on the right hand side for the cities that have higher segregation, the Q1 to Q1 probability is higher, meaning there is higher persistence of poverty at one of the, of the income distribution in those cities that have higher income segregation to start with,

- What's the incidence of single parent families by on either side of this dichotomy.

- So single single parents or families with a single parent are more likely to be in the bottom of the distribution. How extreme is that?

- I mean at the 2080 divide are how, what, what's the percentage of single parent families that are in the 20%?

- They're in the 20%? That's a good question. I don't have the exact number to tell you, but it is into

- That top 20% very easily without a dual parent family.

- Yeah, that's a great point and I I I think that is true. That is not something that our, our

- If if the dual parent family is the driving force for social mobility, which is a lot of data supports that.

- Yeah.

- Then is it this you're attributing income which might be attributable to household formation.

- Yeah, so that's a great point. I agree with you and I think there is a lot of literature documenting the role of the dual parent family. So that is now that is not the focus of what our model would speak to. So we're going to think about the role of the neighborhood exposure. So everyone surrounding these kids outside of the family center, those factors matter too. What we're trying to isolate is how much of this pattern that we see can be attributed to this neighborhood effect outside

- That family circle without taking into account the family structure. Yeah. Whatcha controlling

- For when we're the, the, when we take our model to the data, we're keeping the, our measure of family consistent. So we're always taking the family household with at least with at least one child in the same age. And then for this particular paper today, that won't matter. But we have done several applications in these models, for example, to policy to study policies such as the housing vouchers. And in that case we can exactly control and have all the comparison when we have one parent with one child, two parents with two children. So that is more about mapping our model to the data, but the core of the model itself will not be about the, the family structure. Great. But

- Just to follow up on your comment that there, the Chetty et al group actually find that mobility is also affected by the two parent versus one parent status of the others in the neighborhood. So

- It's Yes, that's a change from their original paper.

- Yeah, yeah. So, so it's not only your own household composition, but

- But your neighbors Yeah, yeah. The the composition of the neighborhood.

- Yes. - I also noticed the upper right hand corner is quite different. The chances of staying wealthy are better if you are in a high segregation neighborhood. I've always wondered about this, we, we say the advantages of take the poor kids and put 'em in the rich neighborhood. Yeah, they'll do better. Looks like they dragged the rich kids down.

- That is a great point. And that is actually what, that is one of the things that come out of our model when you're trying, when you look at aggregated facts of policies that try to move the lower income children to the better neighborhoods is precisely that while those children's gain, there are negative effects that are decreasing the outcomes of the higher income children.

- You may get to this, but this is a big focus of Charles Murray's book coming apart. Right?

- Okay.

- Right. So I mean I'm just worried that a lot of it might be in that book already.

- So lemme say, so the, the main contribu the main contribution that we have to this big topic is going to be a model that micro finds the role of these neighborhood effects. So that's exactly what's coming next. That is our, this is a big feature and that's our contribution

- Just for the rest of the audience. When you scale it up, it doesn't have big aggregate effects, even if it's local sometimes. Right. Letting them know that so they could,

- No, that's really important. Yeah. So you should say that and why it's so important to look at that.

- Yeah, yeah. Thank you. So then with this in mind, let me talk about our theory and as we already started discussing, this is going to be an overlapping generation general equity model with residential choice where the key ingredient that we're, we're thinking about is the local neighborhoods spillover that affects the outcomes of the children. And I'll show you what happens when we put these neighborhood effects in place and think about this and exogenous, but then importantly I'll show you that if we endogenize this spillover, these endogeneity amplifies the effects of the, that the neighborhoods have on the outcomes of the children. And then with the endogenous version of our model, I'll also show you how we map these to the empirical estimates from the applied micro literature on the causal effects of this neighborhood to put to quantify the role of the neighborhoods and think about their macro effects. And the overall takeaway here is that in this model, inequality and segregations are reinforcing each other and shaping intergenerational mobility.

- I'm not sure if you showed, oh I don't showed in the data and it doesn't look like you talk on the model, but one might imagine that there's a lot of segregation, you know, the, the, you know, rich, the richer folks moving into certain places for the, for the libraries, for the the schools. How is that entering into the model? Like is it only through the neighborhood's parents kind of spillovers or is it also possible that some of the channels are gonna be two days or something like that, that are set up?

- Yeah, so yeah, that's a great point. Thank you. And so today the, I'll show, I'll start from the simplest version of the model where the ne the role of the neighborhoods only operates through these children helping children develop their potential. But of course there are other reasons why neighborhoods are different, like you said, libraries, various, I think amenities that you have in mind. So in various applications of this model that we studied, one that was qu more quantitatively oriented, we have also introduced those amenities and quantitatively trying to separate to what extent segregation comes from amenities versus to what extent comes to this role for children. So that is more of a, of a quantitative question. Today I'll focus on the economic forces behind it so you won't see amenities, but I can talk about it at the end if there are questions. So I'll, I'm going to start with the static version of this model where the spillovers neighbors blowers are exogenous so that I can show you the equilibrium allocation but also compare it to the planners allocation and see where a policy would take us there. And then I'll move to the dynamic version of the model where these are endogenous so that I can show you how we take that model to the, the data and think about steady state and transitional dynamics in response to shocks. And at the end I can, I can talk about some of the extensions and applications of this model that we studied. Okay, great. So let's start from the static version of the model where the neighborhoods spillovers are exogenous. That means that we are considering an economy with a unit measure of parents where each parent has one child and the parent is defined by a pair of variables. That is the first one is w is the wage of the parent. The second one, what we call a is what our preferred terminology is thinking about this as the child's latent productivity. So as we were discussing before, this is something that is meant to capture what is the potential productivity that this child may reach on the market. This is the one that sort of that it's called a, because we can also think about it as sort of the ability of the child and there will be other variables that I'll show you later that determine eventually what is the wage that this child can earn on the market. And then we have a distribution, capital F is a joint distribution of parents over this wage and child productivity. And in this version of the model, this distribution is exogenous, we're taking it as given. And we're also starting from an initial case where this ability of productivity of the child is orthogonal to the wage of their parents, they are independent. And later on I will relax this and show you what happens if you said we think there is a correlation between the wage of the parent and the productivity of the then an important piece of this monolith

- Restrictive is that assumption that the productivity of the child is independent of the wage of the parent.

- It is, this is to start from the simplest possible case. It is somewhat restrictive indicate in the sense that of abilities of course very hard to measure in the data, but the evidence we have is that there is some positive correlation. So I'll, I'll dedicate some slide exactly to show you why that matters for, for the equ, for the planner.

- Getting back to the single versus two parent family and the child. Are you looking at any issues for example to, there's a lot of assor of meeting. So the assortative mating reinforces the productivity differences and the potential,

- Yeah.

- Expected outcome of the child given some regression to the mean, I suppose.

- Yeah. So in terms of assert meanings, there is some empirical work that is actually trying to see how much the author man mating has contributed to intergenerational mobility. So that is an important part in the data. Again, that's not what this model is going to speak to. So we're going to try to quantify forces from the neighborhoods that are not going through that kind of channel, but through the child development,

- Well everybody's just mixed up and it doesn't matter who you are, you just got a, you have a wage assigned to you in your model,

- You have a wage assigned to you based on your productivity or skills, you'll, you'll see that in one slide. Just before getting there, let me just describe the geography of this economy that has two neighborhoods, we call them A and B, and each agent is living in a house that has the same size and the same quality and it's been rented. So we call arcade the rent in neighborhood K. And to start from the simplest case, we are assuming that there is a fixed housing supply in neighbor day of measure. H meaning age fraction of the agents can live there. And that means that the rent in neighborhood A is an endogenous equilibrium object in this model. Whereas for simplicity we're assuming that in neighborhood B there is a fully elastic supply of housings. We're normalizing the cost of building housing to zero in neighborhood B, simply meaning that in this model the rental rate in neighborhood B is equal to zero. Now coming to these important points here, the car of the model is going to be the role of the neighborhoods and how they shape the wage process of these children. So let me show that here where we're looking now at the problem of the parents that have a certain wage W and child with productivity A and they're trying to maximize their utility while taking is given the equilibrium rents in every day. And these parents care about two things. So they're deriving utility growth from their own consumption. C but these parents also have warm growth preferences. So they care about the outcomes of their children. In particular here, their care, they, they care about the wage that their child will get once they become adults. And, and this ization project problem is subject to a budget constraint so that the, what parents spend on consumption and rent needs to be lesser equal than their wage. And then the second equation that you see in the problem is the key equation that describes the law of motion for the wage of these children. So W prime indicates the wage that the child will obtain when they become adult and that depends on the following objects. The first one B is just a constant. The two important variables are the first one is A, that is this fundamental late and productivity of the child. The second one is what we call sm. And this is what I've been referring to as the spillover of the neighborhood. For now, in this version of the model we are taking that spillover to be exogenous is exogenously given, but what it's meant to capture is various ways in which neighborhoods can affect the development of the child, such as through the quality of the public schools in the neighborhood or the, or the peer effects of the other children that the each child is friends with the network effects. Or you can think about broader channels such as the culture, the social norms, learning from the parents of the other children. All of these different channels are what we're thinking as being part of these neighborhoods spillover. Now the key point here is that for now we're thinking about these spillovers as being exogenous and we're assuming that neighborhood A is the one that has the highest spillover, as I will show you now in a second. In principle, all the parents would like their child to grow up in the neighborhood that has the highest spillover because that will contribute to their future income. But these, because of this, the demand for living in neighborhood A goes up and in equilibrium that drives out, that drives up the rent and prices out of neighborhood A the lowest income parents. So yes,

- I I get, I get you getting to a more complicated model and, and this is just the beginning, but it does raise this point of why neighborhood B don't, since houses are free and everything, don't they increase, don't they spend the money they would otherwise spend going to neighborhood A to make the schools better? I mean I'm, I'm not asking this as why about the real world, but what are the factors like for example, parents living at home and things like that, that are, that are causing them not to do that? Yeah, that may be the real underlying issues, not the fact that not the neighborhoods themselves.

- Yeah. So that is a great point and in fact some version of this model that we have worked with also introduces an investment in education. This, this one appear in what I showed today. But you can think about parents, can parents have some resources they can invest in their children in other way. One of it could be instead of paying the cost of living in a very expensive neighbor, they can pay the cost to do some investment in in their education of the child, putting them in private school, paying a tutor or something like that. So we have a richer version of this model where we're looking exactly as this trade off. So living in a more expensive neighborhood versus saving that and investing in your child in another way. So that is, you can extend the model to allow for this. The interesting feature is that even in a model of that kind, you still see this force of segregation by income and pricing out the lower income families from the, from the high spill over neighborhood and even more so when we put in education in a quantitative version of this model, we see that that's precisely what happens. So when families are sort of becoming, falling down the income distribution, but they have a highly productive child, they move to the lower income neighborhood to save on rent and to invest in the child education. So paying tuition cost, the result of that in aggregate statistics is you see this increase in segregation by income in the city.

- I thought we were going somewhere different, which is not that Palo Alto is inherently a great neighborhood, but just that even if they're completely the same people will self sort. Now i, I guess that needs, I guess that's coming next that the spillovers depend on who, who else is in the neighborhood.

- That's exactly the next sec, next section. Where, where is this over coming from? I will show you it is an endogenous outcome of the model where people sort

- The next one the neighborhoods will be by themselves. It's not clear which is gonna be Los Altos and which is Palo Alto, but which is, we all agree. Okay.

- Yeah, exactly. But in this, so to get some intuition in this first version, let me say one important assumption here is that there is no borrowing so here. So it is important that the parents here cannot borrow against their children future wages to finance their consumption on their housing.

- Cool. On on the previous slide, how confident are we that the coefficient in front of the log W prime should be one? And if that was let's say less than one, do you have, i I guess any thoughts on if that has any implications for policy or efficiency in some sense? Like let's say that parents are choosing neighborhoods in a way that's, you know, suboptimal for the children. What, what role for policy or efficiency there kind of across generations?

- Yeah, so thank you. That is a great point. And in fact how much parents care about their children here, it's, it's crucial. So here there is a one, or we didn't put a parameter because this is say the simplest possible model, but when we, in in applications of this framework, when we have taken this model to the data, we have parameters that we, we try to estimate that precisely say how much do parents care about their children based on the decisions that they take of where to live and how much they spend on education. And if they, if they did not, the equilibrium would be very different. In fact, as you can see in this simplest version of the model, the only reason for parents to choose different neighborhoods is because of they care about their children. And here choosing the neighborhoods is the only investment they, they can do in the development of their child. As I'm separately from that, as I mentioned in the beginning, one version of this model that we have used in one of our papers that is now forthcoming SUP, we introduced other reasons why you choose neighborhoods such as amenities And we have tried to rigorously quantify how much of residential choices are due to amenities. So I like parents like the restaurants in the neighborhood versus they like the schools and care about their children. Yeah and and just what we got out of that is that the role of neighborhoods essentially explains about one third of the increase in residential segregation and income that we've seen in the data. The others is due to other forces that are not in this model.

- You said you, if I understood right you you sort of have an estimate of the coefficient on front of the log W prime, what, what is that estimate like? Is that, maybe I misunderstood, but do you, do you think that's close to one or less than one?

- We have a quantitative version of this model which is in a another paper that is not the one I'm presenting right now. So it doesn't directly map to this coefficient that you have in mind, but what I said is that the quantification of this model that introduces other reasons to choose this, to choose neighborhoods such as amenities, et cetera. From that we find that about one third of the rise in inequality and segregation that you see in the data can be attributed to the role of neighborhoods spillovers in children development.

- I see, thanks.

- Okay, great. So yes,

- Can I just ask to clarify questions, remember one is how do you think about wealth here? That sort of third generational transmission of wealth and the role of wealth, you know, richer people that's much more skewed actually than income inequality. And secondly, for the first picture you showed like you're considering last 30, 40 years, what do you think about endogenous fertility and the fact that the fertility has been related to income?

- Yeah, yeah those are, those are great questions. So here we don't, we we don't have a lot of wealth. We don't have the consumption savings problem of the parents. The reason is while this static version of the model is very easy to solve when we go to the dynamic version, we keep the model simple on, on some parts in order to solve a dynamic problem with a fixed point that in some of our papers we have extended to a multi, not just to but many neighborhoods. That is why for example we haven't separately considered the role of wealth, but this model can be used for many applications and that will be my conclusions line. And for example, what you're saying about fertility could be one different application of these role of neighborhoods that we haven't done yet. Okay. So in this simple model, the equilibrium is characterized by a very simple residential policy rule and together with a renter rate that sold the parent's optimization, clear the housing market and satisfy the law of motion of these children wage. This is a picture that shows the residential rule. So what I'm point there is on the x axis is the latent productivity of a child and the y axis is the wage of a parent and this blue line is showing the threshold decision rule for parents to choose neighbor day. And so what you see is that in the parents that are to the right of the blue line are going to choose neighbor day and the ones to the bottom left are choosing neighbor B, meaning that it is the parents that have relatively high wage and children with high productivity that choose to go to neighbor day what is the high income ones. Because as, as I mentioned before, everybody would like to go to everyday but this dries up demand and that dries up the rent. So only the higher income parents can afford it. And then conditional on any parental level of parental wage on UY axis, if you cut to the right on the line on on this chart, you would see that for any given income level it is the parents that have a child with relatively higher productivity, they find it optimal to go to neighborhood day. And that's because of this multiplicative component that's says there is an assumption that it is a higher productivity children that gain the most to being exposed to the higher opportunity neighborhoods.

- No, no, I'm getting that the wealthy issue, you can't give your dumb kid inheritance, you have to try to send him to school.

- Yes. - Or a wealthy family could go live in a poor neighborhood, but the dumb kid just give him the money. But you can't do that in this model.

- They're not doing that any small

- And the is there a lot, is that evidence, is there evidence of that productivity? Because it doesn't seem obvious to me it might be that the, that the, that the, that the low productivity kids could benefit most.

- Yes. So that is a great point. That is an important assumption. And so productivity of a child is very difficult to measure. So it's hard to find direct evidence of these in the data. However, there are empirical studies of related policies to these such as the moving to opportunity housing vouchers where they give vouchers for people to low income people to move to high income neighborhoods. And they have randomized that and see which kind of families pick up the voucher and in fact decide to move and they have tracked the characteristics of these children and that shows some evidence that it is in fact this goes in the direction of these complementarity. So positive selection to, could we use IQ

- As a measure of productivity? Patent productivity,

- Yes. That is difficult to measure. There is the, the only measure available of that in the US data would be something like the test score that is available in the analysts. One survey is not exactly an IQ measure and to some extent it already captured the the family fact. So we don't have conclusive evidence of that. But the indirect of evidence that we have supports that assumption

- Is it doesn't, the A FQT other people know, I mean that that's really highly correlated with IQ scores isn't it? And that's given in the analysis y isn't it?

- That is I think the best measure that we have, although my reading of the empirical literature is that people say by the time children are taking this test, there is already a lot has happened to them that is affecting their performance essentially starting from what happened to 'em from birth. Yes. The

- Thing that could help you discipline more is stronger rear them. GSE here he has all the data at school district level above which task scores. And so these are sometimes big like in Florida, sometimes relatively small neighborhoods and so you could use that if you want them cognitive potential or

- Yeah, thank you, that's helpful.

- Task scores when they're like in utero, I mean her model is about a is in utero, anything else that happens after is affected by the sorting. And so I think that's the, the harder part with this.

- I think what what would be helpful there is the neighborhood could be lar so the, the high school sometime has some neighborhood, but Evanston is the perfect example, you know, very rich in the northern but they all go to the same high school. So in this model you're not able to distinguish whether these high school or the culture or the spillovers and so on, if you could actually try to point out whether the majority of the effect is the school versus something else, I think that would be very helpful. And so I think in some way using the kind of variation would allow you to define a neighborhood, you know, you know, different dimension of it.

- Yeah, thank you. I think that's a very helpful comment and in fact I think that is one of the promise promising ways for this literature to make more progress. Okay, great. So then let me show you some, I'll skip this. Let me show you some pictures to understand the mechanism in this simple model. So to do that, I'm doing here a comparative static exercise that is I'm comparing the equilibrium in different versions of my economy where I'm exogenously increasing this spillover gap between neighbor A and B. So that is what I'm plotting on. The x axis of each of these pictures is the spillover gap and then each one is plotting the EQU corresponding equilibrium outcomes. And what you see in the two panels at the top, the one in the left is showing is that as I increase this, the gap between neighbor A and B, I'm getting the the dash, the solid line on the left is showing that segregation by income is going up, whereas segregation by child productivity, that's the dash line is going down. And whereas the panel on the right is showing that this increase in spillover gap corresponds to an increase in inequality in the income of the children. So why is that happening in the model? Let me explain that using the pictures on the bottom. So as I increase the spillover gap, that means that it's becoming increasingly more valuable for parents to choose to go to neighborhood day that gives a larger advantage to their children that is driving up the demand for neighborhood day. That's what you see the bottom left. The excess demand chart is saying what if starting from the lowest gap on the left, if I were to keep the rental rate fixed at this level, but I increase the spillover gap, this creates an increase in excess demand. So demand in neighbor day that exceeds the housing supply. So what has to happen in equilibrium, you see it in the bottom right, the rental rate needs to increase in equilibrium in order to clear the housing market as I expand this spillover gap. But as the rental rate increases, that means that there are some families that would've picked neighborhood day that now don't have sufficient income to pay for their rent. That gate, that gate, they get priced out of neighborhood day and they have to go to neighborhood B. And that is why we see this increase in segregation by income as the spillover gap expands and some family are priced out of neighborhood A now which are the families that are getting priced out. As we saw in the previous picture in neighborhood A, there are families that have relatively lower income among neighborhood A that are there because they have a very highly productive child. These are the ones that are getting priced out and so their highly productive children are also moving to neighborhood B. That's why in the model we see that while segregation by income increases income increase the segregation by child productivity is declining. Meaning we have some highly productive children that are now being moved to neighborhood B.

- Okay. Ask a question about, you said at the beginning you're gonna use like the chetty hendron parameters to discipline.

- Yeah. - Some of this, which is cool, I look forward to it, but my understanding of their estimates is that they're using within family variation of like the parents moving their kids to like a different neighborhood and it's kind of exogenous your model. I mean doesn't it seem to be completely inconsistent, right, which is that the families would choose to have the younger sibling go to a better neighborhood besides because they anticipate better opportunities for that younger sibling.

- So that is a great point. I'm trying to think. So let me answer this now as as a preview of the quantification. So at the end of the day, where do we get the numbers of how much the neighborhoods matter is this charity and under literature? And it's exactly as you said, the key idea of their identification or what is the causal effect of the neighborhood is that they look at families in the US that move to different neighborhoods and different cities and they measure the outcome of the children of these families and they get identification by comparing children that move at different ages. And so they have some children that spend more years or fewer years in the different neighborhoods depending on on what it, what how old was a child when the family moved. And you're exactly right, the core of their identification strategy is that the timing of the moving of the parent is orthogonal to the productivity of the child. This is a actually what they write in their paper, they even have some further identification where they compare different siblings of the same family and say the timing of the move is orthogonal to which one of the two children is relatively more productive. So the the great thing about our model, which is exactly what we've done in the quantitative version of our paper, is that we can, we can run our estimation twice depending on whether we think that these assumption holds in the data or not. Because in our model, as you said, families move in response to the productivity of their children and we can test how much those, the quantitative numbers change if we map, if we think that this measure of the cherry under paper also captures the selection on children productivity or whether it is truly a measure of our technology. And this is an exercise that we've done. It is in the appendix of our other paper in so, so that is the great observation. Okay. Can

- I ask you another from Zoom?

- Yeah.

- But sorry to keep thinking of these in a a pf kind of taxes framework. But I find panel a super interesting and it looks to me, I read that as an important form of misallocation. This seems like an inefficient allocation of kids across neighborhoods. And so should I think of that also as a motivation for why we might wanna think of taxing parents' income because that could help, you know, correct this kind of misallocation across neighborhoods of kids.

- Yeah, absolutely. Thank you for this question That actually leads directly to my next point. So let me answer this as I show our next point here. So answer your points about misallocation that will become clear as we compare this equilibrium allocation with the allocation of a planner. So here we're thinking about a utilitarian planner that chooses the consumption policy for each agent in our model as well as the residential policy. So where should each family go? And the objective of the planner is to maximize a welfare function that puts equal weight on all agents in the economy. And in particular here, the utilitarian welfare function considers both the utility from the consumption of the parents. The first term there is the utility from consumption as well as the utility that comes from this warm growth preferences that parents parents caring about the income of their children. And the planner is subject to a resource constraint. So that total consumption needs to be equal to the total income in the city as well as to the housing supply constraints. So that the agents that are located every day need to be equal to the housing supply.

- Where was the market failure?

- The market failure here is that the parents cannot borrow against the future wage of their

- Children, the borrower. Okay.

- In this, in this version, yes. When the, when I make the spillover endogenous it will be also, yeah,

- It's a borrowing constraint in a, I guess a household you'd build building more houses the other end.

- So here is what the planner designed. So this is the residential choice. So again, the x axis is the ability of the child and the Y axis is the wage of the parent. The blue dash line is the residential choice that we have seen before coming out of the equilibrium. Whereas the red line is the residential choice of the planner. So as you see, whereas in equilibrium meetings, high wage parents and parents with high predictivity children that go to neighbor A for the planner, there is residential choice is a purely vertical line that only depends on the productivity of the children. Meaning that the planner is taking all the high productivity children in the economy and putting them all in neighbor day because those are the children that gain the most from being exposed to the high spill of our neighbor. So the planner wants to maximize the size of the pie by putting all the most productive children aid that generates the most income. And at the same time, this planner can also do redistribution. So then the planner is going to distribute retribute resources across families and give exactly the same amount of consumption to each household to equalize the marginal utility of consumption would

- Decouple your schools from your neighborhoods. So let's assume James Coleman actually proposed this. He, he, it's only a brief paragraph, but he proposes that all schools be able to select their children. So he assumes the schools will pick the brightest kids and then he assumes that everybody will wanna go to the best school. So you'll get all these, you'll get all heavily sorted school population. Yeah. But the school will then benefit from this concentration of very bright kids and others less so and so forth. But if you decouple that totally from the neighborhood so that you could easily be transported to school, then wouldn't you achieve your objective? The central planner could achieve this objective without having to move a lot of people from a neighborhood to

- Yeah, that is a great observation and I think it is that would be the case if these neighborhoods spillover is primarily coming from schools and the pure effects in the school. So that is I think pointing more to an empirical question that we're not trying to address here. So what is the effect of the schools?

- You also affect the peer group so you're affecting a lot of other things than, than just, you know,

- That's coming.

- What's happening within the

- Endogenous, the endogenous building,

- The endogenous is coming and I think that's open question is literature empirically is to understand to what extent this is pure effects that happen in school or the friends that I make after school. There is not definitive answer there yet. But if it's mostly the school, what your point is exactly correct.

- Okay. I wanna object though. This is not a market failure. 'cause what you've done is you've given the planner something that isn't available to people. So that that's, and you've also given the planner the ability to know how smart the kids are, which usually the planner doesn't have. So the remedy to that market failure is, is to, you know, solve the wealth constraint problem not to move kids from place to place. Yeah. So I, if, if the planner has the technology to allow savings that or a borrowing that people don't have, then the right answer in that model is, you know, give them money, not move their kids from place to place.

- So I agree with you in the sense that the role of these parents exercise is to illustrate efficient allocations and and what that means for segregation. That will be in the next picture not to derive from these what is the optimal thing that you want to do.

- So you could say the way you could implement this is not for the government to go in and say that smart kid that goes here, but rather the government to give people the money to make their own choices.

- Yes. And this is in this one application of this framework we have done for the forthcoming macro annual, we are comparing a lot of different feasible things that the government can do and how they compare today. I'll show you the one of giving people money and how far that achieves,

- That the endogenous spillovers is clearly the one most on everybody's mind. Because I know what ha I went to one of these schools.

- Yeah, it wasn't,

- It is just an aside, but I'm, as I listen to you, I worry bad school that we're focused on ourself, you know, we're all smart, we all hang around smart people, people and we all realize how helpful was, and I'm worried that it's, we're such a small group in the economy and that we're really ignoring the possibility that it could be the dumb kids that benefit most. That's what I was know. I, you know, you've built everything on the assumption that's the smart kids that benefit much

- Also marginal,

- Have kids, parents, smart kids will do fine on their own, but the dumb kids might benefit

- And the parents might be more likely to move to the better neighborhood because the younger sibling needs more inputs.

- That's another thing I didn't thought about that. I'm just saying, but you know, the dumb kid is probably a dumb parent as well. So I know they have no resources. I mean it's likely, but the fact is we're always assuming here that the productivity is highest, the marginal benefit pro is value. I think that's our experience. But I wonder,

- Yeah, so I think this is a, this is a, a possible concern. In fact, let me show you, I think some of these will come true in the co in couple of pictures I'll show you what happens in when environment where the wage of the parent is correlated with the ability of the child.

- We, we gotta get to because we, we tried this once, it's called school busing and it was a disaster and all everybody moved out.

- Yeah. Okay, so lemme just then show

- Careful to distinguish it was not value added from value, which you wanna call margin. It was, there's a big debate right now about pure effects and, and all the debates in, in the political battles in the schools. A lot of it's about our, if higher income people move out, okay, is that worsening the education of people left behind? Or is it better for them to be mixed with higher productivity kids or is that gonna make it harder for them? And I think there's no definitive answer that I've seen from that. That's a big debate.

- Yeah. So let me

- Informed those kids because they can't compete. No,

- No, no. I mean this, so we don't really know the answer. So when the, I mean yeah, I agree. I think we think this is, you built it in, but I think we really have to think carefully because a smart, you know, I have a friend who's view is, I don't wanna send my kid to the, to the best university. They're just gonna be depressed there, but I want to send them to a university where they can excel and that gives them, you know, so there's I think a more complicated anyway.

- Yeah, I think that is a great point to, to keep in mind. Let me show a couple of pictures to get to the endogenous spillovers. So

- Educational system is based on the theory that the returns to investments in education are greater for those who have higher levels of ability. That's we would have to rethink our higher educational system. No, no. This is the pur effect theory. I'm talking about the pur effect. Yeah.

- Is that, I think that's what you,

- So let me just show you here, what does the planner optimal allocation mean in terms of segregation and inequality? So here again, this is the exercise where the spillover gap increases on the x axis and on the top left we see, so welfare in bed is the planner that is achieving higher welfare than the equilibrium. But in more interestingly, you can see at the top right what is happening in terms of segregation by income. As we have seen in equilibrium, the segregation by income is increases with the spillover gap for the planner. The segregation by income is exactly equal to zero. Why is that? Is because the planner only cares about putting the most highly productive children in the high over neighborhood. And so far I was working in the case where there is no correlation between the ability of the child and income of the parent. So the result of this allocation is that segregation for the planner is exactly equal to zero. But what happens if instead I put in some correlation? Let me show that in this picture. So this is again, comparing the equilibrium and the planner, but now on the X axis, I'm changing the correlation between the wage of the parent and the ability of the child. Before we were in the case equal to zero and the planner had zero dissimilarities. So no segregation by income in the city. But if you, if there is some positive correlation, then you would see that the planner starts to also generate some segregation by income in the city. So, and again, the planner only cares about putting all the most highly productive children in every day. And it's just a byproduct of that is that if these correlation increases, those children are attached to higher income parents. And the result of that is that the planner would want a city with some more segregation by income. And as you can see here, the, well the planner always has lower segregation than the equilibrium. But as the correlation increases this, the gap is closing and it disappears If the correlation is exactly one that answer these questions about misallocation that there was before. So this picture really highlights the misallocation in this model is the fact that there are some highly productive children that if they're born in lower income families, those families cannot afford to pay the high rent to put them in the high opportunity neighborhood. And this is a sort of, is a, a source of misallocation in the sense that the opportunities for those children are, are missing out.

- What you're saying is true and it's all driven by rent. Then we should see no inequality in rent control places. But if you look at actual descriptive statistics, there's in the US there's more inequality in places that have rent control. We have to be able to get the apartment. Yeah, well yeah. Rent control.

- No, no. But it allows with rent control. I mean one of the big things is it allows the landlords to discriminate. So you're not gonna rent to the single mother who might not be able to pay the rent. You're gonna rent to the two parent household.

- Not clear that rents are actually lower in rent controls, which you're saying, right?

- Well yeah, they just ration by other means and it's often by these socioeconomic

- Yeah, I, I think I agree. I think the interesting point would be there to understand how do people sort into the neighborhoods when the price is not the mechanism that allocates them.

- Well, but you didn't tell us what the other one was. It seemed more interesting.

- So that picture is showing what is happening to children inequality. So what is happening, essentially the, the interesting takeaway there is as I increase the correlation between parental wage and child, child, the, from a point of view of the planner. The inequality in children income remains exactly the same and is again, the same message that the planner is always putting the same children in the same neighborhood. So it, the correlation doesn't affect the inequality of the Children's Act. What can simple transfer policy do in this setting? So here we're asking if a simple transfer policy can improve intergenerational mobility. So to do that, let us, we're thinking about a, a, a lump sum transfer, say of 20% of the average wage. Just, that's just an example number. We're giving it to all the parents in the bottom portal of the income distribution and we are financing that with a proportional income tax on all the other parents. What does the, what does this kind of transfer policy do? Here's the welfare and now the green line is the policy. You will see in all the picture that the policy ends up somewhere in between the equilibrium in blue and the planner in red.

- But I thought at the beginning you said it had to be earned income. It couldn't be transfer income. The

- In our data we are,

- Is your transfer income equivalent to earned income In

- The data? We're we were measuring pre tax and pre transfer income. This is just a, an exercise that we do inside of the model to tax some of the agents to give a transfer to show what is, what would be the economic force at play here. And the key, the, so the key point to illustrate with this exercise is what happens in terms of segregation? That's the top panel. So what is this policy achieving is to allow some, the low income family receive a transfer. Some of them will choose to use the transfer to go to neighborhood A, which they wouldn't be able to afford otherwise. Which ones are choosing to spend the transfer on this rent rather than consuming it for themselves are the, these are the ones that have the highest productivity children. And that is what you see there on the top left. That, so the pol, this is the dissimilarity by income and the policy reduces segregation relative to the equilibrium by allowing some of these lower income families to go to everyday. Although of course it doesn't completely close the gap relative to the planner's allocation where there is no segregation at all.

- Yes. So don't have any labor supply effects here.

- There is no labor supply.

- Okay. So what fraction of that 20% of the average wages consumed by the parents,

- What fraction that is? That depends,

- Say the average wage and they can spend it on getting housing in a or they can spend it on their own

- Consultant. They also, they also enjoy the housing in a well, yeah. Why? Because of the utility, the parent spending? Yes.

- Yeah. That, that, so different parents choose differently. I guess one way to see that to is if in the bottom, at the bottom here I'm plotting income inequality and consumption inequality. So quite a, quite a bit of that goes to consumption. And you see that because consumption, inequality here, the blue line is how large it is in the equilibrium. The green line is what it is inequality. And the red is the planner that takes the consumption equality completely to zero. So you can see that the policy takes you in this example more or less halfway there.

- The basic idea here is a lot of, a lot of that 20% is gonna be consumed.

- It is

- Rather than used to,

- Which is why the

- Better neighborhood gets those skills or the spillover, the schools, whatever's going on there.

- Which is why in these models, even in the quantitative versions of this that we have done for the micro annual, the transfer policy typically is not the most effective one because a lot of it ends up going to consumption.

- And if you care about inequality, you shoot the rich people. I'm serious. I mean why is inequality a in the social welfare function here?

- That is the, so, so part of the reason for putting up that planner problem is also to speak to inequality because, so income inequality now this is the one on the left. This is the inequality in children's income. And you can actually see that the planner is, the red line is the one that actually, that puts in the most inequality even more than the equilibrium that is in blue. And even this transfer policy, the transfer policy in green is increasing inequality in children's income relative to the equilibrium because it is sorting some of these high productivity children to the neighborhood with high spill over. And that will always generate more inequality. Although while also reducing consumption inequality,

- I think misleading. Yeah. Because in the planner's problem there is huge talent inequality in the next period. 'cause all the good people are together, they're getting this great spillover and if you look at their consumption next period, it's, you'll have a great increase in inequality.

- Y that's exactly what is happening here. And that's why they

- No, because you've got inequality staying the same. That's because you're looking today, you're not thinking about the next, the next day. 'cause the way you do it, is it?

- Yeah, that's this bottom left picture is, this is inequality tomorrow essentially inequality the bottom left panel C

- That's exactly right. It's one of the big stories, one of the big stories of the emergency in

- So, so this bottom left is telling you that the planner, the inequality tomorrow, so the inequality of the children's income for the planner, that's the red line is the highest higher than the equilibrium and higher than with the policy.

- But there was consumption, that's when I was looking at the other one. Inequality. That is today's consumption.

- That is today's consumption.

- But if we could look at consumption inequality tomorrow Yeah. And it'll look like the income

- Inequality. So this is, this is a static model, so I I I cannot do that. But if, if the planner wakes up tomorrow and does this all again, the planner again tomorrow would eliminate consumption inequality. I just wanna,

- We, we end in 10 minutes and I think all of us wanna see the endogenous because that's the important issue.

- Let me

- Rich people don't like crying.

- Yeah, let me go there and, and while the discussion

- In about two years, oh in about two years, this entire discussion in line of research, that's gonna be irrelevant. Everybody's gonna have a laptop, everybody's gonna have an individual tutor, an AI individual tutor geared exactly to their current level of learning designed to accelerate it to the maximum where you live is gonna make no difference. The school system's gonna crash completely and we're gonna find an entirely new whole education model. And so I've been listening to all of this and nobody seems to understand the world is changing while we're talking. I also have one small comment apart from that and then you can go on. Have you read any books by Tom Sowell, by Thomas Sowell? Have you read any books?

- I don't think so. They're wonderful. So, but anyway, I dunno.

- I know, but I'm done afterwards. Otherwise it's all I'm saying is because in your paper you want to study race next you better read Tom. So before you even get started, okay,

- I'm the Thomas, so I'll tell you. Okay, I'm done.

- Thank you. Okay, so what happens if these spillovers are endogenous? So now the model, it becomes dynamic. The the entire structure is the same. So let me not repeat that and just highlight the thing that is different now is that what so far I was calling the exogenous spillovers. Now we are making them a function of what here is called five KT five. KT is the distribution of agents that choose neighborhood K at time T. And so in general, we're allowing this to depend both of the income, the parent of the parents in the neighborhood and the ability of the child. One possible functional form for that distribution that I'm using today is the one that you see right here where we're saying that the spillover in neighborhood key in every case simply a linear combination of the average income of the parents that are in that neighborhood as well as the average ability with some weight omega that spans different cases. So omega equal to one, that would be the case where spillover purely depends on parental income, say for example, because of the local public school funding. The other case omega equal zero would be it's purely pure effects as we were talking about before.

- Parents Yeah. Mentors, no models, you know exactly all sorts of things with parents, not just school

- Funding. They yeah, school funding is one example of broad, various broad interpretations that you can put under this formulation. So what do we do there? Well we're, this is the model that we take to the data where the key point, I explained it before so I'll not repeat it again, but we need to quantify how big is the neighborhood exposure effect. And we do that by replicating our model, the chat ender and exercise of thinking about the income gain for children of movers.

- Let's make one quick comment on that. So there's a job market paper by a PhD student in our business school, David Ritz Waller, who an econom attrition like Guido students that serve all these measures are in, in a incorrect things they need adjusting. So just if you take a look at it at some point, I don't wanna interrupt what you're doing, it's fine. You're using that's in the literature but something you wanna take a look at.

- Thank you. We'll do that.

- But but before we get, are we even gonna get to the possibility that what in the poor kids in the neighborhood makes the neighborhood really bad that

- That can happen? So let me, instead of showing this because this was a transition dynamics, let me show that using, using this because that is exactly, this is an application that we have done in our mar forthcoming Macron paper that studies the policies. The effect you have in mind comes out when you study the potential effects of doing a housing voucher policy such as the moving to opportunity program and scaling it up. So while the true program was a small experiment, we have studied what happens if you scale up this program and move a lot of children from the low income neighborhood, give them a housing voucher so they can move to the high opportunity neighborhood. But now the spillover is endogenous like the formulation that I just said. What is going to happen there? Well, for the recipient families. So these families that are moving there are large income and welfare gains. Why? Because now that the, the family's moving to the high opportunity neighborhood, these children, this child is exposed to the goods spillover, the rent is covered by the voucher, this child will benefit and have a higher income as an adult.

- Do you mechanically move somebody else out of that neighborhood?

- You do. So that is exactly the second point however is that scaling up this policy reduces the spillover in the destination neighborhood because for every child that comes in, you need to take somebody out more precisely how many people you need to take out depends on the elasticity of the housing supply in that neighborhood. At the very extreme you can say for everyone in, everyone needs to come out. It is somewhat depending on the elasticity. But the key point is that you're bringing in these lower income families because this spillover is endogenous and it depends also on family income. This, this policy is going to endogenously, recuse that spillover in that neighborhood. So the policy is partly self-defeating because you want to bring the children to the high opportunity neighborhood. But that bringing in all these voucher recipients is changing the spillover in that neighborhood. And so that has two effects. One, it has very large welfare losses on non recipients. So people that were living in that neighborhood otherwise now see the spillover of their neighborhood is going down and while at the same time the rent is trying is going up because there is all these new people coming in. And number two, the policy is partly self-defeating for the recipients themselves. Because the more people you're bringing in, the more the spillover is going down because of the mechanical effect through the income of the families that are going in as well as the endogenous response of the resident of the neighborhood, they're going to start moving out and so they're going to go to another neighborhood in the city. And so the, you're sort of essentially unraveling what is the highest bill over neighborhood.

- You're assuming that the transfer income has the same effects as earned income?

- This,

- Yeah, because if you've got all of a sudden the money to move to that neighborhood, that's, that money seems to be as valuable as earned income in

- This particular policy that I'm talking about, which is inspired by a true policy, this is a housing voucher. So the money they give you, you can only spend it on rent. That is the only thing you can do. They give you a voucher, you pay the rent and you need to go to a neighborhood that is a low poverty neighborhood. That is that policy. That's the only role of the transfer income. And

- You send your kids back to the school in the old neighborhood.

- I think in principle you could, but I don't think they would want to once they have moved to or that's

- What they did. In fact do. The moving to opportunity study showed really that there was very little effect on the children of the moving to opportunity. Mini de minimus effects, if anything be because they did so many different things besides including leaving the kids at the schools where they had left and then they would bus

- It

- Back. So because their friends were at the old

- Time. So I think what you're referring to, a lot of people did, didn't take up the voucher. So that's what the empirical studies of the MTO show that only

- Those who did it, there's a really excellent anthropological study of what people actually did with their children and, and it was quite different from the model.

- Okay, so that is great to know. I will read that too. Interesting.

- Relaxing housing supply elasticity, guns thing early on is that the, among the first best policies, if it could be done,

- So realizing the hand supply would help to some extent in the sense that you, you would have less of the mechanical effect that for every one kid that comes in, one has to move out. But that wouldn't solve general equilibrium forces that as you see the spillover of your neighborhood going down and the rent going up, some people would endogenously choose to move out. That would still happen.

- The key thing if land was completely elastic, both there was no house price variation at all. You could just build at the cost of production.

- Well there you still got, it has to be the case that when you bring a poor kid into the classroom, that he is raised more than everybody else is, is lowered.

- That's right. But that's a negative kind of, they're negative

- Everybody in those classrooms moving out. But yeah, so

- Related question, which is a little bit outside the model, but in the first and in the first statistics that you showed, the people that are in the rich neighborhood, they lose by desegregation. And so my question is to what extent you can imagine that, first of all, the first question is empirical, which is, is it true that in the rich neighborhood you have more housing restrictions? Because that would just prevent, you know, it's like it now by design, by political economy. I'm not, yeah, I know that you, you started with this two neighborhood, you know, in an exogenous way. But in principle you can imagine that the people living in the rich neighborhood, they don't want, okay, we have to go. I'll talk

- What let you conclude.

- Sorry. Yeah, yeah. Thank you. In any case, that was already my last, last plus one slide. So the conclusion is that this is a framework to think about the macroeconomic implications of labor exposure effects. Today I've shown you the key economic intuition on this framework. We have done many applications of it. One, as I said, quantifying the role of neighborhood exposure in the rise in segregation inequality. One is this study of policies that I was showing you one right now with the MTO and our new and work in progress in this research agenda is to think about the role of res racial segregation and to what extent information frictions and endogenous belief formation together with neighborhood effects have contributed to persistent racial inequality. Thank you.