Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. ======================================================================== # Living in a Ghost Town: The Geography of Depopulation and Aging Authors: Elisa Giannone, Yuhei Miyauchi, Nuno Paixao, Xinle Pang, Yuta Suzuki Discussant: None Video: https://www.youtube.com/watch?v=w6ObGDwelyg&t=21030s ## Talk (05:50:30 – 06:24:44) [05:50:30] >> Great. So, we're just about on um time for the next paper and I know Larry has a question, but I also know Ui needs to leave in exactly 35 minutes. So, perhaps we can start your presentation and then Larry, if you want, you can just type up [05:50:44] your question in the chat. That way, we um we make sure UA leaves on time. Yeah. [05:51:06] Okay. So, last paper of the day is living in a ghost town uh geography of depopulation and aging presented by u. [05:51:14] >> Yeah. Uh thank you. And uh as a as a last uh presenter, I I uh especially want to thank the organizers. uh it I've been thinking about this issue for from the perspective of spatial dimension but [05:51:28] it's it's been great that you know a lot of the different aspects has been gathered and I learned uh personally a lot um in this conference uh and I also have to excuse that I I so I recently uh [05:51:43] tried to contribute to slowing the you know population decline in Japan as a Japanese citizen and as a consequence I have to pick up my kid from the daycare [05:51:55] uh a little bit uh earlier than uh the official end of the conference. So I I'm going to uh uh uh have to cut the Q&A a little bit shorter. I apologize for [05:52:07] that. All right. So um but I guess a good thing is that I'm going to build on uh many papers today especially uh uh the the last one to probably skip some of the slides uh to make it on time. So [05:52:22] uh this is a paper with Elisa Nol Shinan Utah um I'm going to skip that everyone knows that the world is depopulating. So um and uh so we're going to be interested in uh just like the uh the [05:52:35] previous paper the spatial uh dimension of uh the population and aging. Uh and there's uh extremely limited evidence uh systematic evidence uh you know which is [05:52:49] uh probably changing uh based on the papers that has been presented in this conference. uh but you know part of the reason is because um uh it's you know it's this process is uh happening uh [05:53:02] slowly uh uh only gradually like there are not so many countries which are actually under population decline and uh you know there are limited data uh to analyze even you know uh without talking [05:53:15] about the aggregate terms but also on on the regional terms. So I'm going to be focusing on Japan which is advantageous for two reasons. um one uh because it's happening rapidly now and second uh it's [05:53:29] a datari environment as as I'm going to show you and uh yes anecdotally we know that rural areas are depopulating agent much faster uh that you know we want to understand uh what is driving these [05:53:43] results and also what's happening in in in these uh depopulated regions uh we're going to also highlight this uh fiscal dimension of of of this issue which is that it looks like the uh sustaining the [05:53:55] public uh service cost of running this depopulated region is increasingly uh becoming more costly because of the scale effects of the public service uh cost. Uh so that's uh another thing I'm [05:54:08] going to be highlighting. So what I'm going to uh do in this paper uh is to document the spatial pattern of depopulation aging within Japan. uh basically I'm going to document that uh there's a widening regional divergence [05:54:22] in the population in aging and uh driven by both natural change so birth and death which is obviously the primary driver at the country level uh as well as the social change uh which is the [05:54:34] migration uh across regions in the context of Japan there's not much uh international uh immigrants at least so far uh so that's uh these two are the main margins and uh we're going to also showing that [05:54:49] uh as I said uh it's going to change the local services amenities uh uh as well as the per public service cost and then we're going to be developing a quantitative life cycle dynamic special [05:55:02] gener model which is uh in the spirit at least not not that spirit that structure is kind of similar uh uh with Tim's uh earlier presentation uh which except sorry maybe I shouldn't say that it's [05:55:16] something that we are going to be highlighting more is this uh end indulgence migration was already there. [05:55:23] But this uh this idea that birth locations are also endogenous because when uh young people move uh they're going to be generating kids there. Uh and so the we're not going to endize the [05:55:37] fertility rate. Uh uh but where people are born is going to be endogenous. And uh and this process is also amplified uh because of the lo uh the the spill or effects on productivity amenity as uh [05:55:51] team also mentioned as well as the uh and also there's a public service cost which is going to have a fiscal implication uh that we're going to be highlighting and uh we're going to use this model to predict what's going to happen in the future and the basic [05:56:06] message is that this divergence of uh the population aging is going to uh persist uh because of this uh feedback loop between indulgence migration and birth location. People move to Tokyo and they're going to have kids in Tokyo [05:56:20] there. The kids are going to be stuck there and uh they're going to have kids there. uh and as population increasing in Tokyo more the those places are become uh increasingly more attractive places and hence uh you know this [05:56:35] process is not doesn't seem to at least in in a model doesn't seem to stop uh in near future and we are going to be also drawing some policy implications uh in particular we're going to be simulating the [05:56:48] placebased transfers uh uh uh uh the in the spirit of the uh policy analysis that uh Tim did. Uh the we're going to be high that the key message that we're [05:57:00] going to be highlighting here is that uh there's a uh uh important uh equity efficiency trade-off in a sense that uh you might want to redistribute uh resources toward uh rural areas uh [05:57:13] because you know to protect those uh those areas but that's going to be uh costly from aggregate efficiency point of view. For one because the productivity is lower uh and for two uh [05:57:26] because the fiscal cost uh public service cost is going to be higher as uh because of this uh scale effects of the uh service cost. All right. So I'm going to skip the literature. I I know there [05:57:38] are some uh comments on that chat, but I I think I'm >> for now there's still comments about the the previous paper, but we pass five minutes, so feel free to jump in anyone if um >> there's any questions. Yeah. [05:57:52] >> Yeah, >> I'll monitor for you. Yeah. [05:57:54] >> Yeah. Thank you. Uh all right. So, I'm uh going to skip this. It's uh Japan is the oldest country in the world. uh just a couple days ago uh the number of [05:58:07] people over 100 years old past 100 uh thousand uh people which uh is amazing. [05:58:17] Um and so again like we're going to be highlighting more on the regional dimensions. So that's the first cut of the of the data. Um uh as as you probably know is Tokyo or Sakanaga are [05:58:31] the major uh metropolitan areas in in Japan like where you see like the concentration of population and the uh f the aging is basically uh uh negatively correlated. So those uh places are where [05:58:45] uh younger uh population tend to reside. Uh this is a scatter plot. Same scatter plot. Uh elder share on the uh on the horizontal axis and population density on the uh vertical axis. I mean some [05:58:58] municipalities are reaching like over uh uh 50% of uh elderly uh fraction of the elderly again like 6 65 now in Japan is actually like you know many people don't [05:59:12] think you know they are old at this point but uh anyways that's the definition working definition that we we're running now uh and uh and as you probably know if [05:59:24] you have visited Japan like it's uh very different between like urban centers and uh rural areas. Urban centers are particularly attractive probably like in particular for young people with you [05:59:37] know lots of uh amenities and rural areas are much more sparse uh with that uh lacks not only those types of amenity but also like some kind of public infrastructure as well. All right. So [05:59:50] let me uh show uh two types of uh descriptive facts. uh one is about the pattern of the a little bit more of the patterns of the depopulation and aging across regions and the second is the [06:00:04] local uh economic consequences of this depopulation. So I'm not going to spend uh much time on this uh this uh slide but the what's nice about Japan is it's a uh it's it's quite datarich uh it has [06:00:19] lots of uh data on population changes, migration, uh income, employment. I have to say income is a little bit trickier uh because uh we we don't have it at the municipality level only at the [06:00:31] prefecture level. So uh that's uh that's the nature of the publicly available data. Uh so so uh that's that's a little bit of a limitation but uh uh there are other uh interesting variables like land [06:00:44] prices housing and also uh something that we're going to be emphasizing is amenity. Uh we're going to be compiling uh lots of different types of amenities. [06:00:54] uh uh uh uh uh you know such as uh local services, restaurants and some of the you know private ones and public ones and we're going to show uh the impacts of those amenities on the local population change. [06:01:08] So the first uh uh set of facts is about the divergence of uh the population and aging. So here we are uh uh stratifying municipalities based on the uh the [06:01:21] initial initial period meaning 1980s uh fraction of the elderies. Uh this uh lowest is the uh the most uh sorry this this this is the uh youngest [06:01:34] municipality and this is the uh oldest municipality at the point of 1980 and again you can roughly translate that to like the population density as well. [06:01:44] they are extremely uh correlated as I showed. Uh so the point is that uh uh the the population density is increasing in these younger uh places and population is decreasing in these uh [06:01:58] older places and it seems to be you know the diversion seems to be you know rather than converging uh seems to be accelerating and the fraction of the elderly uh here we are normalizing this [06:02:11] uh at the point of uh sorry the at the point of 1980 uh Uh uh let me just I think I I I I showed the level of how these lines differ at the point of 9080 [06:02:25] in a couple slides. uh but the point uh that I want to make uh for now here is that the old the older uh uh municipalities are getting uh even uh more older uh and younger municipalities [06:02:40] are actually getting uh they obviously like the country itself is uh increasing the fraction of the elderly but the rate at which the fraction of the elderly increases is much faster in already [06:02:53] older regions rather than uh already uh Anga region and uh and uh so here what we do is a simple accounting decomposition of uh [06:03:06] taking the uh oldest municipality and youngest municipalities in uh uh and uh look at the population growth. [06:03:16] uh so these uh solid lines are what I basically showed uh in the previous figure and I'm doing a very simple uh accounting exercise which is uh what would the population change uh would [06:03:29] look like if we shut down the migration so assume that there's no migration across uh regions within Japan and another experiment that we are doing is to fix the spatial distribution of uh [06:03:41] birth at the point of 9080 so that uh This this is kind of our way of uh disentangling the contribution of the uh you know the differences in the birth [06:03:53] rates um and differences of the migration uh in terms of contributing to the the divergence of the population. So the punch line here is that the migration uh is uh is driving uh uh more [06:04:07] actions. So if there were no migration these older uh municipalities are actually like I mean they are still losing population but it's uh at a much slower rate that uh you may think that [06:04:20] the birth is not contributing much but this is a little bit of by construction of this exercise because uh because I we are fixing the birth distribution at the point of 9080 so you you should expect that there's no diversions at this point [06:04:35] uh but as that years goes on um because the reproductive age population is in a data moving uh toward Tokyo uh relatively more uh uh this uh is going [06:04:48] to this margin is contributing more uh over time and as I'm going to show you uh through uh uh in the using the quantitative model this this margin is is in particular is going to be more [06:05:01] important as we think about the future implications uh and this is uh the same figures uh or same experiment in for the elderly share. So uh again this is the level [06:05:15] that I promised uh in the youngest municipality started out around like 6% of or 7% of uh fraction of the elderly uh and uh older ones uh started around [06:05:29] like 16% uh and now it's like 35%. this uh uh that so they are kind of moving you know the difference is actually increasing um but this uh if we [06:05:42] shut down migration or the changes in birth populations uh it's we see more convergence uh as uh as as uh as we as the time goes on [06:05:56] right so that's the first set of uh facts and the second set set of facts uh is about the local consequences of this process this. So here we are running a regression uh where left hand side outcome is the changes in like various [06:06:10] economic economic activity proxies uh such as amenities uh uh per capita tax payment and you know things like that and the uh we're projecting on the changes in [06:06:23] working age population uh and also the fraction of the aging uh the fraction of the population that sorry the changes in the fraction of uh the elderly [06:06:37] and uh so this echoes to uh uh to our earlier presentation uh today uh I mean the design is kind of similar but we are going to be using a different uh uh variation uh which is coming from the [06:06:51] push and pull IV uh so the basic idea is that uh we have a uh sort of migration flows uh at the point of baseline period and we are going to be asking uh suppose there is a population growth [06:07:05] uh in a municipality which tend to send uh people uh to certain municipalities. [06:07:12] Um so like the recipient municipality is kind of uh you know uh experiencing an unexpected flow of working age population not because of the local conditions uh but because uh the uh the [06:07:26] uh the the origin uh municipality has expanded. So you can do the same thing for the uh pool IV uh where like you know the surrounding uh municipalities [06:07:39] uh become more attractive uh instead of the the location itself become uh worse off or something like that and we have a variation for working age population and aging uh because the uh migration flows are age specific and [06:07:54] in it I I think we have a in in the appendix that they vary a lot as you can imagine. [06:08:01] Uh yes. So let me jump uh into the results uh quickly. So I'm going to first show the set of uh uh results on the amenity proxy. So these are five uh [06:08:12] amenity proxies. We use a PCA uh index for uh for for each category. So it's it's a combination of different uh uh different uh proxies which we made it to [06:08:24] a single index uh standardized to uh to to be standard error sorry standard deviation one. Um so the basic uh message here is that the change in working age population tend to increase [06:08:38] various types of amenities. Some of them are more like you can you might want to think about it like as more like a private. Some of them are uh more like a environment transport like something like roads more like a public uh uh many [06:08:53] regardless of these classifications they tend to uh increase the share of elderly uh it's it's a little bit mixed uh it's not as uh strong as the working age population. uh but you see a kind of [06:09:06] intuitive pattern that for example uh elderly service is marginally significant obviously if there's a uh so it's it's like a you know senior centers and so on if there's a higher share of [06:09:18] population is becoming more uh senior then you know that kind of types of amenity that are provided in those regions are directed uh toward that type we're not going to be modeling like we're going to be focusing more on this [06:09:32] uh part in in this uh uh in in in the quantitative model. But this is kind of an interesting feature that we find. [06:09:40] And uh so the second row uh that we are showing here is uh is the let me start with this one. This is the local government spending per capita. Uh so a [06:09:52] lot of them uh is uh is is going toward uh providing local public services. And what you find here is that as the working age population increases the per capita uh cost of government spending [06:10:06] goes down. And we interpret uh this result as indicating that there's a scale uh effects of providing uh local service cost as if there's a larger population it's easier I think there's a [06:10:20] comment by Ron along these lines. it's easier to provide public services because just having one let's you know one police station can serve more population as opposed to in a sparse areas uh you need to create multiple of [06:10:34] them uh to cover the entire areas and so on and here we find uh this is uh uh on the other hand we don't find any effects on the personal income tax this is not the labor income this is a combination [06:10:47] of like uh all that kind of the the tax base uh including you know capital tax and so on. uh so that's the limitation that I said about uh about this data uh uh that we don't have a very specially [06:11:02] dis disagregated data of wages so that we don't uh find uh much effect uh here but I combine these two one implication that is kind of important to keep in [06:11:14] mind is that the populating areas are running uh deficit fiscal deficit because they are uh they're becoming increasingly more costly to sustain even though their tax revenue is not changing. So that's the one margin that [06:11:29] we're going to be highlighting uh in in the thinking about the future uh aggregate implications and we also find uh find uh find some responses of the housing market uh uh we don't do uh we [06:11:44] are we are asked to to do the robustness about the uh heterogeneous elasticity of housing which uh I completely agree that's kind of makes sense uh but we we in our baseline we don't do uh uh uh [06:11:58] what uh Tim and James did uh but as as as they were saying it's it's kind of complimentary uh forces like both of them them can coexist and they uh kind [06:12:11] of uh re reinforce uh each other. All right. So um so that's the uh basic uh data pattern. Um what we want to uh [06:12:23] do in the in the rest of the paper is to build a general equilibrium uh model uh spatial model uh uh with some life cycle aspect to uh think about the future implications and also some policy [06:12:37] implications. the structure of the the model is is a similar again like similar I I'll qualify uh the part that is not similar uh but the idea is like is is similar [06:12:50] where people are making a discrete choice like uh and um as every period uh they determine location based on uh wages uh housing cost uh we're going to [06:13:02] be also modeling uh some some uh the uh pensions and uh public service uh cost financing. Uh but it's a kind of dynamic uh discrete choice decisions. [06:13:15] Um so there are uh in many number of locations uh we're going to be indexing age by a uh and um yeah income uh comes from [06:13:27] labor income, land ownership income and pension. uh land income depends on uh sorry labor income depends on local age productivity which is subject to the productivity spillovers. [06:13:40] Uh every period uh people consume freely traded goods housing and uh there's also amenity uh that is uh associated with uh with living there and we're going to be introducing [06:13:54] amenity spillovers based on the findings that that that that I mentioned. [06:14:00] Um yeah so productivity and amenities are uh have externalities um labor income taxes finance pensions and local public finance uh services um public [06:14:12] service cost also has a uh scale effects. So it captures the feature that as the region becomes more depopulated per capita cost of uh government spending increases which I showed uh [06:14:26] earlier. So I'm not going to probably show you uh lots of equations. Um yeah pref like this flow utility uh budget constraint. [06:14:39] We're going to be somewhat apologetically uh assume handto-mouth uh agents uh to focus on the uh location choice margin uh rather than consumption saving margin that uh probably uh has [06:14:54] been highlighted in the more aggregate uh type of studies. uh and there's uh externalities uh and the location choice is this uh is uh so every period they're going to be [06:15:07] maximizing the choosing the location that maximizes the continuation value uh subject to a migration costing utility and also idiosyncratic shock. So you know people have idiosyncratic reasons. [06:15:19] Some people have received some attachment to Tokyo and they're going to move move to Tokyo. But some uh person like has you know maybe relatives in uh certain regions and they're going to move to different regions. [06:15:33] Um so it's a forward uh looking problem. So you can write a uh value function uh which where the value is depends on the flow utility and the future uh [06:15:44] discounted uh value after you choose the optimal location and uh uh with together with the functional form assumptions about the iid uh idiosyncratic uh shocks for [06:15:59] migration uh uh cost uh we have a kind of logic expression for uh for the migration flows uh which is uh also age specific uh and it also depends on time [06:16:12] because as you know because the local economic conditions are going to change these uh migration shares. [06:16:21] Um this is uh just showing the flow how the population transit uh changes and uh without you know explaining the every single letters uh what what we mean here [06:16:34] is sorry this is actually important. So like this this first one where like the newborns are going to be generated by the uh the the stock of population at [06:16:46] that point in time in in that location uh by the reproductive uh uh age population. So that's has uh some again feedback loop where if once younger people start to live in a certain [06:16:59] location newborns are going to be uh generated there which uh adds a extra uh source of concentration force and uh uh and the other drivers of the local [06:17:12] population change is survival rate and the migration. Uh so this is this is like a demographic balancing equation so-called in a demography literature while we are basically microfunding these migration decisions uh in a [06:17:26] microfunded model. All right. So housing sector this I'm happy to skip because uh Tim did a great job in explaining these things. So it's we're going to be taking a constant [06:17:38] elasticity uh version. the government is going to run a a a balanced budget. So uh it's uh the uh sorry capati is uh I probably didn't explain this is the [06:17:52] labor uh income tax uh uniform across location uh and that income is going to finance uh all the pension and uh the [06:18:04] public service cost and this public service cost we're going to be modeling basically as a fixed cost of sustaining the region. So it's a mechanical function of the population size. But the key is that the population this is a [06:18:19] decreasing function of population. So per capita cost is going to go down as the population increases. So depopulating regions are becoming more and more costly to sustain. It's it's going to add a fiscal pressure [06:18:34] uh uh in the budget constraint. All right. So um I we take uh the model to the data um uh we do reasonable job uh [06:18:46] we do some model validation as well. Um uh if you're interested please take a look at the paper. So I'm going to sort of uh show you a couple of um I spend [06:18:59] couple of minutes uh for showing the future uh prediction. So sorry future prediction and some policy experiment. [06:19:08] So here uh what we're doing is uh take the model and then run uh the simulation uh uh in the uh forward up to like about 200 years. What you see here is that in [06:19:20] our baseline model uh Tokyo's uh population keeps kind of increasing uh rather than kind of converging uh with uh with the oldest uh most sparse uh prefectures [06:19:34] and uh this is uh our basic message is that this is uh as a consequence of the combination of the internal migration and the endogenous birth location. So if you shut down uh each one of them then [06:19:48] you see basically it's going to stabilize at some point uh uh and it's actually not relatively closed level as we find uh in in in you know at the point of 2015. [06:20:01] So this uh feedback loop between fertility and the migration is uh is uh is driving this divergence of uh population uh concentration. [06:20:15] Uh same uh uh same patterns for the fraction of the elderly in the baseline. [06:20:22] Tokyo's uh elderly uh population fraction of the elderly population stays relatively low while the older uh prefixes basically like gets gets older and older uh without a break again if we [06:20:37] shut down migration or birth uh distribution changes then we see more uh convergence at some point uh and I would say that it's also like behind this it's also important that [06:20:49] there's a a long ration spillovers uh Uh this this this interaction between this fertility and migration is a amplified through migration through through the spillovers because once young people [06:21:03] concentrate in a lo certain location that region becomes increasingly more attractive location which changes the migration patterns we have some kind of actual to illustrate that point as well. [06:21:14] Um at the same time um so Tokyo's uh uh uh the consumption equivalent flow utility is becoming uh so this is what I said uh it's becoming more and more [06:21:27] increasingly more and more uh attractive place relative to the uh oldest uh regions and again that's uh that so this is one mechanism in which this feedback loop is important but this also calls [06:21:40] for a policy question like what should we do with this uh older prefixure which is you know getting worse off uh uh you know lagging behind. [06:21:51] So uh and uh so let me also say this. But so this calls for some kind of like equity concerns but there's also efficiency concerns because once the [06:22:03] population becomes uh con so population concentration in Tokyo is is uh may not be good for equality reasons but it's good for efficiency reasons because uh aggregate you know those are the places [06:22:16] which are more productive. So agreed income per capita increases more especially compared to shalinda migration or part distribution. Fiscal spending per capita goes uh down because uh people are not stuck in a region [06:22:31] which is very costly to uh provide public service cost. So this calls for uh efficiency equity trade-off that we analyze uh in in the policy kind of. So we do some simulation of the placebased [06:22:46] transfers. We don't do uh the uh the the opinion on policy in uh in our paper uh in this paper but uh I have a separate paper paper that work on on that and uh [06:22:59] in in a uh future work I I think it's a very interesting uh question to think about you know building on what Tim and uh James did. So I'm not going to talk about this. This is basically highlighting the efficiency equity [06:23:12] trade-off. more transfer leads to more redistribution but that's going to cost aggregate efficiency loss. So I'm going to uh stop here. So that I I would say [06:23:24] like a central uh and I I kind of want to emphasize that uh you know after working on this topic or this paper like this is a hugely important topic that like there they're very sparse uh papers [06:23:39] and I this is probably true for many papers in the population and aging in general but I think uh this regional dimension is also very very important that you know there there should be a lot of research uh undertaken And in particular, I'm uh particularly [06:23:54] personally interested in optional place policy uh uh thinking about the dynamics of spatial transfers uh uh as I've done in a more general uh environment. Uh another question is infrastructure [06:24:09] policy that came up in an earlier conversation and uh some like policy it you know one political issue is like even if it's it's hard for the government to commit to a policy to kill these depopulated regions because of the [06:24:23] redistributional concerns. So I think commitment is also an important issue uh to to think about for kind of actionable policy recommendations. So, I'm going to stop here. Uh, and sorry, I think I I myself went went over time and I [06:24:38] probably have to leave in couple minutes, but uh uh happy to take a quick question. ## Q&A (06:24:44 – 06:32:06) [06:24:44] >> Thanks so much for a great presentation and I totally agree with the spirit of what you said. This is a really interesting work and we need more work on this, but I'm glad this conference managed to feature two papers on on on [06:24:57] this uh this really interesting topic. So, so I guess you should probably you time I know you have to go in about a minute. [06:25:05] >> Yeah, I I please send the email if you have any questions or comments. I again apologize. Uh but uh >> I think perhaps that's what we we could say that um yeah, anybody that has comments for U should just feel free to [06:25:19] email him directly and you know, thanks so much for um for your presentation and for contributing to repopulating Japan. [06:25:28] We'll let you we'll we'll let you go and pick up your your kid. Um so, um All right. So, yeah. Yeah. So, I guess that's a bit um it's too bad we didn't [06:25:40] get uh to uh uh talk to you much, but but perhaps Jim and uh uh Tim and Tim and James, maybe you can I had a question actually about the relation to to your paper in that it seemed like the [06:25:54] there's a lot more richness about sort of the age structure and so on in in that model, but they're abstracting away from the mechanism that you guys highlighted. And so in a way can we think of these two papers as [06:26:06] complimentary um in that if if um you know if they added the nyps of nomaticities you have in your paper they'd get some amplification perhaps kind of bigger or you know much larger effects. [06:26:18] >> I think that's totally right. Yeah. >> Yeah. Totally. And I think they um you know that we kind of tried to we felt like we had a decent amount to say just about the level of the population and we [06:26:33] felt like it was kind of easier to write a paper just kind of be clear on that portion and I think yeah they're totally complimentary and they would totally amplify each other um in particular because it's the young people who are doing all the migration and their [06:26:46] empirical evidence is showing that like the young people are also the ones who are influencing the quality of the amenities, Right. Yeah, that that makes a lot of sense. Um >> I think also like so quantitatively just [06:27:00] to give them them another compliment. Um uh you know our our mechanism like Tim illustrated is very important in the long run. It has some interesting implications on the on the transition path and he showed you the comparison of like 2000 to 2020 where we can actually [06:27:15] compare what has happened in Japan versus what our model predicts. But there's a big chunk of that recent data that's not explained by our mechanism. [06:27:23] And so I think especially with the importance of the the transition dynamics in their model and what's going on with aging and young people moving um that that's probably also quantitatively important for for explaining uh the recent data and and what's going to [06:27:36] happen in the recent future or the near future. [06:27:41] Can I say I think we had four maybe five papers all of which concluded that uh slower population growth and population aging uh would raise per capita income. Is [06:27:56] that right? But I can't quite remember Riner what you concluded in your paper but certainly uh yeah and four others on education on the baby boom [06:28:10] and people presented it this is going to you know disagree with everyone but in fact I think it's kind of the consensus in the literature now that so far as population per capita income goes [06:28:23] well I think there are cases in which it will lead to some decline, but for the most part, it seems to lead to at least mild increases and in some cases in some people's analyses larger increases. And [06:28:35] I think that's interesting. >> Yeah. I mean, I think one question is about the magnitude of this effect, right? I think Keelin's presentation suggested the magnitude was so big that [06:28:46] perhaps GDP may not decline at all. And that's uh you know that the the the qual like qualitatively that it's going this direction I think is uh is perhaps not as hard [06:29:00] >> to believe as as just the magnitude the elasticity being more than one you know >> and I was going to say you know per capita income is not a good measure to use as we know from Samuelson 1958 where [06:29:14] you have two societies with exactly the same per capita income and totally different lifetime time consumption and lifetime utility. [06:29:23] >> Uh because it also depends on how you arrange consumption across the life cycle. And in rich countries, we arrange it so that the young don't consume much and the elderly consume a lot. And that sort of it goes the opposite direction [06:29:37] that uh Samuelson optimization would uh would suggest. But in any event, these effects are so big that they override anything if we believe them that uh uh could [06:29:51] result from just taking counting every person as one in the denominator of the economic outcome. [06:29:59] Anyway, it I thought it was a great meeting >> and you know I think Ron just as you said that I mean I mean there are a lot of people who would have asked just if asked the question what do you think about declining population growth would [06:30:13] would run reflexively to the Chad Jones style you know we're going to have fewer ideas the innovation rate's going to collapse it's going to be a disaster so I do think that this is an issue where they're you know bringing the the this body of empirical and and simulation [06:30:27] work to bear and talking more about this is is an important thing for us to have on the agenda going forward very much. [06:30:33] And you know, it speaks of course to the to the really insightful activity on your your and Will's part in suggesting that we gather this convenient, but I think we have to do more of this. [06:30:44] >> Well, I hope we can. >> Yeah, definitely. [06:30:51] >> Great. Agree. So, I guess we're wrapping up. So, um I was just going to say uh we talked about a lot of things today. Uh so thanks to everybody for sticking around for until the end. We we had [06:31:03] papers about the implications of population aging for immigration policy you know the college market GDP growth investment in labor saving technology innovation rates of return global imbalances public finance you know [06:31:17] pension systems and the size distribution of cities. So, so it was we covered a lot of topics, but I think we also saw a lot of, you know, commonalities across, you know, u uh uh ideas and research methodologies. So, I [06:31:31] I learned a lot and I I thought it's great um that we managed to to do this event. [06:31:36] >> Yeah. >> Great. Thank you everyone. Thank you to the three organizers for doing this and and seriously I do think we should try to find another way to do some more of this because you know virtual meetings like this are are pretty easy to put together and well I should easy for me [06:31:51] to say uh the three of you had to do it but you know easier than than than other types of things and I do think there's a great deal of interest globally in this set of issues and we can pull the research community together. So >> well let's keep it up. Sounds good.