 Edward Glazer from Harvard University to present the paper, Saving the Heartland, Place-Based Policies in 21st Century America. Thank you. And I'm very grateful to Lou for putting me on the program. This is joint work. It was written for Brookings papers with Larry Summers and Ben Austin. And despite the first slide, it's going to deal overwhelmingly with the US. But I think the themes of some places that have been left behind, which may in fact be associated with certain types of populist backlash, may indeed have resonance in the European Union as well as in the United States. So let me start with my one European slide here, which is a picture of European density and its relationship to both earnings and population growth. So the blue line that you're seeing here is average GDP per capita across the nuts three regions, where I've ordered the 1,114 nuts three regions on the basis of their density levels. And what that line shows is that the densest tenth of Europe's regions have incomes that are about 100% higher than Europe's least dense regions. There's just an enormous difference in this. There's, of course, a vast literature that my discussant has been a major contributor to on agglomeration economies and trying to estimate the extent to which this is actually a treatment effect of density or this reflect selection or omitted variables which both attract population and generate productivity. But it does certainly, without any doubt, point to the stark reality of enormous differences in wealth across place within the EU. The red line shows the relationship between population growth between 2000 and 2010 and initial population density. So whereas in the US in the 19th century, we left our dense enclaves on the eastern seaboard to populate the empty spaces between the oceans, the generally rising slope of that line shows that instead of spreading out at the start of the 21st century, we appear to be clustering in. I will tell you, this graph for the United States has a flatter line linking density with incomes and, of course, a much steeper line linking population growth with density. And these population growth density relationships, of course, reflect greater migration, not greater fertility in dense areas. I'm gonna be particularly focused on one type of spatial heterogeneity today and thinking about policy responses to it, which is heterogeneity in not working rates. So I'm not gonna be looking at unemployment, which I think has become an increasingly misleading figure about the state of labor markets. I'm gonna be looking at one minus the E-pop ratio, so the not working share. And I'm gonna focus primarily on men, almost overwhelmingly on men, simply because the labor force participation number for women is just more complicated and it just doesn't have the same meaning and many things about it are different. What you see here is the total not working rate between 1980 and today and the share of Americans who have been, share of American men, prime age men who have been not working for more than 12 months. So when I was born 51 years ago, less than 5% of prime age males were non-employed in the United States. For most of the last decade, more than 15% of non, of prime age males have been unemployed and have been non-employed, have not been employed. So that really is a huge change in the US and it of course moves with the cycles, but after every cycle it fails to come back in a full way and I think the rise of the long-term not working is particularly striking. In this way America is becoming more European in a sense. There's heterogeneity across space, this is the difference between, this is the E-pop rate for metro areas and for non-metro areas and what you can see here is that the non-metro E-pop rate is considerably lower than the E-pop rate for metro areas. So the non-metro E-pop rate for the US has been under 80% for the past couple of years. So you have more than a fifth of American, prime age American males who are not working in these locations. I want you to visualize it though even more clearly. So this is a map of the United States and the darker areas are the ones with higher joblessness rates for prime age males. The really dark things have numbers that are over 25%. So what you can see here is a broadband that begins in Mississippi and Louisiana, heads up through the Appalachians and ends in the Rust Belt which is the epicenter of American economic despair. So this is the eastern heartland of the United States and it's an area that sort of has been hit extremely hard and has failed to come back from these areas from those hits. There also is of course a concentration of joblessness in the West Coast and also in Arizona and New Mexico but two other things I want you to take away from this of course the prime coastal metro areas are quite light so they really are very different labor markets so New York or San Francisco but also much of what we'll call the Western heartland looks pretty good as well. So joblessness is not high in the Dakotas or in Montana or in Wyoming and that puts the lie on a simple view that high joblessness is associated with populist political revolt because these areas in the Western heartland where joblessness is quite low were also very enthusiastic voters in 2016 for the more populist candidate. This is the same map with the same colors in 1980, okay? And as you can see it's not that the correlations aren't there but the numbers are just so much lower in 1980 that the map just looks totally different. There's no place in 1980 where you've got anything over 25% except for a tiny, tiny sliver of Appalachia. The rest of America all looks quite different. This is the same map today for women and in this case it looks much more like a North-South divide than it does a divide linking the Eastern heartland with the rest of the US. So here the Northern states have higher labor force participation rates for women. The Southern states have lower ones and it really seems like it's more of a straight cultural phenomenon than one that actually captures large-scale economic distress. Now this is sort of my picture but I want you to combine it with a view that America is becoming in some sense a more geographically sclerotic nation that whereas throughout most of American history in contrast to much of Europe America was a country in which we would move very rapidly in response to local economic shocks. Whether or not there was the declining transportation costs of the early 19th century that enabled the farmers of New England to go and take advantage of the far richer soil in the Ohio River Valley or the move to cities in the late 19th century or the move to the Sun Belt in the mid 20th century or the move of African-Americans North during the Great Migration. All of this was a nation that was highly on the move. For the 40 years prior to 1992 the inter-county migration rate had never been below six percent meaning the six percent of Americans changed counties in every year. For most of the last, for the entire last decade the inter-county migration rate has never been above four percent. So it's quite a dramatic, it's a third in terms of reduction of the migration rates. It's really a very substantial decline in the mobility of Americans. This is not just about house lock just so the decline also shows up in renters as much as owners. It's a sort of larger scale cultural phenomenon. When we do see migration and this somewhat makes it difficult to think the right answer is just to build more homes in New York City or greater San Francisco. And I do believe that's part of the answer but migration also comes with a curse which is it is strongly skill biased in a way that leaves the people left behind with even fewer high human capital people around. And if you believe that skills are an important determinant of the long-run success of localities the out-migration of the skill is particularly troubling. So what you're looking at here is along the x-axis is the share of these are for PUMAs public use, micro sample areas in the US. The share of college graduates among the non-migrants and along the x-axis are the people who left that PUMA five years ago sometime in the last five years. And as you can see particularly among the less educated PUMAs that the out-migrants are much more educated than the people who stay there. So that's the difference between the dot and the 45 degree line that you see that. So there really is a very severe de-skilling that occurs without migration which makes it hard to think the right answer is just we should make it easier for people to build housing in San Francisco and let them all move there. That may be an attractive long-run solution if it ever could be remotely feasible but at least in the short or medium run it leads to a hollowing out of the places where the skill to leave behind. There are added changes that are important. One fact which is associated with Peter Canong and Danny Shoag is that migration has not only declined it has ceased to be directed to highway areas. So historically Americans had moved to places that had higher nominal incomes, higher nominal not worried about the CPI but worried about local price of the seats. That has really changed particularly for less skilled Americans. So there's no sense in which the migration rates of Americans are any more particularly targeted towards the highway parts of the US. Secondly successful areas have made it increasingly difficult to build low-cost housing which leads to a spatial mismatch of the United States. And this is a point that I made 13 years ago and Shay Amireti attempted to put a quantification on it but some significant fraction of US GDP is probably lost by the fact that we don't allow people to build in the areas that are more productive and of course that is one explanation of why we no longer move to the productive areas that when the farmers moved west in 1830 they could build balloon frame houses for pennies on the dollar in cheap areas. Today you wanna build a starter home in Palo Alto which is five million dollars. Okay three I think I'm actually underestimating this. There's a strong correlation between the change in the share of the population with a college degree in an area and the initial share of the population with a college degree. So you can see this Amireti originally identified this in 2004 and you can see this quite strongly in the data. And the last point which I sort of really wanna emphasize for this for anyone who's sort of more macro oriented is that income convergence at the local level has stopped. In fact the last time that we saw it was in 1990 and then Barrow and Sally Martin wrote their paper and then it went away. But it was a phenomenon in the US for 130, 140 years but it really has ceased to be. And just if you remember all of those graphs from Barrow and Sally Martin that showed just a powerful negative effect this is what that graph looks like. So it today 1980 to 2010 so it's still slightly negative but if you do even the slightest thing to correct for the measurement error induced mean reversion like for example instrument for the median earnings with the 75th and the 25th percentile earnings this becomes positive. So mean reversion of income levels at the local level in the US really has stopped being a phenomenon. And even more striking is the persistence of not working rights, the persistence of joblessness. So some of you may remember table two of Blanchard and Katz's paper from the Brookings papers from 1992 or so. Table two showed a stark relationship between unemployment at the state level in 1985 and unemployment at the state level in 1975. There was none. So it really led to the view that America ironed out its unemployment rates incredibly quickly. This is what joblessness in 2010 looks like when we regressed on joblessness in 1980. That's a correlation of over 80% at a coefficient of more than one. So there's no sense that America's high joblessness areas are somehow or other disappearing. We become a nation in which these things are much more fixed. Last point I want to make in terms of preliminary facts is just there's a reasonable view that not working is a much larger social problem for men than purely low incomes are. So we've worried a lot about inequality over the past 10 years. We've worried much less about the rise of joblessness among prime aged men. There at least is some data which suggests that the problem of joblessness may be a much more severe social problem. I have a big exponent of the view that we should never use happiness data and think that we're capturing utility. There's no reason why human beings should maximize happiness and there's not a clear sign that we do. But the fact that the non-working men are so much less happy than the working poor and this is what this shows. So if you ask what share of the population is dissatisfied with their lives? Earning more than 50K, that number is about 2%. Earning between 30 and 50K, it's 4%. Earning less than 35K, it's 6%. Non-employed, 18%. Quite a large difference and it's particularly striking if you look at those who are both non-employed and working alone. So you remember 2% among the employed males who are earning more than 50K. What's the number for long-term unemployed who live alone? Over 30%. So this is a phenomenon that's associated with social isolation, a deep sense of purposelessness in the world and a deep amount of misery. There are some other attendant things that go along with this. This is the geography of opioid deaths in the United States. It doesn't match perfectly the joblessness rate but it certainly is suggestive. So that this drug fatalities are linked with this. Divorce rates, suicides have all been linked to joblessness in a serious way and of course the government spends where the people are jobless. So this is a US government expenditures on a per capita level across states and again this matches up reasonably well with Eastern Heartland. Just to show you a couple of dynamic facts here to simplify things and not give you every state and the union, I'm gonna just carve things up into the Eastern Heartland, the Western Heartland and the coast. And my definition of the East versus the West depends upon when the state was admitted into the union. So pre-1840 year in the East, post-1840 year in the West. This is joblessness by the three regions. So the Eastern Heartland has been above the other two regions since the early 2000s and then again it was in the 1980s. The gulps are between the Eastern Heartland and the Western Heartland which is the biggest difference, the coasts are actually between, is about six or seven percent in terms of the joblessness rate overall. If you wanna look at the three regions, all of them have seen their GDP go up of course in real dollars but the growth rates have been much faster in the coasts and the Western Heartland that they have been in the Eastern Heartland. That comes from two separate phenomenon. What's happened in the coasts is that GDP per capita has gone up faster and that's what you're seeing here. The coastal states are the blue ones and they're the ones with faster GDP growth per capita. The Western Heartland is barely ahead of the Eastern Heartland but in terms of total employment growth, it's hard to make it out from this graph but the rate of growth in terms of employment has actually been a lot faster in the Western Heartland. So bodies have been going up faster in the Western Heartland, dollars per body have been going up faster in the coast. And finally in terms of the trans, this is a, many of you may have read the Case and Deaton paper on the stalling of life expectancy gains for prime age men in the US. The epicenter of that again is, of course, the Eastern Heartland and that's exactly what you see. So these are prime age mortality rates. Historically, the Western Heartland was the healthiest. The coast in the middle and the Eastern Heartland was slightly worse. For a brief period in the late 1980s, early 1990s associated with both crack cocaine and AIDS, the coasts went above the Eastern Heartland in terms of prime age mortality but now the coasts in the Western Heartland are both much lower whereas the Eastern Heartland has been treading water or even slightly rising for at least 20 years in terms of mortality. So it really is a deeply troubled region. I am not gonna try and sell you on any particular story for why the Eastern Heartland looks different or why joblessness has increased in the US. There's a very nice survey paper by Catherine Abraham that attempts to decompose how much of this is about labor demand and how much of it is forms of labor supply. She finds more emphasis in favor of the labor demand side than the labor supply side. The labor supply side is both about various federal government programs which make not working less painful but also the ability of friends and family to provide support for non-working men and 35% of non-working men in our sample are actually living with their parents. So there's a fair amount of family transfers that are going on in various ways. I have a simple two factor model of economic growth that I always carry around in the back of my head which emphasizes the combination and interaction of education and institutions and these are areas which both of these areas are areas in which the Eastern Heartland is lagged. It's traditionally been a less educated area and here you can see that this is true in 2015 as well and this reflects two different phenomena. In the Southern States, this is the legacy of slavery and the Jim Crow South where there was less investment in education because it was a highly agricultural area and schools were seen as being a potential route towards rebellion. In the North, this is the legacy of the industrial past that workers could get good jobs working on the assembly line and didn't need to get college education to do it. The Western Heartland by contrast was the birthplace of the American High School Movement and the Western Heartland has typically been actually better educated than the coast, historically. So this is the prime college, primal college education rate in the three regions. As you can see, as of the late 1970s, the coast and the Western Heartland were about equal. If I went further back, the Western Heartland would be ahead. Over the past 30 years, the coasts have surged ahead. The Western Heartland has been relatively stagnant but it's still above the Eastern Heartland and the Eastern Heartland has caught up. And I don't know if that means that the next 30 years will be better for the Eastern Heartland because this education is coming up or the next 30 years will be worse for the Western Heartland because their education is lagged behind. There also is at least some suggestive evidence that the institutions of the Western Heartland are, the institutions of the Eastern Heartland are somewhat weaker. I wrote a paper maybe 15 years ago called Corruption in America where we use federal charges against local officials for corruption violations of federal law. These tend to be concentrated in the Eastern Heartland. So Mississippi and Louisiana for most of the last 30 years have routinely led the beauty contest as being the most corrupt places in the US. And this is from a more recent paper by Liu and McSell which sort of shows again that the Eastern Heartland has this institutional problem. And there's a bit of speculation in the paper about why the institutions would be weaker in the Eastern Heartland. This shows occupational licensing. The pattern lines up a little bit less well but at least there's some suggestive evidence that at least this type of excessive regulation, perhaps excessive regulation of, in this case this is a state regulation of opticians. So those are not the guys who actually check your eyes who one can reasonably argue should be licensed. These are the guys who tell you whether or not you should have ground glasses or square glasses. So the regulation presumably makes sure that you don't have someone recommending ground glasses when you should have square glasses for your eyes. Okay, now all of this leads us to the question, is geographic sclerosis, is the hardening of America's geographic mobility an excuse for revisiting place-based policies just as the greater sclerosis in the EU was always seen as at least something of a justification for place-based policies in the EU. Okay, so let me just go through the reasons why for most of my career I have argued strenuously against place-based policies of any form, particularly the large infrastructure ones advanced by the EU's. Counter argument number one, subsidizing declining places keeps people in dysfunctional local economies. It's not false, but this becomes less important with lower migration rates and somewhat less important if you're particularly worried about the outmigration of the skilled. Counter argument two, subsidizing any place leads to capitalization in rents. The poor tenant, I mean the poor renter, who doesn't like contemporary art may well be worse off because of the Bilbao Guggenheim. The rents went up, they don't like Gary, and again, as people are less mobile, this would be less important. The relative importance of capitalization versus distorting migration depends upon the elasticity of supply of housing. So if you're looking at a declining place like Detroit, where essentially you're on the vertical part of the supply curve, meaning that they're not building any new housing and it's just declining, are places in which you expect it to show up in capitalization rather than distorted migration. Places where more elastic, you expect it to show up in terms of migration. Counter argument four, some place-based policies can create pockets of high unemployment and low human capital. This is still true, but presumably doesn't need to be, it depends upon the design of the strategy. And counter argument number four is that when infrastructure gets decided not on the basis of rigorous cost-benefit analysis, but based on myths of place-based rebirth, all sorts of horrors can occur. Okay, well, that one's still true. Okay, so this is the Detroit's Peeper Mover monorail, built in the 1980s, which as you can see from the picture, it glides over essentially empty streets. Okay? You can run a bus at any speed you want in Detroit. It's a city that was built for 1.85 million people. It is less than 800,000. And the fundamental fact is that the defining feature of declining areas is that the population declining is they have an abundance of structures and infrastructure relative to the number of people, relative to the level of demand. More than 90% of the homes in Detroit were valued by the market at considerably low construction costs in 1980. It never made sense to have policies to subsidize housing in Detroit. And it certainly doesn't make sense to subsidize a monorail of this form. Now, the case for place-based policies, right, has there are three elements of it. The first is about various forms of local externalities, okay, which is, you know, the decentralized spatial equilibrium is unlikely to be a Pareto Optum. So agglomeration economies are generally accepted, at least by people who are in this world. A typical number would be, we think that delog income or delog value added, delog productivity per delog density is like 0.06. That would be the sort of number, okay? Congestion externalities on the other side are also quite real. There are a variety of different estimates of those. Human capital externalities may be more contentious, but certainly if you believe any of the Moretti numbers, they're big numbers, right? So they're potentially quite large. Typically as the share of adults in your metropolitan area with a college degree goes up by 10%, your earnings go up by 10% as well. Holding your years of school in constant. Now that's more bedeviled by the selection of high human capital people into these areas. But these externalities mean that a decentralized spatial equilibrium is unlikely to be a social optimum. But as I argued in a Brookings paper 10 years ago on these things, that doesn't mean that we know how we should actually want to shape people, shape the movement of people. So it is certainly true that we're probably not at an optimal, but unless you know not just the existence of these local externalities, but the full shape of the curve, I don't know if I wanna move people from West Virginia to New York or New York to West Virginia. Their externalities from both moves and knowing that there is a spillover from this productivity is not enough to know the direction of where that externality is larger and where I should shape people around. I don't believe that we are ever going to have good enough sources of exogenous variation to pin down these local externalities to be confident about the shape of these things enough to move people around. So I basically think that we just can't use this as a justification for place-based policy. It's often advocated, but we just don't know enough about whether or not where we're gonna be gaining welfare triangles and where we're gonna be losing them to be at all confident that we're gonna be gaining versus losing by trying to move people around. Place-based argument number two, which I have no argument with in terms of the logic of it, is just an insurance or if you want to be rousing about it, a behind the veil of ignorance kind of equity argument. So in 1969, Detroit was slightly richer than Boston. Today, Boston incomes are 40% higher. Surely ensuring individuals against shocks to the local economy would be well-fair in enhancing in some ideal world. It's almost impossible to think that that wouldn't be. Of course, there's some counter-move in terms of the cost of living, but for the owners of Detroit real estate, that counter-move isn't exactly a plus. It could be pretty non-distortionary based on place of birth, although place of birth is pretty inconceivable as a policy outcome. A related argument is that place itself may be a marker for low income and in some cases less distortionary than low income itself. But the big limitation on this is that the only really natural political unit on which you could base these insurance levels are states. And states account for a very small amount of the variance of incomes within the United States. About 1.2% of US income variability occurs at the state level. So even if you could wipe all that out, you're achieving very modest welfare gains for a huge political lift of actually making this thing occur. Now, of course, if you could go down to micro levels, if you could go down to Pumas, or if you could go down to blocks, then you'd get a much higher variance. And indeed, the recent work of Raj Chetty and Nathan Hendren shows us that you get a significantly higher R squared, maybe nine or 10% going down to the block level based on birth. But how could you possibly imagine an effective government policy that bases things on block of birth? So that brings me to the third argument. And that's the argument I'm gonna try in advance. Which is that there are different labor markets within the US. I'm sure there are also different labor markets within the EU, right? Different labor markets with different elasticities, call for different social insurance policies and call for different employment subsidies, okay? That's just a flat-out application of any sort of formula that you want for this. If you have a different labor supply elasticity in one place than another, you're gonna want different formulas. I'm gonna try and put some numbers on that going forward. But let me give an extreme example of the need to have place-specific policies and we know we less disdain are different. The US government has a uniform federal construction subsidy and the low-income housing tax credit, right? There's a reasonable argument that this tax credit is Pareto improving in places which have artificially constrained housing through local regulations. So in New York, in San Francisco, and LA it's not a crazy policy. In a city like Houston where there's essentially completely elastic supply of housing, these added subsidies are clearly related to one-to-one crowd-out. They're clearly doing absolutely nothing to housing supply and it would be crazy in a city like Houston to think that you actually wanted to subsidize it. In Detroit, there may not be any crowd-out because you wouldn't build without the subsidies but it's clearly not optimal either because you don't want more housing there. Example number two is hotspots policing. There's a long claim in the policing literature that you wanna throw the cops where the crime is. Similarly, you wanna subsidize employment or design your social insurance policies more in places where there's more of a response to that. So if you have a limited pool of employment subsidies you wanna throw it in places where you think those subsidies will have more effect. So the argument is that in high employment markets, policies that deter employment may not matter. In high non-employment areas, policies that deter employment may have awful consequences. So think about a $15 minimum wage in Seattle versus West Virginia. So it's gonna be tiny in Seattle in terms of its effect as recent studies have shown and perhaps Mazum in West Virginia. So the question that we ask now is is the marginal impact of an employment subsidy higher in West Virginia than in Seattle? And this is just my view of the world. I have a distribution of the net return to working and it's moving around over space. And I've got the places that I have, lots of people are not working, the density is higher. That's just my view of the world of this. Now, the explicit spatial proposal is that we want to subsidize working more in places where the response to that subsidy are gonna be larger. The nice version, which I associate with my always generous and kind co-author, Larry Summers, is that he wants to just hand out more money in places where we think the employment response will be larger in places like West Virginia. The mean version is I'm gonna take away something else and effectively tilt the benefits. So I'm gonna do more to subsidize employment in West Virginia and do less to subsidize non-employment. So as you think about it, you can do this in a way that it's actually spatially neutral in terms of where people locate, but just work it tilt that responds to local labor mark conditions. And that's the mean version of it. I do have a point that I wouldn't trust the locals to do this because it could easily lead to a race to the bottom where you try to deter poor people from coming to your areas. So we now do a bunch of different work on trying to estimate whether or not there are greater responses to a variety of labor demand shocks, which are seen as being mimicking what would an employment subsidy look like in places which have higher levels of joblessness in the U.S. So we do this with three things. I'm not gonna walk you through the numbers, but they all give figures which roughly point in the same direction, although none of them are completely nailless. And I think this is really an invitation for more work, not a claim that we finished it. One of which is just the classic bardic industry composition shock interacted with national trends. This shows the bardic shock. This is 1977, I think, to 2006. And we're interacting the bardic employment growth number with the historical not working rate. The thing to pay attention to are these negative coefficients, which tell you bardic is reducing, the positive bardic shock is reducing non-employment more in places that have high historical non-employment rates. This is, and it goes in the opposite direction on housing prices, this is at the Puma level again, a strong negative interaction there. This uses Auderdorn and Hansen shocks. Again, in this case the shocks have the opposite sign, these are positive shocks, but again, they have more of an effect on not working. They have more of an effect on joblessness in places that have high joblessness levels to begin with. Third shock is the Nakamura Steinsen shock that comes from federal spending. Again, they have more of an impact on joblessness in places where the historical joblessness level is high, which shouldn't surprise you. When everyone's employed, the impact of a labor demand shock on employment is gonna be zip, right? In places which have a 30% unemployment rate you would expect to see a larger thing. That's exactly what we're seeing in this data. So higher joblessness areas seem to have higher joblessness responses to various shocks. Now, we then, in the paper, modify the Bailey-Ceddy formula in a way that tries to incorporate this. And we then are going to, you end up with a formula that relates the marginal utility of consumption in the employed and non-employed state with a formula that takes into account what you think the employment response will be to a more generous payment to being non-employed. So you plug in this elasticity into this formula and you get out predictions about what things should look like. We then calibrate this, I'm running out of time, and we're gonna be calibrating this off of an externality from not working, which is just based on the fiscal externality of not working. So lost taxes, increased government benefits. So our excuse for an employment subsidy is just straight pegoo 1912. You're gonna subsidize employment because there's a welfare loss that you're not incorporating, which is the fact that you're imposing costs on the government by not working. If we wanted to throw in the cost on your parents as well, depending on how well you think cozy and bargaining works within the household, then that number would be much larger in terms of the externality of not working. We're gonna use CRA preferences and range of values for risk aversion. And this basically gives you what the ratio of the, this is the employment subsidy relative to wage, should be based on how responsive things are and based on how risk-averse you are. So the more risk-averse you are, so the higher gamma, the less responsive you have, the more you wanna keep those employment, those consumption levels constant. Obviously, the less concave you are, the more comfortable you are having larger differences. We calibrate this using the Arctic shocks to get these different elasticities. And the type of thing we get is that the, let's say if you have a coefficient relative risk aversion of one or 0.5, you get in Wyoming, which is a very low non-employment rate, your consumption when non-employed is supposed to be 92% of your consumption when employed as opposed to 72% in West Virginia. So it implies at its most extreme level a fairly large gap between the states in how nice we should make working versus non-working depending up because of those very different labor supply responses in those areas. So as we push towards a full bore spatial policy, I think the right answer is that both social insurance policies like disability, like UI, should take into account local labor market conditions and in some sense should do less to subsidize not working in places where the not working response is larger. We should have employment subsidies that are larger in those areas and probably place specific educational interventions until they don't really have time to discuss that and things like place specific regulatory reform. I don't really have time to reflect on US EU differences. The case for EU spatial policy was always somewhat stronger than the US partially because migration rates are so much lower and partially because just the difference in climate between the areas the US just so much smaller than it is in the US. Yet from your perspective, many of the social policy decisions are based at the nation state level rather than the ECB level. Nonetheless, I hope some of this will be useful for the EU context. And just I wanna end on a caveat which is as much as I'm arguing for place-based heterogeneity and thinking about insurance policies, I am not engaging the full-throated support for any of the large-scale place-based policies that the EU has advanced over time. And the possibility for waste from these policies is enormous. So even when they work, so Guggenheim Bill Bow is a spectacularly successful museum but Bill Bow's unemployment rate is still 19%. It didn't fix unemployment in Bill Bow. And that's a good museum. That's choosing the max in terms of the quality of success for a local cultural thing. Anyone remember Sheffield's National Center for Popular Music? That one was also a place-based intervention that closed within the year. So again, have policies that are targeted towards local conditions but don't think that you wanna arbitrarily pour money into areas or that you can magically by spending it in an area on a museum cause some area to reverse its economic trends. Okay. Thank you very much. So we have the discussion Diego Puga from Senfi. Okay, thanks. So I'm happy to hear that it's gonna be a paper. You know, the papers are really about place-based policies. You know, this can be a wide variety of things but basically any place-based policy, what has in common is that it's going to target specific geographical areas as opposed to targeting specific people, specific firms. So, you know, as said, as told us, and as we all aware of, economic views have not traditionally been very favorable towards these place-based policies, at least in the US. For two main reasons. First, because we have this perception, you know, stuck back in 1992 and not just Barot-Sala-Martin but also, oh, sure. Is that here or is this a microphone? I'll send you the gate. Partly from Barot-Sala-Martin but mostly I think from Blanchard and Katz that migration was a very effective adjustment mechanism for, you know, inequalities in the US. Whenever there's a negative shock, people moved out and things adjusted fairly quickly. And also more generally, worries that in any case, anything that we do in terms of place-based policies that are going to shift things around from one place to another, we thought really having a big aggregate effect. Two things have changed recently. First, the magnitude of both gross and net migration flows has dropped. And I think this is a point that Ed has stressed a bit more in the presentation. Maybe in the paper I think it would be worth pushing this a bit more. In fact, what's important is not just that migration flows have fallen but the fact that they're less directed, right? So of course you could have a following migration but still, you know, the net effect is that 2% of people are going towards places that are growing. You can still have adjustment, right? The key here is that not only overall flows have fallen but actually they're less directed than they used to be and this is a point that's been made clear in the presentation. And second has been this rapid rise in the share of primate males specifically who are not working and this is a rise that's concentrated in the eastern heartland of the US. And the authors argue that this may make place-based policies something that they're worth reconsidering, right? So of course it's a very useful paper. First because it brings attention to this fact in the US that disparities are rising and migration is falling and this is something that, you know, if you think about non-working rates I think there's some brother awareness that this is happening but probably much less so is that this is a phenomenon that's geographically concentrated and also I think in terms of migration we still tend to think of the US as a fast adjustment, high migration country and maybe we're not so aware of this fast decline that's been happening. The paper also has a very careful discussion of what is needed for place-based policies to make sense and this is something that I think is, you know, that there's lots of literature discussing place-based policies and most of it, almost all of it is not nearly as careful, right? So just to give you an example this argument for agglomeration economies and justification for place-based policies well, you know, as Ed has made clear what you need for place-based policies to make sense because of this is not that there are agglomeration economies but actually that there are different across space and we really know there are agglomeration economies we have a good sense of the magnitude when you get in terms of spatial differences that's a much more difficult thing to assess, right? And he stressed, you know, they need to know about differences in agglomeration economies in fact you also need to know about differences in congestion costs and this is something that we know even less about, right? How fast is congestion going to rise in San Francisco if we let more construction take place? How fast is, we really don't know even less than with agglomeration economies? Also the paper is useful because it has an empirical analysis of whether the key conditions are met in particular it looks for evidence of heterogeneous employment elasticities and the evidence is just suggested but at least it's making an attempt to get some numbers on that and then finally has a balanced discussion of the potential place-based policies that you can have so it looks at different alternatives looks at what policies make sense which don't make sense, feasibility politically in terms of magnitude, different alternatives, right? So it's I think a very useful contribution in that sense. The paper, so let me then spend most of our discussion just giving you both some additional comments maybe a few European numbers or more Spanish numbers you will see than European generally. And just some quick reactions quickly partly because I've only been asked to discuss this over the last few days so it's been a relatively, I have read the paper before fortunately but it's been sort of a quick putting together of the discussion. So starting point is that networking rates in the US are high and they're heterogeneous so they've gone up from around 10% in 1980 to close to 20% today with, and they're also heterogeneous so you have huge differences like around 50% in Flint, Michigan, 5% in Alexandria. This contrast with this perception of the US as a country where non-working rates are low and you think of the levels I mean just getting to Spanish kind levels of non-working rates, right? Which is quite amazing, right? So the non-working rate for primates meals in Spain is also about 20% and the comparison of differences is not quite right because I'm doing this at the level of autonomous communities as opposed to finer areas as there but there's also a broad range going to close to 30% in Andalufia versus 10% then in La Rioja, right? So this is not just that there's a rise in these differences, the levels actually are getting up to the levels of high unemployment in European countries. And it makes sense to focus on this partly because I agree with Ed that the line between being unemployed or being out of the labor force is really blurred, right? And we see lots of people crossing that line frequently in the data. There's also this evidence that in terms of happiness, in terms of well-being non-working might be more relevant than the level of income and in any case, even if we worry about income not working is going to be a big component of income inequalities, right? In the case of Spain, for instance, there's this recent paper in the EJ by my former colleague, Stephan Bonham, now at Chicago and Laura Speedo from the Spanish Central Bank showing that the rise in income inequality in Spain after the great recession is mostly due to non-working rates. In particular, lots of primates males with low education levels who are working in construction getting around median income who have dropped out of the labor force. So you have this hollowing out of the middle of distribution, but it's happening through precisely non-working rates. So even if we care about income inequalities, non-working rates are going to be an important component of that. So this is the drop in migration rates in the US, just taking the lower chunk of that diagram in the paper. You can see this fall from around 7% for a long time to under 4%. And you know, a first thing that you might think when you look at this kind of picture is, well, if part of what's been happening is that adjustment has slowed down. We have this fall in migration rates. Shouldn't before we start thinking about place-based policies, shouldn't we think about how to recover those migration rates, how to get them high again and maybe recover that adjustment? So why are migration rates low? Well, you know, out of all people having been telling us about this for 10, 15 years, how important regulation is in preventing high potential growth cities in the US from growing even more. However, I think that there's some perception that this kind of regulation is something that you have some fast-growing cities like San Francisco that have high regulation, all those fast-growing cities like Houston that don't. Actually, no, I want to just show you quickly a few pictures showing that this regulation is something that's systematic, right? So this is a picture of house prices as a function of distance from the center of the city for different American cities. Some things here, I think you would expect this from a standard urban model, monocentric city model, right? So of course, house prices are declining with distance from the center, prices of the center are higher in bigger cities, bigger cities expand further. What you really wouldn't expect from a standard urban model in this picture is the fact that not only prices at the center of the city are higher in bigger cities, prices at the periphery at the edge of the city are also higher in bigger cities. And this is not something you would expect in a model without regulation because the price of land at the edge of city is very similar across the US. Construction costs are also not very heterogeneous. So unless there's big differences in regulation, you wouldn't expect this. And you do not need not just big differences in regulation, but this regulation to actually be systematically related with the size of different metropolitan areas. And in fact, that's what you see. So if you plot the house price at the periphery more generally against population, you get this very clear positive relationship that's been driven by stronger regulation in bigger cities. And as you can see, by plotting this water regulation index against population, again, a very clear relationship, that regulation is keeping periphery prices high that you see here. And as a result, you know, whereas normally you would expect if house prices go up in some region, the construction isn't going to react by building more there. If you look at construction against periphery prices, you basically see no relationship, okay? So there's really an important limitation imposed by regulation on the expansion of cities. Now, can we fix things by addressing this? I don't think so. I don't think so because first of all, few people who are not working move to look for a job elsewhere. This kind of migration is something that as economists, we like to think a lot about. We have very nice models like Ken and Walker looking at this sort of thing. In practice, you know, less than 6% of all moves across U.S. states, less than 3% of moves within state are by people who don't have a job and move to another place to look for a job. Okay, so this is just a very, very small fraction of migration. If we free up housing in cities, the highest killed are maybe going to move there and take more dens of this. They're not working. They're really not going to react a lot to this, I think. Okay, and also, you know, if we look at this drop in migration rates, we saw this picture for the U.S., you know, I can show you the picture for Spain. It doesn't look very different. Okay, and here we don't have the same set of differences in regulation across cities, right? So it must be something else also that's driving this in part, okay? Still, you know, this doesn't mean that there are no gains to be made from freeing up regulation, but they're not for the kind of problem that we're looking at today, these non-working rates, okay? How about other barriers to migration? Well, you know, there are certainly local welfare benefits, housing vouchers, locally administered federal programs, those sort of things. Those sort of things probably are deterring mobility. Now, given that those, you can describe as type of place-based policies, does this mean that this is an argument against place-based policies? Well, not generally. What this argues against is place-based policies that are based on residents. You don't want that sort of thing because that's precisely going to the term mobility even more. But I think the paper is clear about that. It's not pushing for residents-based policies, rather it's pushing for policies that are based on where you work rather than where you live and that's not the same guy yet, okay? So, you know, let me just end up with a few tweaks I think I said. I think I agree generally with the message of the paper. So what I have are more tweaks a bit on the emphasis and things that I maybe are worth looking at a bit more. So I think the paper exposes really clearly the symptoms. I think it really doesn't take a very clear stand on where the problem is coming from. I think that's partly because it's a very difficult issue. So, you know, non-working rates rising is a very important problem, but it's something that we don't understand very well. But I think it might be useful to take a stand of where it's coming from, and especially where differences and where this concentration these in Harlem are coming from. So I think, you know, the paper provides some evidence on this. I think at the end of the day, the authors are telling a story about the combination of negative employment shocks with low education. But I think it's probably worth making that more specific. They're taking a stand saying this is what you have in mind if that is the case. Also, the paper focuses on Eastern Ireland. Now in the presentation, Ed gave us a different kind of data in terms of large metro areas versus smaller metro areas and non metro areas. I think not as a substitute versus a compliment, it might be useful to look at things that way because this may be also largely not so much a problem of just Eastern Ireland versus the rest, but maybe large metro areas versus the rest of the country. So maybe I think that that cut that you made in the presentation is actually useful. A useful one to have as a compliment. I think it's also important to understand what is the market failure that we're trying to address, right? So there's certainly a discussion in the paper of why workers prefer not to work rather than working for low wage, and why we don't have more mobility of non-working primate males. But, you know, why is economic activity not going to Eastern Ireland? I think it's important to think about and to understand if we want to think about what kind of place-based policies might work. The paper makes an effort to try to assess different processes of employment, and for doing that, it does three exercises. It analyzes military spending shocks, okay? And that I think is very relevant and it's careful, but it's probably something that's, I think the paper is also clear about that. It's something that's relevant for thinking about that kind of program, but not so much transferable to other kinds of welfare programs. It talks a bit about, you know, where we would really want the evidence to come from, which is looking at the past, looking at place-based policies that have been implemented in the past in different regions, and seeing what the effects have been different, depending on the area. Unfortunately, there's not a lot of literature there, and there's little they can do more than calling for more work in the area, right? So given that that work is not there, they turn to some analysis of essentially Bartek, the MAC shots, or the China syndrome kind of shock, to look at what diseases might be. Now, this is nice and interesting. My worry with this is that the fact that some place got hit harder from rising imports from China, or the fact that, you know, some decline in traditional manufacturing has hit more a region where there's a concentration of high, a high concentration of non-working primate males. That doesn't necessarily mean that we subsidize employment there. It's going to react more, right? Because maybe some jobs were lost in traditional manufacturing. The workers that were working there who lost their jobs have low education. Now, if we subsidize jobs, maybe the jobs that are going to be created are not going to be the same type of jobs. Maybe the jobs that could be created are not suitable for this kind of workers with low education. So maybe the effect is really not going to be the symmetric one to this negative shock, right? So, you know, it's a just if evidence. It's interesting. But I worry about drawing the elasticity of that to design the, I think that's my main caveat in terms of the analysis. Then I'll finish with this. If we think about the different kinds of policies, I think I basically fully agree with that on the first two. So in terms of targeted location decisions for public employment, I think they're possibly effective, but the scope is probably not going to be something that gives big effects. And I think we agree on that. Interest infrastructure, I think infrastructure can be very useful. We have evidence that it can have large effects. But I think there are two reasons why this is not the route to go. First, you know, although we found, the literature has found, big effects of infrastructure in the past, I think there are no projects like, you know, building the interstate highway system in the US, or it's defying the appellations that are going to be left. So as we build more and more infrastructure, the marginal products have smaller and smaller effects. But also this kind of place-based policies that tries to help areas by building big infrastructure, whether it's that modern music museum in Sheffield, or whether it's high-speed trains that go to nowhere, these things are not very effective, right? And one way to see that is if you, you know, if you look at work in urban economies, looking at the effect of highways, whether it's through a turner or an eight-bump snow, there, they worry about the indogeneity of infrastructure, the fact that you might build more infrastructure which is most needed. And for this reason, you might want to instrument for what infrastructure is actually built. In fact, when the instrument for infrastructure, they find that the IV effects are not smaller than the OLSX, they are greater. This tells you that you're building more infrastructure, not where it's most needed, but where it's less needed. It also tells you that those projects that you build in areas where it's not needed have really two of an effect. Okay, so there's really not a big scope, I think, for fixing things by building lots of infrastructure. So that leaves us with employment subsidies or education retraining. The paper focuses more on the subsidies. You know, my caveat there is what I mentioned about the facilities. I think I would probably put a bit more of a focus on education retraining. And here, let me, and then I'll be finished just with this comment. Here, you know, my worry, does not worry about the paper, does not worry about my own statement about focusing on retraining, comes back to the picture that I showed us about corruption and the quality of institutions, right? Because if I think of my own country, Spain, and the effort it's making on retraining, we have this, again, high non-working rates, people who are working in construction, who have low education. We now have, you know, this worry that, increasing worries that our vocational training programs are not up to standard, certainly far away from what's been done, for instance, here in Germany. So there's been some work towards making those programs more relevant to skills that are needed in the market, to things that are more reactive, where firms are more involved. The, you know, there's no systematic work yet comparing how this is happening across Spanish regions, but you do have work on specific programs in the specific regions. My reading of the literature is that this has been very successful, precisely in regions that are really doing relatively well like the Basque Country, where they're very careful to involve the firms in terms of seeing what skills are needed. They react very rapidly. So we're missing this kind of programs that have this skill. Okay, so let's do some courses that do that, and they react very quickly, and that has a very big effect. Whereas in other regions, precisely the ones where this would need to be more effective, where non-working rates are higher, you know, this is much more the tax. Programs have been designed like traditional programs, without really thinking about what is really needed. And then the rates of people who actually find a job and then stay in a job for a long time are much much lower, right? So here the devil is in the details, and I worry that precisely what is needed is a quality of institutions that's correlated with non-working rates. And I'll finish with that. Thank you. Thank you very much. So let's collect some questions from the audience. Second row. I have one question, like, how, Ed, do you think about how the sort of great recession fits into this picture? I know that mobility rates based on your picture state declined before the Great Recession, but obviously great recession is associated with millions of foreclosures, decline of housing wealth. A lot of these regions, I presume when you have high share of people that are essentially not working, we're also fairly heavily hit by recession. So to what extent is this also like a story of a decline of housing wealth and credit worthiness that prevents migration to better places, in the sense that, A, if I suffered foreclosures, my credit score is very impaired, my credit card lines get cut, instead of moving to a high productive place, I'm gonna move to a place that's like incredibly cheap, but they're also not maybe jobs. B, if I'm really a young male and my parents suffered foreclosure, or for example, their house is worth way much less than it was, and in slow recovering, they cannot really help me financially to move to a better place. And to what extent is this partly an artifact, partly just an effect of the Great Recession and the slow recovery? If you could hand over the mic, and then to the back. There's a complimentary set of facts, I think relevant to your analysis that involves the general decline in labor market dynamism, that is job to job transitions have steadily fallen. So the question is, could the decline in migration contribute to the decline in job to job transitions, or how much could be the reverse, the decline in these transitions could affect migration? If you could just pass the mic, thank you. So you've shown some trends and some facts about the not working population, and you used these trends and facts to justify maybe some place-based policies. The not working variable has a big advantage, it's very well defined, it's binary, zero, one. But it has a disadvantage of living out of the analysis, the working pool. People, if you add those people in your analysis, the working pool, does the picture change a lot and do your arguments for place-based policies weaken or are they stronger? For the questions. Pepeck. Thanks a lot. The one dimension which you just briefly touched upon is especially from a European perspective, the heterogeneity of state policies. So if you look at, for example, California, they are very effective at moving people from their own state unemployment insurance to the federal disability insurance, while some of the states that are actually in your eastern heartland are not even making full use of federal resources available to them through Medicaid, through TANF, et cetera. So the one thing I'm wondering is, would those place-based policies that you advocate for, would they actually find the, would there be some ownership in those states that really need them? Any further questions? No? Okay, so first of all, let me thank Diego first. His usual superb discussion. And I don't think we differ very much, and I just want to call it one major thing, since there are maybe some macroeconomists in this room, that the opinion of infrastructure, right, is very different among macroeconomists and among economists who actually study infrastructure. So most of the time that one talks to macroeconomists, it is taking a sort of an unbelievable, unshakable thing that we need more investment in infrastructure, whether the U.S. or somewhere else. I have never met an infrastructure economist who actually believes that, right? Now, they may very well believe that we certainly need targeted maintenance expenditures, where the returns are very, very high. They may certainly believe that we need policies to encourage better use of that infrastructure, but like this article of faith that we need, what we need was more infrastructure just as clearly as we need more education, that is not something that's typically shared. I think Diego's comments brought that up. Secondly, on Diego's final point on education, two points. One of the reasons is why we didn't emphasize education more relative to employment subsidies. The first thing is that I think for most of my life as an economist, which is about 30 years, the routine economist's response to every economic dislocation is to invest more in education, right? That does not make that wrong, right? It's absolutely right that we have good evidence on pre-K programs, but telling an unemployed coal miner in Kentucky that like, you got nothing for him, but his granddaughter will get a great pre-K program is not the most satisfying response politically that one could possibly do. Now, so in some sense, I was reacting to agreeing with education, but thinking that we wanted to say more. And I actually think that the case for, and this relates to the point about the working poor, the view that there is a flat-out externality associated with joblessness, which comes fiscally, and maybe a more arguable one that comes through other forms of transfers, or maybe psychic internalities, if you believe in that sort of thing. That seems to me to call for an off-setting Pagoovian subsidy. I mean, that's what Pagoo said in 1912, and I think he was right then, and I think he's still right. So that's why we are arguing that. And if we're supplemented to that a little bit, the past 40 years of research on adult retraining programs, primarily in the U.S., is vast and dismal, right? I'm not saying that every single one has been a failure, but overwhelmingly, this has been a retraining 50-year-old guys for new jobs is hard. Now, Diego's comment is quite reasonable. Is it possible that anyone can retrain the 50-year-old guys or use the 50-year-old guys? And I'm not sure what the answer is on that. We sort of know what less skilled people do in cities. They work in the vast service sector. It's less clear what less skilled people are gonna be doing in areas that are far away from more skilled people. And whereas in the 1960s, a manufacturer would move capital to a low-wage place like Mississippi to take advantage of cheap labor, no fancy Silicon Valley tech guy is gonna move to West Virginia because he can get cheap haircuts there, right? So you don't have the same sort of relocation that comes about this. And by the way, this issue also comes up in terms of the future of like sub-Saharan Africa, right? Even if transportation costs continue to fall and accessibility continues to fall, if manufacturing mechanizes sufficiently quickly that the East Asians are basically able to produce at a cheaper rate, even with higher labor costs, then there aren't gonna be manufacturing jobs in Africa either, and that begs the question for what will those jobs take? But what will the jobs look like, if anything? I still think that as economists, if we see a market failure, if we see a Perguvian externality, we can call for a subsidy even if we're unsure, even if we're unsure about what the behavioral response will be. So a couple of other points. So the comment about the working poor I wanna sort of emphasize two things. We saw there as being, we actually wanted, I mean, we obviously want the working poor to be working less poor, but we actually think that getting people from non-employed into the working poor is actually a really good thing from most of the data that we have. What we don't have is we did do an interaction with not working rather than interaction with any other thing about the labor market. That was fairly arbitrary. I mean, the logic of the model tells you that any place for whatever reason you have a higher labor supply response on the extensive margin, you wanna have a higher employment subsidy. So we wanna identify those more clearly. The Tomasha's point about the Great Recession having some role in it, surely it's some. Tomasha, I mean, surely that's some role, but the two pieces of evidence to go against her, the timing doesn't quite line up well. And secondly, there is all this evidence that like renters have had exactly the same decline in mobility that owners or past owners have or permanent renters haven't. So it can't all be house lock. I mean, house lock is surely some dimension for this chlorosis. Mark's comment about sort of dynamism of, I thought you were going somewhere else on this, but you're about the movement of workers from firm to firm. Another similar decline in dynamism is the declining level of entrepreneurship as well in the US. And I haven't fully wrapped my head around how these things fit together and whether or not as it's best seen as a third force which is leading to all three of those things. But they are a fascinating change and an important change in the US. I will say just one thing about looking for work and changing things. Historically, you think about Americans as moving first and then finding jobs later. And that was historic difference between the US and the EU where it's like no German is gonna move from one city to the other until his job is lined up in the new city. You don't just like show up in Stuttgart and say hire me. But you did for like decades, for centuries you showed up in Minneapolis and said hire me in the US. Now what could be happening actually is that the rise of various job search platforms mean that you no longer do that. So it may lead to less job search like migration but that's a speculation. One last point and then I'll just end. Will states own the policies and the heterogeneity and state implementation stuff? I don't know and it's not quite clear that the political path forward which says you're gonna allow some degree of state flexibility and employment subsidies that would be a relatively easy path forward which is sort of similar to the wet path forward that we have with Medicare and Medicaid. I would worry about that that in fact if you gave states these options that it would be implemented in a completely wrong manner. So I'm very skeptical about the ability of the states to do this right on their own. I favor state heterogeneity but not necessarily state control. So let me just end there and again, thank Luke for including me on the program. See you. Okay, thank you very much. So we go for second coffee break. Thank you.