Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. ======================================================================== # Population Dynamics and Economic Growth: From Demographic Dividend to Demographic Drag? Authors: Rainer Kotschy, David E. Bloom Discussant: None Video: https://www.youtube.com/watch?v=w6ObGDwelyg&t=15621s ## Talk (04:20:21 – 04:51:16) [04:20:21] great um presentation and discussion. So our uh next paper is uh population dynamics and economic growth uh from demographic dividend to [04:20:33] demographic drag and I guess Rainer U will be presenting. [04:20:39] >> Yes. Can you see my slides? >> Yes. [04:20:44] >> And I hope you can also hear me loud and clear. Thank you so much for the introduction. Thank you so much for the moderation of this paper. [04:20:52] Uh my name is Reno Cotchi. Uh it is my pleasure to present this uh paper which is recent work on how the conceptualization of aging may shape our thinking or understanding of how [04:21:06] population aging might influence economic growth. [04:21:10] So many societies around the world are aging because of increased longevity, lower fertility, and also the progression of large-sized cohorts into the older ages. And you can see this aging trend for instance when inspecting [04:21:23] median ages which in OCD countries are projected to be about twice as high in uh 2050 as they were in the 50s. [04:21:35] And as a result of this aging trend there will be relatively fewer younger and relatively more older people in the years ahead. [04:21:45] A key question in this context is what are the effects of changes in population population age structure on economic growth. And so far there's a consensus in the literature especially in the social demographic literature that [04:22:00] falling youth and growing working age shares offer the opportunity for demographic dividend in terms of fast economic growth. [04:22:08] The idea behind this concept is very simple. An increase in the working age population raises the labor potential which in turn promotes savings, innovation and overall productivity. And there are [04:22:21] quite a number of mechanisms also including directed technical changes we have seen earlier today. However, looking ahead, working age shares will contract rather than expand and old age ratios will increase. [04:22:35] So consequently, resources that otherwise could be used for production will be bound for old age. And this raised the question of whether the demographic dividend that has propelled our living standards in the past. So over the last 50 to 80 [04:22:50] years will turn into a demographic drag that might slow the pace of economic growth in the years ahead. [04:22:58] In this paper, we examine the role of population age structure in economic growth and also study how sensitive the estimated effects of population aging on economic growth are to different [04:23:13] conceptualizations of population aging. And I will be more explicit about what I mean by that in a moment. [04:23:19] Specifically, we answer two research questions. So first will contractions in working age shares slow the pace of economic growth in the years ahead and second to what extent do the estimated effects depend on the concept of aging we employ. In answering these questions [04:23:34] we will add to the debate about the size of the potential effects of population aging on economic growth and add an additional perspective or an additional nuance on how the measurement of population aging might influence our [04:23:48] think methodologically. we proceed in two steps. The first step is quite straightforward. So we estimate a simple um empirical growth model that we fit to [04:24:01] cross country panel data in 5-year intervals through 1950 to 2015 to get some estimates of the potential effects of population aging on economic growth in the spirit of the demographic dividend literature. [04:24:15] In the second step, we then input demographic projections into this simple fitted model to gauge the consequences of population aging for growth uh in 2020 to 2050. [04:24:27] A key novelty of our approach is a comparison between retrospective and prospective measures of population age structure. I'm going to explain them in a moment. In contrasting our results for [04:24:40] these two different measures, we calculate back of the envelope estimates of a potential demographic dividend or potential demographic drag under fixed and evolving population health. [04:24:53] So what do these measures mean? So we distinguish between retrospective and prospective measures of age structure. [04:24:58] Retrospective or chronological measures of age structure rely on people's chronological age. So you can think for example of the conventional old age ratio which is defined as the ratio of [04:25:12] people 65 and older relative to the working age population. [04:25:16] These ratios are very simple to estimate. However the issue with these measures is that by relying solely on chronological age they ignore the age patterns of health that is age patterns in mortality age patterns morbidity and [04:25:30] also age patterns and functional capacity more generally. [04:25:35] And because of that, they're not necessarily well suited for comparison across countries and over time. [04:25:42] So one potential solution to address this issue is to use a prospective or functional measure of population age structure that allows entry into old age to evolve with changing age patterns of [04:25:55] health. And one way to do that is to use a prospective oage threshold that is defined say as the age at which remaining life expectancy falls below 15 years. This is an approach that has been [04:26:08] pioneered by Norman Ryder and further refined and pushed by Warren Sanders and Sak Sherb in a couple of contributions in recent years. [04:26:18] The idea is that this prospect of old age threshold or POT increases as people live longer and healthier implying that we have a supposedly more realistic representation of people's actual labor [04:26:31] potential. The prospective old age threshold significantly correlates with measures of population health at the aggregate level. So we show that it is uh [04:26:46] conditional on country and time fixed effect significantly correlated with mortality in terms of life expectancy at age 65 with people's physical capacity in terms of their maximum group strength their cognitive [04:27:00] capacity in terms of immediate work recall and negatively correlated with mobility in terms of average health deficits per person age 50 to 79. [04:27:11] And what this means is that we have a measure that is actually reacting to change to changes in population health if these different quantities change over time. [04:27:25] Retrospective and prospective measures of population age structure can produce qualitatively and quantitatively different demographic trends. Let's begin with the retrospective old age ratio for instance. So this ratio is defined as the population 65 and older [04:27:40] relative to the population 20 to 64. And as you can see it's almost monotonically increasing much more in OCD countries than in non OCD countries will be maybe two to three times as high by 2050 as it [04:27:53] were in the 1970s. Next, consider the corresponding prospect of old age ratio defined as the ratio of the people who are older than [04:28:04] the PAT relative to the people aged 20 up to this PAT. In contrast to the retrospective measures, there are prolonged phases during which this ratio is relatively constant or even declining, implying that the population [04:28:18] is growing younger rather than older. And this can happen when improvements in population health outpace changes in cohort structure. [04:28:29] And if you combine these two figures into uh one graph, what we can see is that the distinction between these two concepts can be meaningful depending on the time at which we uh uh measure aging [04:28:43] using these measures. So for instance, if you look at OCD countries here illustrated by the red and blue lines with the circles and the squares, we can see that after 2010, uh the prospective [04:28:56] old age ratio is increasing at a much slower pace than the retrospective old age ratio suggesting that aging might not as fast as the conventional retrospective or chronological measure would lead us to [04:29:09] believe. Moreover, we can see that OCD countries tend to be relatively younger under the prospective measure after 2010 [04:29:22] whereas the nonCD countries tend to be older under pro under the prospective measure. And the reason for this discrepancy is that um population health is worse in many nonocc countries than [04:29:34] in OCD countries which means that the implicit old age ratio of 65 and the retrospective measure might be simply too high for some of those countries say think of Afghanistan or Syria. [04:29:48] >> Uh Rainer just just a clarifying question. So just to give us a sense of the orders of magnitude here. So this POT measure that you have by how much like what is it in 1950 versus 2050 like by how much how much of an increase are [04:30:02] we talking about? Um the increase depends on on the countries that you're looking at but we we can think of between 5 and 10 years older than than 65. So for some countries this might be 75. For some [04:30:16] other countries it might be 67 or so. If you go into the non OCD countries it can even be slower than 65. So you could be at 60 or 55. [04:30:24] >> Okay, thanks. [04:30:26] >> Thank you. Great question. Um, oops. Let me quickly go back. So the distinction between these two measures also matters for labor potential as measured by the [04:30:39] working age share. Again focusing on OCD countries uh in blue and red we can see that the working age share is decreasing at a much slower pace after 2010 under prospective aging than under [04:30:52] retrospective aging. And the differences are quantitatively meaningful. So by 2050 um the gap amounts to about 5 percentage points or 10% in relative [04:31:05] terms. So we use those prospective and retrospective measures of age structure in a very simple uh and standard growth accounting model. We deliberately [04:31:20] um use such a simple model to study how the conceptualization of age structure might affect the pot uh the potential effects of population aging on economic [04:31:32] growth. So we really try to get around most of the complications that would arise in a more difficult model. [04:31:40] Good. So specifically we assume a very simple production function model that expresses output as a function of TFP total factor productivity I should say physical capital human capital and [04:31:52] labor. And we reformulate this simple model in terms of output per capita by dividing by L and then using a very simple accounting identity. [04:32:09] This transformation incorporates the working age here indicated by the letter W. And this working age share serves as a proxy for the economy's labor potential. And it depends on the [04:32:22] population age structure. Notably, the effects of the working age share can be positive or negative. So if you have an increase in the working age share, it can increase output per capita. If you have a decrease in the working age share, it can uh lower output per capita. [04:32:37] And taking logs and first differences gives us a very simple model in growth rates. [04:32:43] I should also mention that this very simple accounting identity that we used to incorporate the working age share imposes a parameter restriction for the effect of the working age share on output per capita. So this coefficient is equal to [04:32:57] one and this implies essentially the accounting effect of changes in the working age share on output per capita and we will later use this restriction for interpreting the empirical estimates that I'm going to show you in a couple [04:33:10] of moments. Now our simple model we model productivity growth as a condition conditional convergence process. Um which means that uh changes in total [04:33:24] factor productivity depend on the productivity potential in the current period t and the realized productivity in the prep period t minus one. Moreover there's a potential uh productivity [04:33:36] shock absylon which will serve as an error term in our empirical model. [04:33:44] We model the productivity potential uh as a function of time fixed effect capturing uh changes in the world productivity frontier. Uh the human capital stock which uh is included to account for the fact that countries with [04:33:59] more human capital might uh innovate more than those with less human capital. [04:34:03] and some additional uh country and time varying uh characteristics captured by the vector x that can include uh or that can determine growth. [04:34:14] We further followed the um empirical growth literature in that we model uh past productivity by past output per worker and then recast this expression in terms of [04:34:27] output per capita in order to also introduce a lacked level of the working age share. So this allows us to also account for the level effects of a working age here. Let me briefly check. [04:34:39] Was there a question in the Okay, good. Um so if you combine all these uh elements we arrive at our empirical estimation equation which essentially expresses [04:34:54] changes in income per or output per capita as a function of productivity changes in input factors and an error component. [04:35:04] We fit two specifications of this model. One is a simple model in which we assume this parameter row to be equal to one. [04:35:12] So that means that we have uniform growth trends and uniform convergence. [04:35:15] So this keeps the model very simple. A second specification um allows for richer heterogeneity by uh allowing for country specific growth trends and convergence. So to do that we additionally incorporate country fixed [04:35:30] effects in this already first differences model and then also incorporate uh a lag human capital term that allows for faster adoption of existing technologies in the spirit of NES and fels if you have a higher human [04:35:44] capital stock. Please note that this additional structure that I'm showing you here is not strictly necessary for answering the question of how the conceptualization of population age structure affects the [04:35:58] economic growth effects of aging. But we incorporated in additional tests to gauge a little bit how the model um the model fit in within and out of sample projections. [04:36:11] Good. So we fit this model um using crosscountry panel data that are measured in five-year intervals in 1950 to 2015. [04:36:24] The key challenge or the key econometric challenge that arises when fitting such models is that uh the key uh variables might be endogenous and specifically this might apply to changes in the working age share which might be [04:36:37] affected by all kinds of variables. So the approach that we are using here is um that we instrument changes in the working age share between periods t minus one and t with the predicted inflow based on the cohort size in [04:36:52] preceding periods. So the way in which this works is that you have a measure of the working age share here defined for ages 20 up to the PT [04:37:03] and you can predict the change in the working age share by using say the population share of the age core 15 to 19 5 years earlier to see how uh this uh [04:37:17] quantity is changing one period later. And of course you can also go further back. So you can go 10 years back or 15 years back or even 20 years back and use these different population age shares to predict how the working age share [04:37:32] changes. And apart from that or apart from using the population age share, you can also use fertility rates as you as we have seen earlier today. And we are doing all of that in in um [04:37:45] in our robustness tests to convince you that none of these different approaches are really crucial for the key results that I'm going to show you. [04:37:56] In simple and plain English, what uh does this approach assume regarding the identifications? So it means that the unobserved heterogeneity in the current period does not affect the population age structure in the pre periods or put [04:38:10] differently people do not systematically anticipate productivity shocks that would lead them to alter their behavior in response to those shocks. [04:38:20] Good. So let me uh show you the results. I begin with a very simple parsimmonious uh specification that follows the spirit of the demographic dividend literature. [04:38:33] So this the specification regresses income per capita growth on the growth of the working age share the log of the the lack level or the lack lo of the working age share and also the lack lo income [04:38:48] per capita in addition to time fixed effects in a first differences model. [04:38:54] And what the estimated coefficients show you is that the growth effect of the working age here and the level effect of the working age here are larger than the corresponding parameter restrictions that are implied by the accounting [04:39:08] identities. So for the working age chair this or growth of the working nature this would be unity and for the lack level of the working age chair this would be um minus this coefficient and the fact that the estimated coefficients [04:39:21] are larger than than the parameter in restriction implies that in addition to the accounting effects of change in population age structure there are also behavioral changes in response to population aging. So people save more, [04:39:34] people innovate more, um people might even trade more like all these different mechanisms that in some form we have seen today and discussed today. [04:39:44] So all of that essentially um reproduces what has been shown in in previous work on the demographic dividend. Um so once we include uh the additional coariantss that might capture these behavioral adjustments say the capital stock say [04:39:58] human capital we find that the estimated coefficients for uh the growth and level effects of population aging go back to um parameter restrictions that we find in the model that is uh we all what we [04:40:13] have left is the accounting effect of changes in population age structure and all of these results uh are stable for different samples. [04:40:22] and also for different uh configurations of the specifications with respect to coariantss and uh alternative estimation approaches. [04:40:31] So what we can take away from this exercise is that um there are meaningful effects of population age structure on economic growth. And next we want to see how the potential effects of population aging might look in the future and to [04:40:45] what extent these effects depend on whether we use these retrospective or prospective measures of population age structure. [04:40:52] So to do that we use the fitted model to project income per capita in 2020 to 2050. And in these projections we make a couple of structural assumptions which [04:41:04] we lack subsequently in um in sensitivity analysis and we input the demographic projections that we take from the data. So we have demographic projections about future population age [04:41:17] structure also future human capital um attainment based on uh education or uh based on educational uptake. And in our baseline [04:41:29] specifications we assume that uh capital stocks evolve exogenously. This is something that we are um relaxing and that's an assumption that we are relaxing later. So in these projections we compare three different scenarios [04:41:43] that model income per capita. So the first is a counterfactual no aging scenario in which we force the working age share to remain at its level of 2015. [04:41:56] The second is a retrospective aging scenarios. Essentially the standard approach in which the working age share varies according to its projections um for the conventional definition of working ages 20 to 64. And the third is [04:42:11] a prospective aging scenario in which working age shares vary according to the evolution of working ages 20 up to this prospective old age threshold. And the third scenario allows essentially the [04:42:25] working age population to grow if there's a little bit of improvement in population health. [04:42:32] We are not interested in the level of these different in the levels of these different scenarios because the levels crucially hinge on the assumptions that you make about the growth the underlying growth processes. However, we're interested in the differences across these scenarios. So the difference [04:42:47] between the first and the second scenario gives you an estimate of the potential growth effects of changes in cohort structure. So this is something that has been done in one form or another in the literature before. [04:43:00] The difference between the first and the third scenario gives you an estimate of the potential growth effects of changes in cohort structure plus changes in the age patterns of health. And the difference between these two differences, the different diff if you [04:43:14] will, gives you an estimate of the potential growth effects of changes in um the age patterns of health. [04:43:24] So how do the results look like? So this is um an overview of our projection analysis uh for all OCD countries pulled together. [04:43:35] The level is not so much interesting but the ordering of the three different projection scenarios is interesting. So according to this projection income per capita is highest in the counterfactual no aging scenario here indicated by the [04:43:48] orange line with the uh small diamonds. It is lowest in the retrospective aging scenario illustrated by the red line with the small squares with the prospective aging scenario um lying in [04:44:01] between those two. And this ordering illustrates our two key points. First, income per capita's highest under the counterfactual no aging scenario indicating that on average OCD countries [04:44:15] will in all likelihood experience a demographic drag in the years ahead that will slow the pace of economic growth. [04:44:24] And second, the fact that the prospective aging scenario leads to a higher income per capita level than the retrospective aging scenario suggests that there are appreciable economic gains associated with expansions in [04:44:38] labor potential due to improved population health. [04:44:44] Good. So next uh let's have a look at the effects for the different countries in our sample. So this uh figure shows you that the projected change in annual growth rates for each of the 34 [04:44:57] countries that were members of the OECD in 2015. [04:45:02] I begin with the retrospective aging scenario here indicated by these blue bars into small squares. [04:45:08] And what you can see in this figure is that practically all OCD countries are expected to experience an appreciable demographic drag which um ranges from about 0.4%age 4 percentage points in the US here, the fourth [04:45:23] country from the left, up to 1.7 percentage points for the Republic of Korea on the right. And the estimated demographic drag tends to be larger for the fast aging countries including Korea [04:45:36] and Japan, but also Germany, Austria, and Switzerland, as well as some southern and Eastern European countries. [04:45:44] What all of these countries have in common is that they had relatively low fertility rates in past in the past decades and only limited immigration to offset um the low fertility rates. In [04:45:58] contrast, the estimated demographic drag tends to be somewhat smaller in countries with higher fertility rates or higher immigration, including the United States, also the Scandinavian countries, [04:46:10] and France. However, the estimate or the the actual demographic drag could be appreciably smaller than what the retrospective aging scenario indicates. And this is [04:46:25] something we can um measure by looking at this prospective aging scenario which accounts for the fact that the prospective old threshold might increase say from 65 up to 70 or 72 in some of [04:46:37] those countries. And the difference between these two bars, that is the difference between the red bars with the diamonds and the blue bars with the squares shows you how much improvements or expansions in labor potential due to [04:46:51] improved population health um or how much of the demographic drag could be offset by expansion in labor potential due to improved population health. And this counterwaling effect is larger for [04:47:05] countries with relatively or with relatively healthy populations including for instance Sweden or Chile or Japan which are countries that are at the top um of the pop of like have the most healthy populations in their respective [04:47:19] world regions. And uh the counterweailing effects are much more muted in countries whose relative whose populations are relatively unhealthy. [04:47:26] For instance, the United States, but also Hungary, which is a country struggling with high cardiovascular disease mortality. [04:47:34] Notably, the two projection scenarios that I'm showing you here constitute bounding cases. So the retrospective aging scenario assumes that none of the improvements in population health will lead to higher labor potential and [04:47:49] therefore any economic gains. Whereas the prospective aging scenario assumes that all of the improvements in population health automatically translate into higher economic performance. And in reality we will probably end up somewhere in between um [04:48:03] depending on people's preferences as well as the economic and um political conditions prevailing in the given countries. [04:48:11] Good. Before I uh conclude, let me very briefly speak about low and middle inome countries. What can we say about those countries? Um the population in these countries tend to be much younger than in rich countries, suggesting that there [04:48:25] might be an opportunity for a demographic dividend. However, there's also the risk that these countries are quote unquote aging faster than rich countries because they have lower living standards and worse population health. [04:48:38] Now our analysis suggests that the retrospective measures of population age structure tend to overstate the potential for a demographic dividend. [04:48:46] Specifically, this means that the blue bars that uh capture our retrospective aging scenario tend to be higher than the red bars capturing our prospective aging scenario. And what this means is that [04:49:01] what this means is that the magnitude of a potential demographic dent might be overstated under these conventional retrospective aging scenarios. And moreover that the window of opportunity for potential demographic dividend might [04:49:14] be slightly shorter than um these retrospective measures for age structure would lead us to believe. However, I should also say that any of those comments come with additional caveats. [04:49:24] So especially um referring to the low and middle inome countries um many of those have underdeveloped pension systems and large informal sectors. So it's less clear whether people can enjoy a long retirement period. So aging might play out slightly differently than in [04:49:39] rich countries. So let me conclude here. I have showed you that the contraction of working age shares will likely slow the pace of economic growth in the years ahead and that a demographic drag at the end of [04:49:52] the demographic transition appears very likely. In fact, we find the scenario for about 70 out of the 145 countries studied in our sample. [04:50:03] Moreover, I argued that healthy aging could offset a significant portion of this demographic drag if the economic and political conditions uh uh allow um the improvements of population health to translate into [04:50:18] higher economic performance. We believe that these results make at least two important contributions. [04:50:25] First, they indicate that the retrospective measures of population age structure tend to overstate the economic consequences of aging because they ignore population health. And this effect goes in both directions. That is, they tend to overstate the negative [04:50:39] effects of population aging in developed countries and they tend to overstate the positive effects of population aging in developing countries. [04:50:47] Moreover, the results also point to the possibility that policies that promote healthy aging might have economic gains above and beyond improving human welfare. However, these economic gains do not um occur automatically. Rather, [04:51:02] the economic and political conditions have to be in place um for uh the market to essentially retain this additional labor um productively. Thank you very much. ## Q&A (04:51:16 – 05:07:54) [04:51:16] Great. Rainer, thanks so much for uh finishing just on time. So, we have 10 minutes uh for questions. The floor's open. [04:51:38] I guess I'll I'll take the prerogative of a a question. My prerogative as organizer. So um so I think this uh this approach is is really interesting to kind of qualify like simple shift chair [04:51:51] type calculations uh for the demographic drag you get um but I wonder whether there's a possibility to extend it to not just think about when is it that working life [04:52:05] ends which I think is what your poe measure does right it measures I guess the it just it's just based on remaining life expecty See um but also say productivity improvements at a given age. So if we think that the sort of [04:52:20] total workforce effective workforce is is based on the amount the the product like productivity by age you know weighted by the share of the [04:52:33] population that age. Then as demographic uh change unfolds, you have also uh this additional effect which is that maybe as people get healthier and so on, they get more productive at a given age. So I was wondering whether [04:52:47] you'd you'd thought about that both empirically as some measure of say productivity improvements by age and and what that would do to a calculation such as the one you're doing about you know [04:52:59] um you know demographic drag under under this you know prospective aging versus this more u sophisticated measure of prospective prospective aging that takes that into account. [04:53:11] Yeah, thank you. This is a fantastic question and I I have been thinking about this a lot. Um, so far I only have limited answers to to your question. So I view this paper sort of as of a proof of concept that in general we should [04:53:25] think more about different conceptualizations of population age structure. And you're correctly pointing out that this me measure focuses really on the mortality margin and therefore is somewhat silent on [04:53:40] the different productivity effects at different ages. So optimally what we would what we would love to have is um an age specific measure that accounts essentially also for a mobility uh um [04:53:53] schedule. Um what I've done so far in a different paper is to try to understand how changes in in function capacity measured by people's hand grip strength translate into different uh uh labor [04:54:08] supply and different engagement in um in productive non-marked activity. [04:54:14] However, I as of now I don't have a macro analysis and I think like this is high on my agenda and I would be very interested also in in collaborating with other people on on this issue to to make this sort of exercise maybe for one [04:54:27] country rather than for many countries so where we have a much richer micro database to implement uh these uh productivity schedules that you just mentioned. [04:54:36] Great. >> Thanks Diego. Oh Diego I think yeah Diego and then Ron. Yeah, >> thank you. Um, first I want to say if you're trying to find people to [04:54:50] collaborate on this, happy to uh to uh offer myself. I am I I work at the World Bank. [04:55:00] >> Excellent. Fantastic. I work at the I work at the World Bank. Our team is trying to do very similar modeling and we're actually trying to develop scenarios where we can build up uh policies that will impact health and how [04:55:13] those can buffer the effects of aging. So I'll send you an email after the conference how we can think about that and particularly important for my team and for for the World Bank is how these uh issues affect low and medium uh low [04:55:28] and medium income countries. No. And uh a way to put it for those countries is uh people tend to use the phrase work until you drop. Labor participation in those countries for older age groups is [04:55:41] really high already. Partly because what you say there is no special protection system in place. So maybe they have to work until because they don't have any other source to finance those extra years of life [04:55:55] that come with a population agent. The question there is how do you apply then the model in those in those cases um when labor force participation age at 65 or out of 55 is already way higher than [04:56:08] for high income countries. Does this concept of prospective dependency still apply? I would expect these countries can lower level participation at all age in the future because they can uh stop working until they are dropping [04:56:23] if that makes sense. >> Yeah. Uh thank you for these two comments. First of all, um let's keep up uh uh or let's catch up after the conference uh on these issues. Um second uh regarding the low and middle inome [04:56:37] countries, I agree that the absence of of a strong social safety net and the existence of large informal sectors uh severely complicates these predictions. [04:56:49] So um I'm I'm not I I would not say that um that our model is really tailored to these countries. In fact, I think this the model is essentially tailored to the OCD countries. However, the message that [04:57:04] I I think that applies to low and middle inome countries nonetheless is that um once we consider differences in population age structure, the um potential window of opportunity for demographic dividend appears much [04:57:18] shorter than the chronological measures of age structure would us lead to believe. Um more generally I believe that when it comes to these countries it might actually be a good way forward to think about um essentially two economies [04:57:33] within a country that is um one economy that is more formal um for instance in the in the larger cities where a social safety net is in place and one economy that is more informal and characterized [04:57:46] essentially by by rural and urbal areas. But again, I think like there's more work that needs to be done on these issues and I I concur with you that that it's difficult to to model that um or in [04:57:59] a way that's comparable to developed countries. [04:58:02] >> Great. Thanks. So, I think Ron had a question and then we have Joe and Will. [04:58:07] >> Yeah. Uh thanks. So, I enjoyed that uh paper. um getting to this question of whether you know retirement age can rise, [04:58:20] whether labor force participation rates at older ages will rise and so on. Um so in the US we see labor income at older ages has gone up really an extraordinary [04:58:35] amount far more than the labor force participation rate has because although some people are working in old age because they are compelled to by economic necessity [04:58:47] most of the cases I think are people with more education higher skilled jobs higher wages satisfying work, who are uh continuing [04:58:59] to work. And uh for lower um education, lower lifetime income people, on the one hand, their benefits are relatively [04:59:12] inexpensive. On a second hand, they have unfortunately much shorter life expecties in retirement. And um [04:59:25] I think in fact uh their retirement ages should stay more or less where they are or rise slightly. I think there should be heterogeneous um [04:59:38] retire normal retirement ages in public pension systems that reflect uh differences in lifetime earnings. And for the most part, it's the high lifetime earnings people who have very costly pensions who have uh better [04:59:52] health, longer life expecties and the ability to work longer and often the wish to work longer. Uh so I also want to call attention to the fact that there's enormous variation across [05:00:05] countries in the arrangements for support of the older people. In Singapore, for example, there are essentially zero public transfers for [05:00:17] health or pensions to older people. And older people in Singapore work quite a lot and they also get quite a bit of support from their families, but not from the government and only to a [05:00:30] limited degree from the mandatory asset accumulation they've spent their lives achieving. And the last thing I wanted to say was about the lower income and lower middle inome countries and upper [05:00:42] middle for that matter. Um the older people often have very little education compared to younger workers. They have very low productivity and uh their [05:00:54] participation doesn't directly add much to the national output or something of that sort. But I think it is valuable to begin for people to begin to get the [05:01:08] idea that old age in well whenever we think it begins it involves a continuing work. That's part of life and it won't be long a decade or two before more [05:01:21] highly educated young people will be becoming old and then their labor force participation will matter a lot. Anyway, thank you very much. [05:01:32] Thank you. Uh I would take the questions of Joe and Bill also and then maybe reply in chat if that is corre if that is fine. [05:01:39] >> Yeah. And and just to keep track of time so we're just about on time. Uh we have 10 minutes uh until the next presentation. Uh in principle this is a break but this discussion is really interesting. So um so let's maybe continue with Joe and Will's question and then perhaps we can bring Bob Gordon [05:01:54] in also on this debate about pension reform if he's here. [05:01:59] >> Thanks. Uh so my question it's it's it's related to the one that that Adrian raised at the at the beginning and I guess thinking about if you were to move from just shifting the retirement age and instead having this kind of productivity or effective labor force [05:02:13] that can that can evolve across the age distribution with um changes in in health and even with thinking about kind of what kind of labor people are doing, right? if if there's changes in the demand for labor, moving away from [05:02:27] things that are sort of bad for health, you know, like really intense physical uh labor type jobs. I guess my question is if you were to move towards that, would you expect this to be something that lives in between your kind of results of the retrospective or [05:02:41] perspective or is there scope within this framework to think about how that would actually sort of go even beyond the sort of results you have with the sort of all labor is kind of treated the same and you're just moving the the cuto [05:02:54] off here. Um, but I don't know if I'm even framing that correctly, but that's just a thought I had. You mean like automatically linking the retirement age to to >> No, I guess what I mean is if you were to be able to sort of think about sort [05:03:07] of effective labor across the working life as being more or less productive because of these health improvements or from uh like would that just sort of live in between these two results of the retrospective or perspective or is there [05:03:20] some kind of scope there for there to be additional kind of benefits >> coming through that? Um I guess is my question. Thanks. [05:03:29] >> Mhm. >> Yeah. So I had a a somewhat related question and I think it makes a lot of sense to start it thinking about the [05:03:41] retirement age and end of uh end of the career because we have strong institutions which can affect this behavior. But I would love to think more about the labor supply side of this and the demand for leisure at different [05:03:55] ages. So you as people are expected to to live longer, if they're going to be expected to work longer in, you know, past age 65, we're looking at our dependency ratios where we're starting [05:04:09] at age 20, say, but we're seeing in uh increased youth unemployment and underemployment. And some of that is preference-based. Uh, and then we have [05:04:22] increased demand in wealthier countries for uh, uh, paid leave, family leaves, and that kind of thing. So, I I'm just I'd be curious to see how much of this might be offset by decreases in in labor [05:04:37] supply throughout the the life course. It's just to put some some numbers on it as you're you're continuing to flesh out this agenda. [05:04:49] >> Wonderful. Um great question. So let me let very briefly reply. Um so in response to Ron, I absolutely agree that there is a tremendous amount of variation and [05:05:03] heterogeneity when it comes to both people's productivities and also people's wishes for for productivity. So it might be that the way ahead from a polit policies perspective is more flexible finding of more flexible [05:05:17] schedule when it comes to retirement rather than just sticking to one or two retirement ages. So when I redid a similar like it's not a similar analysis but when I investigated the role of chronological and functional aging for [05:05:32] labor supply and retirement in a different paper what I found is that for many people it seems that chronological aging is really determining labor force exit around the retirement ages but it's really health problems that um leave out people that push out people earlier or [05:05:47] maybe even at later ages. And what this suggests is that um there need to be disability um uh schedules in place that or disability insurance schedules in place that help those people who really [05:06:00] need help because of their health and then there needs to be a solution to the general chronologically aging related retirement that people at some point expect because they have quote unquote deserved it. Um with respect to Joe's [05:06:15] question um in principle yes so if we if I were to sort of incorporate um an age um productivity schedule I would get different um uh sort of like bounding [05:06:29] cases for retrospective and prospective aging um and then we would again still end up somewhere in between but uh that would require me to incorporate such a schedule in the prospective and retrospective aging scenarios. Um so I [05:06:42] think like without that there is no no trivial comparison to the existing results that I have shown you and yeah again regarding to to to Bill's question I I I think this is a very [05:06:57] important question I don't think I or I'm not sure whether I'm having a convincing answer to all of that um in the sense that you're right that with lives becoming longer essentially there's also [05:07:09] more heterogeneity in how people uh approach this life. So some people start working earlier, some people start working later. So we are starting here at age 20 because we need to make an assumption um about at what age we should start. But in principle there [05:07:24] could be a lot of heterogeneity about that also. And um we not really speaking um to any of those um pauses or gaps that you might have during your life cycle. But I think like that might [05:07:37] become increasingly relevant given uh heterogeneity in in fertility patterns in in careers and also um preferences for working at old age. [05:07:48] So thank you for all those questions. I think these are really insightful questions and also comments.