Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. ======================================================================== # Human capital in developing countries (STEG 2021) Authors: Discussant: None Video: https://www.youtube.com/watch?v=8PCada1BR1I&t=0s ## Talk (00:00:00 – 01:23:23) [00:00:02] when we get started here um thank you for coming um to this third lecture uh first lecture on thursday the first supplemental lecture [00:00:14] of the class um my name is joe kobaski i'm uh one of the co-leaders with doug golin of the steg program this course is being offered by organized through the steg program [00:00:28] which is supported by the united kingdom's foreign commonwealth and development office as part of the uk effort we have people in the front row as panelists you can interrupt use the microphone we [00:00:43] encourage you to interrupt with questions just as in a normal classroom but otherwise if you're not interrupting please mute yourselves those of you who are viewing as a webinar you can address questions in the q [00:00:57] a you can upvote them we'll be trying to um answer those as we can and if there's you know particular popular questions uh [00:01:07] um todd shulman will um leave some time at the end to answer the questions uh the lecture notes are posted at the stag website uh there will be a youtube recording this is also being broadcast [00:01:22] via youtube so you can go back and watch it on youtube i want to thank todd schulman from the minneapolis fed for teaching the class today again he's doing this out of the kindness of his heart i want to thank [00:01:34] mandy and kirsty from cepr for everything that they've done for logistics and um i want to announce that tomorrow we have uh bertolt herrendorf who will be talking about structural [00:01:48] transformation at the same time 4 p.m london time um okay todd go ahead all right thanks i want to say first thanks to both of you for organizing this whole enterprise and and [00:02:03] these lectures in particular and for inviting me today so what i'm supposed to talk about is human capital in developing countries since i work for the fed i have to stop and make sure you all see here that these are my views about human capital in developing countries and not the [00:02:18] federal reserve banks so i like to start by just making sure we're on the same page i think we often talk about human capital but here's going to be my formal definition of it i want to think about it as being the stock of productivity enhancing [00:02:33] attributes embedded in people and for clarity i want to make sort of two points right up front the first is that in general we should think about this as being a vector as being a multi-dimensional stock it includes a wide variety of attributes [00:02:46] it includes the knowledge of workers their skills the habits they develop over the course of their life their health which helps them be more productive on the job so i want to think about this as being a long vector of attributes that people have [00:02:59] the second kind of key attribute component of human capital is that it's generally produced through investments okay so there's some active uh expenditure of time or resource that goes into enhancing human capital [00:03:13] for example education is used to produce knowledge training or practice to produce skills and habits exercise to improve your health etc etc etc i think it's really common when we talk about human capital to jumble all of this stuff together [00:03:28] primarily to be honest for measurement reasons we can't really measure human capital and often we can't really measure your knowledge or your skills exactly what we find it easiest to measure is how long you spent sitting in classrooms [00:03:41] how many years you went to school maybe we can measure how many minutes a day you exercise even if we can't measure your health but at least at the outset i want you to kind of be clear these are distinct objects human capital has a long history in economics although it wasn't always [00:03:55] referred to by that name if you go all the way back to adam smith you'll see he talks about the skills of workers and how workers with different skill levels have different productivities that's the same idea but it was really brought to the forefront of economics in [00:04:08] the late 50s and early 1960s when becker minster and others started to work out a full theory of human capital what i mean by a full theory of it is that they have a models where workers or their families are [00:04:22] making investments and they have rational cost benefit type considerations you know how long should i go to school i should go until the return on going to school is equal to the return on other activities i might undertake things like that okay from that time [00:04:36] there's been a huge explosion in the literature and there's you know a great deal of work on human capital in general what i want to do today since there's been so much work because i'm going to focus on two very specific things that i think fit well [00:04:51] with the theme of macro development and stag and that you're working through the first thing i want to do is measure as best we can or really estimate might be a better word cross-country differences in the stock of human capital [00:05:05] when i do that i'm going to try to convince you of two things i think there's sort of two emerging uh there is an emerging consensus here which is that cross-country differences and human capital stocks are large that's a statement that has to be relative to something and i'm going to [00:05:20] think of it as human capital stocks or large differences in human capital stocks are larger than what the early consensus held the early consensus would be like what julieta taught last week and i'll try to be more precise about [00:05:33] that in a minute the second is i'm going to say that it's large relative to cross-country income differences that is to say that human capital plays an important role in thinking about why rich countries are much richer than poor ones [00:05:46] there's relatively less evidence on the role that it plays for example in structural transformation i know that's a key theme of stag so i'll try to highlight a few references and some of the work that's being done in that area but i think that's an area [00:06:00] that we need a lot more work on since i'm going to be focusing on these cross-country issues in other words comparing human capital stocks between rich and poor countries or middle income and poor countries i'm going to have relatively less to say [00:06:14] on the vast and excellent literature on within country human capital differences that's just a function of how much time we have today i don't have time to get into it do you take as given that um there are large differences in human capital stocks [00:06:27] across countries or across sectors within a country then as somebody who works on macro development or who's interested in structural transformation you might want to build human capital into your thinking and one way that we we of course make progress here is by building models and so i'm going to try to give you some [00:06:42] ideas the second thing i want to do is illustrate the key model mechanisms that you can use to think about human capital differences across sectors and across countries so i'm going to build on some very classic results especially from becker and ben parath i'm going to show you [00:06:56] three maybe four if i have enough time of the main mechanisms that people use to generate human capital differences as the results of workers and their families endogenously deciding how much human capital to acquire that's kind of the [00:07:08] underlying idea i think it's always useful to at the outset say sort of what's the underlying challenge that we confront and with human capital i think it's really a big one and i like to pose it [00:07:23] as a simple thought experiment which is imagine i ask you even pick within a country take two workers imagine i ask you to explain why some of those workers are more productive than others what drives productivity differences between workers [00:07:37] even if you focus on a country like i'm an american i live in the united states you know why are some americans more productive than others i'm hard-pressed to tell you and i don't think the economic literature has settled on an answer how much of that is human capital [00:07:51] versus how much of its other characteristics like luck or the job you type you happen to work in or how much of it's driven by things like discrimination you know those are questions that people are wrestling with now in other literatures and i don't [00:08:05] think they're settled issues even if we agree that it's human capital i think we're also going to have a hard time putting our finger on exactly what it is and an easy way i can say this to you is that i have the good fortune to work with ed prescott [00:08:19] ed is like me a phd educated economist so we have the same job we have the same employer we have the same education level and yet ed is vastly more productive than i am at producing economics papers why is [00:08:34] that i think we can all agree it's something that ed has it's some human capital characteristics that he has that i don't have but if i ask you to put your finger on exactly what those are to measure them you would have a hard time doing that what we're going to do today is [00:08:47] something that's even tougher than that we're going to try to investigate these differences across countries the set of relevant human capital characteristics that we might be interested in is going to be even larger and the set of confounding or environmental factors that we need to think about and control for is even [00:09:02] larger and so in general i want you to just acknowledge right up front measuring cross-country differences in human capital stocks is really tough i'm going to give you my best guess you know the best approaches that i think we've discovered so far i'm going to give you the answers that they lead to [00:09:17] but i want you to be throughout this vaguely dissatisfied because i think a vague and persistent sense of dissatisfaction is a great tool for research you know if you're never quite happy with what's been done and you're always a little uncomfortable with the assumptions that's a great way [00:09:31] to get started in your own work all right so let's start with part one how do we measure human capital differences the techniques i'm going to show you two different ways to go about doing this [00:09:44] and they're both going to be closely tied to development accounting so just like julieta's lecture last friday so she's done a great introduction to this i'm just going to remind you a little bit of the basics that i need and then we'll get going [00:09:58] so for development accounting what you really need the main assumption you start with is about the aggregate production function so i'm going to think about capital y that's output that's ppp adjusted gdp of country c [00:10:11] as being a cobb douglas production function of the physical capital measured in ppp adjusted dollars raised to the capital share and then tfp total factor productivity and the total human capital stock of [00:10:25] country c okay and julieta walked you through last time we think that we have good evidence that this production function provides a useful fit to the data we think we roughly know what these factor shares the alphas are so this is [00:10:38] sort of a good starting point for us just like she did i'm going to divide both sides through by the stock of workers in the country so then i can think about gdp per worker in country c and that's going to be determined by three components that she talked about [00:10:53] the capital to output ratio raised to a power total factor productivity and the human capital stock per worker people will sometimes call this last piece the average human capital you do actually technically need extra assumptions to say that [00:11:07] i may slip and call it that myself but anyway human capital per worker so generically in the development accounting literature what we're trying to do is understand differences or variation in this gdp per worker as a function of [00:11:22] how much capital to output every country has the tfp in every country and the human capital per work in every country today i'm going to be really focused on what can we say about human capital so i'm going to lump the rest of it [00:11:35] together and just call it z other stuff that i'm not interested i'm not going to try to measure i have nothing to add there and so we're just going to ask what can we say about human capital like julieta did i'm going to take logs [00:11:48] of this equation so basically we're going to investigate variation or differences in log gdp ppp adjusted gdp per worker that's pen world tables 10.0 that's going to be in part a function of capital output and tfp differences [00:12:03] and part human capital per worker differences okay so that's just to get you back to where julieta left us and what we're going to do now is consider what what is sort of the the best evidence about how we can measure this object h and think about its contribution to the [00:12:18] ppp gdp per worker difference as we see i've already established that that's a challenge and and julieta did a nice job with this but i do want to take one extra minute and point out that regressions are probably not going to be the way that we [00:12:32] identify or measure h or its importance for cross-country income differences she gave you one perspective on this which is that you should think about this as being a production function and the estimation of production functions is a very difficult task [00:12:46] another way to think about this is that i like to borrow from mancus bookings papers he has a very nice discussion of the problems with cross-country regressions and his view is that this is really a [00:12:59] very simple and practical problem at the end of the day if i want to investigate human capital using cross-country regressions how many countries do i have something like a hundred is typically how many countries you're going to be able to get data for and if i want to think about [00:13:13] cross-country differences you know how many right hand side variables would i want to include how many things do i think might generate cross-country income differences including human capital skills capital institutions etc etc and his point that he makes very nicely [00:13:28] is just that if you only have 100 observations and you have hundreds of things you might think about as being determinants of gdp per worker it's kind of a hopeless endeavor and so we need to find a way to circumvent this we need to find a way so that we can have more than just 100 [00:13:42] observations to help us learn about human capital so how are we going to do that sorry 10 a.m here so i still need coffee um we're going to build on two [00:13:57] assumptions that are going to allow us to make substantial progress and this was first worked out by bills and clean out a 2000 aer paper those two assumptions i'll just state them first and then we'll talk a little bit about their significance the first is that there's perfect [00:14:11] substitution among different labor types which is sometimes called the efficiency units assumption what this assumption means is that for the purposes of producing aggregate output we all supply labor there's only one input called [00:14:25] labor some of us just happen to provide more or less than others so you know the way to think about this is okay if you add up enough todd shulman's eventually you can have a bill gates and we just have to figure out how many todd shulman's it takes [00:14:39] to provide the labor input of the bill gates or whoever your famous your favorite famous person is but there exists some number and we know what that is we can solve it okay the second assumption is that labor markets are competitive [00:14:53] and what that really means is that workers and firms take prices as given or that prices are set by marginal products okay under these two assumptions we can use the firm's first order conditions to characterize the demand [00:15:07] for workers who have hi units of human capital in country c so the competitive labor markets assumption we're going to use that first to say that the wage of worker i in country c is equal to their marginal product of their labor then we're going to take the derivative [00:15:24] we're going to just find what that marginal product is and it's going to depend on the labor share the capital output ratio tfp and the human capital of the worker and here's where the perfect substitution assumption shows up [00:15:37] it says that the marginal product of labor depends on how many units of human capital they provide in a world where there's not perfect substitution but there's different types of labor there's an extra term that shows up here which is the price of my type of labor [00:15:51] so for example if you're used to thinking about models where there's skilled and unskilled labor and they're different inputs maybe with a ces production function then you would have a term here that says what's the price of skilled or unskilled labor the fact that we don't have that comes [00:16:05] from the perfect substitution assumption so that's what those two assumptions buy us if you stare closely at this equation you'll see that there's the capital output in t of p terms just like before that's what i labeled z on the last slide so we've got that same [00:16:19] z that's what this really tells us is that the wage of worker i in country c depends on that country stuff and the workers human capital i find it most useful to explain the power of these assumptions by [00:16:35] turning this whole thing around what this says is that if a worker has twice as much human capital they're going to make twice as twice as high of a wage but in practice what we really want to do is [00:16:48] invert that and use it the other way around we want to say if i see a worker making twice as high wages they must have twice the human capital and the reason that's really useful is on the last slide i was saying boy if we want to measure human capital using [00:17:03] cross-country evidence we only have 100 or so countries maybe 200 at the most here i'm saying well if i see a worker with twice as high of wages i know they have twice the human capital now i'm using evidence on workers wages how many observations do i have on the [00:17:18] wages of workers billions at this point it really is billions of observations and so this is going to give me a lot more evidence it's going to allow me to draw on a lot more data to discipline the measurement of human capital [00:17:32] and now you can really think about okay so what i need what i want to be able to do i said is i want to be able to say if i see a worker who has twice as high of wages he must have twice the human capital now you can really see what those [00:17:45] assumptions are doing for me if there was imperfect substitution across labor types that inference no longer holds because i have to worry well maybe he's just a different type of worker and that type of worker has a different price if labor markets are not competitive [00:17:58] then the fact that a worker makes twice the wage could just be due to discrimination or market power or luck or all sorts of other factors okay so that's what those assumptions bias they're both really important assumptions i'll come back and talk a [00:18:13] little bit about them later on i don't have time to dig into the whole literature on these today but i think i want you to understand that's what they get us they allow us to take this from being a cross-country problem to a cross-worker problem and that's [00:18:26] really what's unlocked the measurements of human capital sorry todd can i ask a question on this yes please so just irrespective of the two assumptions we are already we also make an assumption on kind of ruling out human capital externalities [00:18:41] here right we are not as we are already assuming that h aggregate h doesn't have a direct effect on a right on on z yes there is a small literature that [00:18:57] tries to measure those human capital externalities and i'm going to skip it today and and there's really it's a practical thing um it is extremely difficult to measure externalities if you think about what i'm doing here the underlying idea is that i'm saying [00:19:12] uh i can learn about the productivity of a worker from looking at their wages and their marginal products and sort of the defining feature of externalities is of course that they're not priced in the market which makes them hard to measure through this approach so i can point you to the couple of [00:19:26] papers that i think have made the best progress on that that is a much trickier issue and one that i'm not going to get into today thank you all right so suppose i take this equation very [00:19:40] simple equation right and i just take it in logs then i have a very nice log linear wage equation which says that the log wage of worker i and country c is additive in two terms there's a part that depends on the country you live in [00:19:55] that captures tfp and capital output ratios and there's a part that depends on your human capital what's really nice about this is if you lay it up next to okay let me do this differently there's [00:20:09] a very nice connection here to the enormous labor literature in particular the micro labor literature that starts with jacob mintzer and his 1974 book in that book [00:20:21] minser argues that log wages in the united states are approximately linear in schooling and quadratic in experience so he investigates a whole sets of different ways to model wages in the united states [00:20:35] and he says the best way to model wages is to think about them as being the log wages linear and schooling quadratic in experience this was enormously influential so tons of people have estimated equations of this form which is now just [00:20:48] called the mincer wage equation not just in the united states but in countries all over the world and so what's really another neat feature about these two assumptions is that they allow us to have a wage equation that looks very [00:21:02] much like what people have been doing in micro labor for 40 50 years so if you kind of stare at those there's a bunch of connections i want you to see if you estimate a wage equation of this type within a country [00:21:16] you'll see that the intercept captures all of these country specific effects okay so that's the first piece of connection the second thing is that of course this huge labor literature that's been estimating this minster wage equation [00:21:31] they're including things like schooling and experience on the job those sound like attributes of human capital those sound like things we want to include in our measure of human capital so they have been estimating exactly these types of [00:21:44] ideas and what they're doing when they estimate them these coefficients beta 1 beta 2 beta 3 they speak to the value of those things now the way somebody in the minster literature thinks about this [00:21:58] is they tell you the wage value how much does it increase wages you know how much does a year of schooling increase my wages that's what beta 1 picks up but under the assumptions i've made those also tell me how much more productive a year of schooling makes me [00:22:13] and so those connections says that now i can bring to bear all of the evidence from this minster wage literature not just in the united states but really around the world at this point to tell me what sorts of characteristics [00:22:26] of human capital i might want to include and what are the value of those characteristics and in this specification is experience just measured in years of experience yeah i'll be more precise in a second what almost everyone ends up doing is [00:22:41] something called potential experience which is the years since you graduated school and then you're going to do a quadratic people i mean the literature is built from that over the years but that's really going to be the the [00:22:54] starting point for everything okay thank you um excuse me so um by looking at this equation i can imagine that this is going to have huge errors terms so [00:23:07] i'm i'm wondering how well this um like human capital factors are actually going to explain the differences in wages i'm just saying like if these factors explain only 10 percent of the dispersion in wages maybe it's not i mean they might be contributing [00:23:22] factors but they're not going to give us a good measure that's a great point it's not so much about the standard errors it's about the r squared in a typical application the r squared of this regression will be 20 to 30 meaning you explain about 20 to 30 [00:23:36] percent of the variation in log wages in the sample and you're exactly right now you have to think okay but then what one hypothesis is that there's non-human capital stuff going on that drives wage [00:23:50] variation right block the employer you happen to meet with discrimination etc another is that maybe i'm doing a lousy job of measuring human capital and i don't know that we've ever figured [00:24:03] out exactly what it is but those 20 to 30 percent r squared's you know they are a bit worrying in that sense and i agree completely yeah okay so these really tight connections are going to lead us to the to the first approach to measuring human capital stocks which i'm going to call the [00:24:18] constructive approach and i don't mean that as opposed to destructive but i mean it to point out that you were literally going to construct human capital piece by piece with this approach so what you're going to do it's just a simple four-step procedure you're going to select [00:24:32] some dimension of human capital that you're going to try to measure so think about it is i'm going to select schooling i'm going to measure each nation's stock how much schooling does every country have then i have all of this evidence from [00:24:46] the menstrual waves equations about the value of schooling so if i know how much schooling every country has and i know how valuable it is how much it raises productivity then i can use that to aggregate and find a measure of human capital from schooling and i can in theory do [00:25:01] this for a whole variety of attributes and so that's what i want to build on first and show you first i like to show this one first because this is the way the literature actually developed and so it's a good way to kind of get a sense of the intellectual history of the [00:25:14] measurements of human capital so the way this literature started is from measuring years of schooling and i think the reason is kind of hopefully obvious which is the years of school years of schooling is [00:25:29] it's the most obvious it's the most visible it was the first thing that mincer put in the wage equations and talked about right it was right there from the beginning it's also one of the earliest things that we got data on and richard hit on this in his in his [00:25:42] first lecture data availability and the growth of data availability has been a fundamental driving force for macro development research so human caps right years of schooling is one of the very first things we have because it turns out that many countries [00:25:56] collect data on how many students or what share of young people are enrolled in school and knowing how many people enrolled in school if you just kind of add that up through a through a cumulative process you can figure out sort of the average [00:26:08] years of schooling in the population barrow and lee are kind of the most well-known people for providing these estimates and so that's what i've plotted here on the left this is simply average years of schooling in the population age [00:26:21] 15 or above from their data plotted against gdp per worker measured in 2017 us dollars from pen world tables 10. [00:26:29] both easy to get data sets online so what you can see is that there's enormous variation in years of schooling and this makes this a very likely and obvious culprit for including in our measures of human capital the richest countries in the world especially those that started getting [00:26:44] educated early they have something like 13 to 14 years of schooling on average in their populations the poorest countries in the world that might be as low as two years of schooling okay people don't talk about it a lot [00:26:58] but i think it is kind of interesting to look at the long run dynamics of this process and so uh lee and lee and one of these leads is the same as this lee built on the barrel and lead database but went back in time and measured or took the best possible estimates of [00:27:12] enrollment as far back as you could go and they argued that it was possible to measure educational attainment from 1870 onward for many countries so what i've shown you here is just sort of the long-run development of education which i like to do to give you a sense of the [00:27:26] historical perspective so you can see for example that part of the reason why the united states is very educated today is that it's been educated for a long time way back in 1870 it was already the most educated country in the world [00:27:39] you can see here japan has caught up fairly quickly it used to be relatively less educated and has grown quickly china is also becoming educated quickly the nice thing about china and one reason i want to show you this chart is that i want you to understand human [00:27:52] capital is very slow moving because it's human capital of the workforce and so if you start investing a lot building you know secondary schools building colleges paying people to go to college you know making tuition free all of those [00:28:07] policies you do that today you're going to affect the workforce but very slowly because the workforce includes some people who are 60 65 70 years old and so your 60 65 70 year olds will not [00:28:20] be highly educated for another 30 40 years and that's sort of what's slowing china down they're investing a huge amount in education but it takes a long time for this to play out and then i've included down here a couple of sub-saharan african countries just to make the point that [00:28:34] even within regions there's huge diversity of experiences so you can see here ghana has had phenomenal growth they're investing a ton in education and in fact they're up around china and brazil at this point whereas niger [00:28:48] is lagging badly and is way behind okay so this is all to say that education i think is sort of the most obvious most visible it's something we've been investing in a lot for a long time this is the first thing we're going to want to include in our measure of [00:29:01] human capital the second thing we need is an estimate of what's the value of education now right from the very beginning in the united states there was the literature argued that wages were [00:29:15] about log linear in schooling if that's true then the value of schooling is just the slope it's just sort of what's beta 1 for my regression before what's the return to schooling as people in the micro literature often call it [00:29:30] mincer argued from the very beginning that looked like something in the range of eight to ten percent was right for the united states that value seems to be fairly persistent over time we get fairly similar numbers now [00:29:44] what's nice about this is that as i said people have run regressions of this type in many countries around the world and so people have been collecting what is the return to schooling for countries around the world and that's what's plotted here on the right this is the data from the most [00:29:58] recent database i've tried to always include the source of the data and you can actually click all of these links if you have the slides open and it'll take you to the reference if you want to get any of this so this is a database that has collected these returns to schooling and i've just [00:30:13] plotted what is the return against gdp per worker you can see that there's substantial variation right you see some countries where it seems like the estimated return is two three four percent you can see some countries where it's like twenty thirty percent [00:30:27] what's maybe surprising but shows up persistently in studies that do this is that the return to schooling on average does not vary much between poor and rich countries sometimes you'll find it's a little higher in poor countries sometimes you [00:30:42] find it's the same but not that different so i think what many people have done is take this as being evidence of saying that the return to schooling around the world a good estimate would be eight to ten percent okay so this leads us to the classic estimate [00:31:01] of human capital i should have put some references here i'm sorry but this is what a classic estimate would be like holland jones 99 or caselli's handbook chapter from 2005. [00:31:11] their estimates of human capital that you get these very traditional estimates are built on years of schooling and in particular they're built on the observation that there are large differences in years of schooling across countries and it seems like a year of schooling [00:31:23] raises wages by eight to ten percent under those assumptions i described that would mean that a year of schooling makes a worker eight to ten percent more productive okay um here there's a bunch of references there's you know people have [00:31:37] been building these databases on the returns to schooling so i've given you a bunch of different references that i think all end up pointing in the same direction poor and rich countries have similar returns maybe poor countries it's a little higher and that'll get us to a measure of human [00:31:51] capital like what juliette gave on last friday which says that log wages and log human capital they're about linear in years of schooling so if i want to measure a country's human capital stock i'm going to say that the log of human capital is just 10 [00:32:06] times their average years of schooling and this is what people are in practice frequently doing okay if you do that just to give you some statistics for what i think is the classical measure of human capital [00:32:19] you'll find that human capital in the 90th percentile country divided by the 10th percentile country that differs by about a factor of two whereas the total income difference is about a factor of 19 between those two countries okay so human capital that this is [00:32:33] important you know i think we would all like to have twice as much human capital twice as much output that would be great but relative to the size of the thing we're trying to explain it's small and sort of that's you know here's the metric for that the 90 to 10 ratio of human capital [00:32:48] divided by gdp worker that's like 11 so we haven't explained that much and then people said well it seems like maybe human capital is not so important i prefer a little bit different development accounting metric everybody's got their favorite this is like one of these never-ending debates [00:33:03] i kind of like to look at the covariance of log human capital log output divided by the variance of log output but it really doesn't matter what metric you use this gets you 19 at the end of the day if this is your measure of human capital you're going to find it's not that important questions before i start showing you [00:33:20] more components all right i have a i have a question may i ask yep please so i have a question about the years of schooling so is there any way that [00:33:33] researchers include their the quality of schooling also give me two slides not for the basic thing but eventually people started to think about that that's the right answer all right so the second thing i want to [00:33:50] include and juliette did put this into her measure of human capital on friday is experience think about this as right from the very beginning minser argued that experience was an important component of human capital [00:34:03] more experienced workers have higher wages under the assumptions we've made that implies that more experienced workers are more productive up here i answered this question earlier i'm sorry i can't remember the names from flashing them in [00:34:16] the screen but what is experience really almost always that means potential experience we would love to measure real experience like how many years or hours or months have you been working to do that really you need panel data i [00:34:31] don't think anyone trusts the you know imagine asking you how many hours have you worked in your life for months or years you know the answer would be nonsense let's be honest i have no idea how many months or years i've worked in my life we wouldn't trust that answer [00:34:44] and we don't have panel data for poor countries so instead of using actual experience what almost everybody does is they measure what we call potential experience that's how many years has it been since you graduated the idea is that's when you really started working you could have really started working after you finished your [00:34:59] last degree all the time since then that's how much experience you could potentially have that's where the word potential comes from and then in practice that ends up being measured as something like age minus schooling minus six [00:35:13] okay if you measure experience that way you're going to find that the level of experience is really similar in poor rich countries and the reason is actually very simple very intuitive poor countries you know you are [00:35:26] developing countries have are younger they have very young populations they have young workforces and so they have lower ages on average among their workers but they go to school less so they have lower schooling if you construct experiences age minus [00:35:41] schooling minus six what you're going to find is they have lower ages they have less schooling but the experience they have is almost exactly the same as what rich countries have on average okay so that's not what's going to be different [00:35:54] it's not how much experience they have it's the value of that experience that we see in the labor market so what i'm plotting down here is the experience profiles for a select set of rich countries these are the five richest countries in [00:36:07] the database from this paper and then profiles for select poor countries which is the five poorest countries that were in the database for the same paper so the way you read these profiles let me just pick one to start with so this dash gray here the highest one is [00:36:22] germany you always estimate a profile so that it starts at zero and then what you're measuring is how much do wages go up as you get more experience and get older and so that's what's being plotted here when you get to five years of experience [00:36:36] in germany the wages have gone up by 50 percent as you get towards 15 years of experience their wages have gone up by a hundred percent and eventually in germany the highly experienced workers make about 120 percent more than brand new workers [00:36:50] okay you can see here united states canada france australia you're getting between 70 to 120 higher wages as you become experienced on average among rich countries you about double your earnings over your life [00:37:04] um just when they're when they're doing these wages what are they using for wage like is it like pre-tax post tax like how are they defining your wage that's going to make a huge difference for something like us versus [00:37:17] canada even like yeah so they as we i'm in here somewhere in the at all now but now it's embarrassing because i have to remember i think we're using pre-tax for most countries and the reason is that pre-tax [00:37:31] is what is measured in most countries yeah that would seem to that would make sense to me okay thanks yeah um on the right hand side we're doing the same thing we're measuring these same wage profiles but in a group of poor countries and you can see [00:37:46] in poorer countries earnings wages go up much less over the course of your life so you know older and experienced workers make maybe 30 more than the youngest workers in those countries so we can show this for as many [00:38:02] countries as we can get our hands on so what i'm going to show you now is what is the height of this profile what's the value to having between 20 and 24 years of experience that's what's plotted on the y-axis so basically it's the height of that profile at 20-24 years [00:38:16] plotted against gdp per worker and so you can see that that what i just showed you is a fairly persistent finding are richer countries wages go up a lot more over your life as you get more experience and in the poorer countries [00:38:30] they go up a lot less and you can see here you know bangladesh vietnam india guatemala earnings are going up by wages go up by about 20 as you gain experience over your life not very much now in practice you have to be a little [00:38:46] careful about saying that that's definitely human capital right there's many things that could be going on in in in the paper we we talk a lot about other hypotheses of what could be happening here i think a lot of ancillary evidence has [00:39:00] come together to lead people to conclude that this probably is human capital that workers are developing more human capital over their lives in rich countries than in poor countries sort of a recent piece of evidence that i like a lot that points in this direction [00:39:14] comes from a paper by ma it's all it's brand new it might even be 2021 at this point which looks at how much training workers receive over their lives and so what they do is they collect data from a wide range of countries on [00:39:28] the share of workers who get training from their employer and they're plotting what share of workers get training against gdp per worker you can see there's a humongous gradient here in poor countries basically no one gets training from their employers and in which countries maybe half of the [00:39:42] workforce typically gets training from an employer that's a function of two things one is that in poor countries many people are self-employed you can't get training from your employer if you're self-employed i think that's still fair though in the sense at least to me it seems like [00:39:57] i think it's probably legitimate to think that a lot of self-employed workers aren't learning much as they do their jobs i mean who would they learn it from who would teach them anything the second thing of course though is that even for people who work for employers who work for wages are more [00:40:12] likely to get training from their employer in rich countries so i think about this as being one likely culprit for why we see wages rising more over their life maybe workers are getting more training they're doing more investments they're learning more over their lives they actually provide a quantitative [00:40:26] theory that suggests this is probably a big component of it if we take oh sorry if we take this as reflecting workers learning more over their lives and rich than poor countries we want to add that into our measure of [00:40:38] human capital and i'll do that in a second do these results change dramatically at all if you separate it out between like men and women for example have you done that [00:40:53] a little bit but not much surprisingly women have lower returns to experience in almost every country but that is mostly because the potential experience is a less good measure of [00:41:07] experience for women than for men men work almost continuously since they graduate women have many more interruptions over the course of their lives right yeah true yeah exactly all right yeah can i ask a question about the uh [00:41:22] about this graph actually um i guess is it like you you say that it's not exactly well no sorry uh let me let me reword it [00:41:36] i guess are we could it also possibly be that just the there's kind of like a cap to the capacity of a country to use experience of a worker where like in the more rich countries [00:41:51] having more experience actually lets you move up like a wrong or a ladder or something but in a poorer country having more experience doesn't really do anything for you if there's less capacity for the [00:42:04] country's like uh like tfp to kind of use that experience so i guess i'm like years of experience maybe isn't completely translatable across countries [00:42:19] i think that the the paper that we had to read kind of talks about this a little bit the one that you did with hendrix sorry if that was very vague no no i understand exactly what you're driving yet [00:42:32] um it's a good question and it's a deep one so if i go back a little bit really this goes back to the production function if you look at this production function it says look [00:42:47] there's constant but decreasing returns to capital constant and decreasing returns to human capital they're completely separable and there's really no interesting interactions between these factors here if i double your human capital in the united states you get 2 raised to the 1 [00:43:01] minus alpha output if i double your human capital vietnam you get 2 raised to the 1 minus alpha extra output it's the same everywhere and then i think i could give you a list of papers and especially it's a common thing to see in the conclusion of a paper to say [00:43:16] yeah you know it's really interesting what if there's an important interaction between the technology we have and our ability to use schooling or experience that's an interesting idea for future research and that's kind of where it always lingers okay but that's [00:43:30] that's really where this point right here is where you can really see why why i'm not talking about ideas like that it's embedded in the production function right from the very beginning thank you all right i don't want to bore you to [00:43:46] death by showing you a million of these things so there's a couple more things that people are going to keep track of time too there's a couple more things that people have looked at including education quality and health would be the two other things you might think about adding into this measure so on the left hand side here i have [00:44:00] plotted a measure of education quality which is math scores on the pisa exam which is a very common and very popular exam if your country takes it you'll know about it because they'll report the report they'll report the scores in the newspaper and complain about how you're [00:44:15] not doing well enough on it unless you're in singapore or finland where they do great so what's plotted here is the score on math in 2018 and you can see yes people in poor countries score much worse on the math exam [00:44:28] health is probably the trickiest thing to measure and again it's like even in the united states you know how do i measure the health of two people that's a very difficult thing so here's one proxy for health adult survival rate that's the [00:44:42] probability of the 15 year old will live to 60. [00:44:46] and you can see that in rich countries most 15 year olds will live to 60. but if you go to poor countries you know on average maybe only 70 80 percent do we think that's probably correlated with bad health with the prevalence of [00:44:59] diseases with their inability to get enough nutrition and to afford proper medicines you know it's not ideal measure but that's what we have to work with you can see though that as we go along in this constructive method [00:45:13] you know what's the value of 50 points of ps of math score for productivity now we're starting to get this is getting a little tricky this gets a little harder and i'm going to show you the papers that have done this in a second but this is more challenging what's the [00:45:28] value in terms of productivity of living in a country where you have 0.1 higher adult survival rate okay people continue to build on those minster wage equations and they have you know approaches and tricks in back of the envelope calculations to make that [00:45:42] happen so let me show you what those results are what i'm going to do here is accumulate one step at a time a measure of human capital so i'm going to construct it literally as the literature has done over time and i'm going to ask how important is it [00:45:56] for cross-country income differences using this covariance metric so i told you before if you measure human capital as years of schooling you're going to you're going to explain or account for about 19 percent of income differences if you add quality of schooling sort of [00:46:13] a typical estimate is that that's actually about as important as the years of schooling for cross-country income differences that'll get you to about 38 and i don't know if this is coincidence or not but when you add experience you almost always get about another 18 to 19 [00:46:28] percent here it's 18 so this is 56 health the leading estimates of health are that it's much less important which i don't know i just always feel like it's important but anyway i have no i have no evidence i've never done any [00:46:42] research i feel like it should be more important than that but i i've never proved it and someone said that schooling experience health we're up to maybe like 59 of cross country more than half of cross-country income differences are due to health if we add it up this way here's your list of references if you [00:46:57] click on them they'll take you to the right papers to see see where these estimates come from being part of health is accounted for indirectly through like schooling experience like if you don't have good health then you're not going to school [00:47:11] yep give me one second that's the next slide it's exactly i'm going to summarize and that's exactly what i'm concerned about so this first approach which i'm going to be done with now uh measures cross-country human capital differences constructively like [00:47:24] literally constructing building it piece by piece you account for differences in stocks or rates of return and those resulting estimates are large maybe 60 percent of income differences but you know i think you should have some concerns at this point you should not be [00:47:38] satisfied or happy with what we've done one thing is you have to be worried about those assumptions so here's a reference to the first paper that starts to rethink those assumptions we use for development accounting there's a whole literature that builds from him but he was the first [00:47:52] the other two concerns just kind of conceptually it's hard to be exhausted what i mean by exhaustive is do i think we've accounted for or measured all of the different pieces of human capital probably not right i mean we added up [00:48:06] years of schooling quality of schooling experience health if there's anything else that's not in there so we might imagine this number could be even bigger but at the same time and this is what is it jolene is that who that was yeah okay good um we may be double [00:48:20] counting here right because health might very well affect the quality of schooling that a country has if the students have better health they may do better on the tests and that may show up as better quality of schooling likewise you know being in school a long [00:48:35] time may help you learn more on your job so when i add up the value of schooling and the value of experience i may have counted something twice because some of the value you having experienced may have come from your schooling and so there's a concern here that by adding things up this way i may be [00:48:49] misattributing miscounting or double counting things that's going to lead me to my second approach which i'm going to call the deductive approach so are there any questions before i meet [00:49:02] um excuse me yeah hello prof yep somebody go hey hello i hear you yeah look [00:49:16] yeah hello i have a little bit of question respect to a recent port contrary especially technological uses and that of that with the taste so in terms of measuring uh the human capital specifically in most of the developing countries like [00:49:31] you gave an example like in ghana and other different countries and etcetera don't you think that the taxonomy might be perhaps determined more easily based on the technological uses that the people have within the country they say because i think uh with the use [00:49:45] of technology the technical know-how especially in this uh environment of uh intensive uh economic that we are in so with all i mean uh the technical uses uh how do you think that poor control regardless of violence [00:49:59] especially in the cultural project in terms of converging bigger convergence uh it is possible whenever poor countries are trying to educate the populations because i think there is always a peak in most of the western nations especially developed country [00:50:13] and as a result agenda is always an opportunity for the poor country to catch up because they're convergent but how would they catch up when they in terms of the transfer of technology rnd i mean research and development is not advanced [00:50:27] and the taxonomy issues also has a little bit of gap so with respect to the quality of education so what are some of the research you think that may be appropriate to ensure that an advancement or most of the least developing country can able to catch up [00:50:41] more faster okay that was a lot to chew on um i think i heard two distinct questions and you can let me know if i missed something one thing i heard was um a general [00:50:55] question about the interaction between human capital and technology yeah and i think my answer is going to be the same as the answer i gave a minute ago which is that if you look right here technology is a part of tfp here's human [00:51:09] capital if you look at this you'll see that there are no really interesting interactions between these two things having more technologies good having more human capital is good but there's not really there's no idea that you have to have [00:51:23] technology to make use of skills there's no idea that having skills makes technologies more valuable in everything i've done so far that's been missing if you send me an email i'll send you i mean there is very recent work that's trying to investigate these things like how much [00:51:37] does technology aid the use of human capital and vice versa that's really frontier work at this point so i would say it's not very subtle that's why i didn't want to teach it but people are thinking about it the second piece was you know if [00:51:51] education quality is important how do i get more of it i think that's the way i would interpret that question the uncomfortable and sad truth about this literature is that no one knows [00:52:04] i can point you to some books and some papers and reports that's not really something that what's a polite way to say i'm somebody who's worked on this and i'll say it about myself that's not something my research has [00:52:18] ever been very good at i can tell you that it's important but if you ask me like at a nitty gritty level what should i do in our schools to make sure that they actually get this quality that seems so important my papers are completely useless on that part that's the truth [00:52:33] um i can point you to some manuals and like if you read the pisa reports they'll give you some ideas or best practices of countries i don't know how reliable those are but that's the best that's out there truthfully because what made me to have asked the question is simply because i'm also [00:52:48] writing my uh this irritation solution on human capital development it will develop in developing countries so but uh i came across some uh i mean some issue with respect to the i mean the country productivity based on mainframe [00:53:02] that stays to this stage that is to say from the before the turns after another advancement like in asia example like in korea like people select taiwan you know they started with uh radio activities like a [00:53:15] radio receiver and all the different things you know before they try to advance but in the same vein and in certain countries in asia and so like malaysia where i'm stationed currently and basically china they were not advanced and typically [00:53:29] compared to other countries like ghana was empowered with other korea by the end of the day ghana left behind malaysia has advanced korea has very intense technology i believe that technology i mean observe like a sort of leaf rock [00:53:43] of the economy like in india okay because of the incremental human capital in india i think they tried to refrog we're in it of living over world today so if maybe a technology vis-a-vis human capital a main [00:53:58] improvement or advancement per se can can be uh somewhat depressed let's say because we have different processes in terms of human capital such as f as is one of the deposited percent and income is one of them but can we measure the gdp per capita [00:54:11] based on the individual percent based on the popular educated population within a country or based on the input of the country the total production part of a dpf so total factor of production tpf [00:54:26] tft runner sorry i don't think i understood that question yeah what i try to say is that can we measure the country gdp [00:54:39] i think we're going to have to go on i'm sorry okay okay if asked me again during the during the follow-up session i'm just not quite sure i'm grasping it um [00:54:52] can i make a quick question yes please um so is there a sense here um we're making a causal statement somehow so we're saying human capital causes gdp to be larger but it can also be true that um richer [00:55:07] countries have like more opportunities to increase human capital right because it's not only valuable for develop as a productive resource and thinking in cases such as latin america where [00:55:20] like recently the like the fiscal government like the the country governments make a huge effort to increase education in all levels and it was only possible because the countries were a bit richer than before or they [00:55:34] experience some kind of commodity bloom etc so right from the beginning actually this has been an obsession in this literature and if you go back and read bills in greenhouse paper the people who worked out these assumptions that we're using you'll see that that was exactly what [00:55:48] they were wondering which is does getting more schooling make you rich or does getting richer lead you to get more schooling what i'm going to say today and the way i try to phrase these slides i sometimes mess up on this is we always try to say that [00:56:00] human capital accounts for a lot of cross-country income differences and then that's a fancy way of saying we don't know where the schooling comes from it could be that tfp leads people to get more schooling and leads them to get richer i'm not going [00:56:15] to be able to make a statement that's causal i'm trying not to make any causal statements it just seems like there are really large differences across countries in human capital and we want to think about that more all right so the second approach uh that [00:56:31] i kind of want to investigate a little bit is going to be called the deductive approach and this is built on the idea that human capital you know by my definition at least is embedded in people and so what we're going to do is we're [00:56:43] going to utilize the fact that migrants of which there are many carry their human capital to new countries when they move okay and so this is going to allow us the opportunity to deduce the importance of human capital what's different about this is that [00:56:57] we're not going to try to specify what human capital is or to build it up piece by piece we're just going to say whatever human capital is the migrant has it when they move they take it to the new country let's see how important it is in that new country [00:57:11] so if you remember you know the wages we specified under those maintained assumptions were that the log wage of a worker with human capital i in country c is additive and a country effect and a human capital effect and so the way i want you to think about [00:57:25] this is well if i have a worker who migrates they take that human capital they used to have a country with this z they move to a country with a new z and that's going to give me some opportunity to see how they do even more intuitively the way i like to [00:57:38] think about this is in general what would be really cool is if we could take the united states and give the united states the human capital investments of ghana and then say how does that affect the united states now of course we don't get you know we're macro economies we don't [00:57:53] get to do randomized trials on entire countries but an immigrant gives us the best shot we have because we can see somebody with ghanaian human capital move to the united states and see how they do and that's kind of the underlying idea behind this exercise [00:58:08] now of course there's a trade-off nothing comes for free in economics since we're going to be talking about migrants we have to deal with two additional concerns that are just going to be pervasive throughout this literature that thinks about migrants and human capital which is that the migrants of course are not randomly chosen you have to worry [00:58:22] about the fact that they're selected they may be different than the random ghanaian and the second is that their human capital may not be fully portable right if you look at this equation what i'm thinking to myself is well if they take this human capital to another country what happens [00:58:37] but of course the skills and the knowledge they have may not be as useful in the new country they may be the wrong type they may be discriminated against all of those types of concerns have to be dealt with okay so there's been a at the bottom here i talked i gave you [00:58:52] some references for other papers there's a there's a sort of a small literature but a long one that long has been going on for about 20 years that investigates this idea the approach i'm going to show you today and that i recommended for the reading was to think about what are the wage [00:59:06] gains at migration okay so the idea here is imagine that i could see the wage of a worker who moves from c to c prime both after he migrates and before he migrates [00:59:21] if i go back and take this wage equation and just take the first difference of this just difference across two countries that worker has the same human capital in both countries so that's going to drop out and what i'm going to be left with is just the z the difference in z's between the [00:59:35] two countries but that's exactly one of the two things i wanted to understand from my development accounting exercise which is how big is the z difference across countries so this would be a really easy way to [00:59:49] quantify one piece of the development accounting puzzle the underlying intuition here i like to just say this like plainly suppose i have a worker who migrates from a poor to a rich country let's say the rich country is 10 times richer and [01:00:02] make everything easy if that workers wages went up by a factor of 10 i would say wow switching countries fixed everything it made this worker much richer it must be that country is what matters something about the institutions the [01:00:17] capital the infrastructure i don't know but changing countries fixed everything i'm going to conclude that z must be what matters for cross-country income differences on the other hand suppose that that worker switches countries and their wages don't move at all they stay really low [01:00:32] well then i would say boy the country didn't do anything the country must not really matter much for the wage this worker receives boy it seems like their low human capital must explain why they have such low wages now you'll notice that that really [01:00:45] hinges crucially on thinking about who these workers are and whether their skills fully transfer and in the paper i'm going to try to get through this relatively quickly so i can show you a little bit of models in the paper we do try to do some exercises to think about how big those effects are but i want to kind of get to [01:01:00] the main results to show you what's different about this let me emphasize this one more time this is almost like a completely different way to come at the same problem because here i'm not going to try to specify what human capital is [01:01:14] i'm not going to try to specify how they got it or what attributes of it are important i'm just going to say whatever they have they're going to take it to two different countries and that's going to let me to think about how big the z differences are that's what i mean by deductive it's like i'm going to deduce based on [01:01:29] how much their wages change how different human capital must be you need the the kind of one tricky thing for this is that you need data where you know how much workers how much wages workers made before and after they migrate [01:01:44] most data sets don't have that so here's a couple of data sets we found that did that's what we're going to use to quantify this and all we do in this paper is just convert wages to ppp adjusted wages and then we're just going to ask how [01:01:56] much do ppp adjusted wages change relative to pppgd people worker gaps that's the notion of how much does your wage go up how much is the country matter and then what i've highlighted here is [01:02:11] sort of the main results from that paper what we find is that workers from countries that are less than a sixteenth as rich as the united states so these are the very poorest countries immigrants come from to the united states when they move they get about three times higher wages the gdp gap between those countries in [01:02:28] the united states on average is a factor of 32. [01:02:32] so switching countries makes them richer right their wages go up by a factor of three but that's a small amount relative to the total gdp gap and so when you do it you'll find that that would imply that human capital accounts for about two-thirds of [01:02:44] cross-country income differences you can do the same thing for different groups we have two different data sets but as a whole when we do this we consistently find that workers wages go up by [01:02:56] a small amount relative to the gdp gap and that that would imply that human capital accounts for something like 60 of income differences 50 to 60 percent so the deductive approach i think has [01:03:16] strengths and weaknesses relative to the constructive approach it's exhaustive because it measures whatever human capital is wherever it comes from but it's not constructive and so somebody already asked me you know if education quality is so important how do i get it you can see that this is even [01:03:30] worse now right this one is like if human capital is so important where do i get it from you know this would really not be very helpful for that it avoids double counting it does require these additional assumptions about migrants and i'm going to refer [01:03:44] you to the paper if you want to ask questions you know that's it honestly it takes a whole seminar but i'm going to present the paper and really talk through all of that it's it's anyway what i like about it is that although it comes from a completely different direction you end up getting fairly similar results you get the human capital [01:03:59] accounts for half to maybe 60 percent of income differences okay and so we have two different ways of getting to the same point both of us lead to think that human capital is really big i still think there's room to explore on the measurement side [01:04:13] sort of my pet peeve with this whole literature is that we're really talking about measurable stuff we're not really talking much about skills i would really love to know more about what exact skills are missing in poor country like they don't go to school as long and the schools aren't as good so they don't [01:04:28] know how to do what what specifically is it that they miss when they go to the job i've never had any sense that i know what that is i would love to know more about that anyway i think there's more to be done on the measurement side what i would like to do now let me see how am i doing on time [01:04:43] yeah about what i expected what i'd like to do now is show you a little bit of models about human capital differences and the idea here is if you are convinced or at least intrigued to think that human capital might be vary a lot across countries and it might be a big piece of the [01:04:57] development puzzle and also of the differences in across sectors during structural transformation you'd want to build that into your model how do you do that what is like the off-the-shelf simplest model you would start with so i'm going to give you a very simple [01:05:12] version of these models and then give you references if you want to go and dig in and do the fuller bigger versions of them all all right so the goal here is mostly i [01:05:28] first and foremost i want to illustrate the mechanisms that can help you explain these large human capital differences all right and the way you want to think about this is human capital as i've been emphasizing is an investment that people make people in rich countries are investing [01:05:41] more that accounts for a lot of cross-country income differences but it doesn't explain it to explain it i would want to know why why do they invest more so what i want to do is give you some models i'm going to start with a benchmark model that can't explain it which is a [01:05:56] useful result like oh if i do that i'm not going to get anywhere then how do i tweak that model to explain some of the differences all right this is the simplest way i could think of to write this model down so let's start with the firm it's a [01:06:10] representative form so you can think of it as a bunch of firms if you want the firm is exactly what i showed you before nothing new here they operate this cobb douglas production function it's the same one i showed you i'm going to insist on the same two [01:06:23] assumptions perfect substitution competitive labor markets that's going to give me the same wage equation i had before nothing new on the firm side what i want to do now is add a model of the worker [01:06:38] really the person who's making decisions about investments and think about what would lead their decision about how much to invest to vary across countries so here's the simplest model this really is a dumbed down version of becker's [01:06:53] book so let's have a consumer who lives for one unit of time and they're going to maximize utility and the utility is just i'm going to integrate over that period i have log preferences i have consumption of t [01:07:07] at every instant t during that one unit of time and then they're going to decide their consumption profile the other choice they're going to get to make is how long they go to school before they start working [01:07:20] so the simplest version possible of this is to say all right they're going to attend school for a length of time s schooling develops human capital so when they leave school they're gonna have human capital h of s the remaining unit of time one minus s [01:07:35] they're going to go to work and they're gonna earn a wage then their budget constraint simply says that the integral of their consumption has to be equal to their income right the money to finance what they consume has to come out of their income [01:07:50] and then their income piece by piece this is the z from the wage equation so literally i'm just copying the wage equation to start with right i'm plugging in their human capital that's a function of [01:08:03] how long they go to school and then how much do they get to work well only one minus s of their life the rest of their time they spent in school if this is your model of the human capital choice you're going to get two [01:08:17] main results this is this is well known from becker's book the first result you're going to get is that workers in this model or people really want to smooth their consumption over their life in this model i've assumed that people don't discount across the period [01:08:31] or within the period so their consumption is exactly the same over their life they basically spend their income smoothly over the course of their life the second result that you're going to get is what's the optimal school choice [01:08:44] so just take the first order condition and it really naturally gives you an equation that says the marginal benefit is equal to the marginal cost what's the marginal benefit of going to school well this first piece here [01:08:58] is the wage you get for working one minus s periods the benefit says well i have a little more human capital if i go to school a little bit longer i'm going to have marginally more human capital that's going to marginally increase my earnings what's the cost of getting a little more [01:09:13] schooling well this is the wage i could have earned during that instant right i could have made h of s instead but i spent that extra instant going to school and getting more human capital in the future that's the trade-off that a worker faces in this model should i get a little more [01:09:28] human capital and get that benefit over the remaining one minus s or should i just go to work and enjoy extra earnings for that instant now what's really neat about this model [01:09:41] uh is that if you look at it carefully the one minus alphas and the zcs cancel they're on both sides of the equation and so that tells me that the optimal school choice in this model is independent of z this [01:09:55] is an idea that shows up in becker's book already i've always called this becker neutrality from our perspective this is a problem right we're trying to understand why people in different countries are investing different amounts in schooling [01:10:09] and there's nothing about the country left so this is a theory that gets us nowhere i like it because it's a benchmark it says all right if i start with this model i can see that i'm going to go astray so i need to tweak something here i'm going to give you some ideas about what to [01:10:23] tweak in the next few minutes but i want you to understand the intuition for this because i think it's really powerful so what's going on here imagine i live in a country with a better government or better institutions or technology or whatever your favorite [01:10:37] theory of tfp is i think most of us kind of intuitively would say ah well if there's a better technology it makes it more beneficial to invest in human capital right but if you look up here what this tells us is that yes the benefits of human [01:10:52] capital investment go up right because i'm going to earn more off my human capital investment but in this model the other force that shows up is that if i live in a country with a higher tfp better technology whatever it is that also raises the cost because i [01:11:06] could have earned higher wages during the time i spent in school and those two things exactly cancel okay and so the underlying idea here is that if time my time [01:11:20] is the main investment i put into human capital then the opportunity cost of my time is proportional to z the benefit of my time is proportional to z that's not going to get me very far [01:11:35] all right so now how can we tweak this and there's going to be i think i have time to do all three let me check my time one last time might have to cut one off there's three main ways that the literature is used maybe four to tweak this model so we can start to get some actions we can start to get [01:11:49] cross-country differences in human capital and so if you're building a model of human capital i think you want to make sure you have some of these elements one is borrowing constraints go back and look at what i said maybe you didn't notice it [01:12:03] it sneaks by people a lot what agents in this model want to do is they want to consume smoothly over their life including while they're in school where does the money for that come from [01:12:16] well the way this model is written in the bar in the budget constraint you're allowed to borrow as much as you want there's perfect credit markets so the first thing you can do is say well [01:12:29] i'm not so sure about that i think there's actually probably a lot of limits to the ability of children and their parents to borrow and finance these investments and i've listed here some of the major impediments you know in general i think most of us know that credit markets [01:12:44] aren't perfect that it's hard for people to borrow large sums of money to finance all of their consumption and expenditures on schooling particularly since we don't let children write contracts we don't let them sign binding contracts [01:12:57] right hi i'm 14 i'd like to borrow the money so i can go to heist to a really good high school to benefit myself over my life you know we don't let people sign a contract that agrees that they'll repay that you might be able to work around that if [01:13:11] you think oh well the parents can invest on the behalf of the children but that requires that parents be really altruistic towards their children okay and so if you go and read becker and tones this is a beautiful paper in the journal of labor economics that just works through exactly what you would [01:13:26] need in terms of altruism or credit markets to get the socially optimal investment and they're pretty tough restrictions for our purposes since we're trying to think about cross-country human capital [01:13:39] differences the main idea you would want is maybe people in poor countries have a harder time borrowing and that seems like a pretty natural restriction if they have a harder time borrowing in poor countries because the credit markets aren't as good or because their parents can't [01:13:54] borrow on their behalf or because you know they're they're they're near a subsistence constraint consumption is really low any of those types of mechanisms um then what you're going to get is that people in poor countries are going to have less schooling because they literally just [01:14:07] can't borrow enough that's one force that you could put in the model and hello and um i think perhaps it's also impossible to [01:14:19] hedge against unemployment after getting the education yes um or more generally there's a there's a literature um tom krebs [01:14:33] is the big reference for this about human capital risk that you can't there's no way to hedge against risk we often talk about human capital as if go to college in the united states and you get sixty percent higher earnings but of course there's a huge thing [01:14:45] yeah okay yeah yeah but there is no possibility of putting that in this model for example yeah no i mean like i said if you go read krebs uh krebs and write [01:14:59] send me an email and i'll send you the exact papers people have put these into models nobody in macro development that i know of so there's a literature in the united states for example that says you know americans might not be willing to borrow for college because of this risk [01:15:13] i don't know anybody who's taken that across countries yet i think that's kind of the current state of the literature that would be an interesting idea yeah because the and this is quite different if you look at the trend of what is happening in many developing countries in developing countries there is a [01:15:28] growing uh commercialization of education especially at high school level people spend a lot of money acquiring education and and usually without credit and i i really have not really seen [01:15:45] so much in terms of return to that uh such investment because this is an environment where you have very high unemployment rate and all of that uh so i i i i like to get the [01:15:58] paper maybe that way it's a proposition that if maybe governments will be able to put in some better policy if it is possible to hedge against unemployment after making those investments in human capital [01:16:12] yeah yeah if you send me an email my name is my email address basically my name at gmail i'm easy to find okay i'll send you that paper i want to say one thing real quickly though there's a distinction between two things one is just risk [01:16:26] i don't know if i'll find a job i don't know if the job will pay i don't know if my degree is valuable there's a great literature on that there's a second phenomenon which is kind of strange in some developing countries where the college graduate unemployment rate might be 50 [01:16:40] yeah that's not risk that's like a misallocation i don't think anybody has i don't know i've not seen a good i've always been confused like crazy about that and i don't have any good models why that would be [01:16:55] so anyway that would be so if somebody could explain to me just at a really basic level what's going on there i have no idea you you're right you're right i'm i'm worried about that too because that's that's actually that is the reality in many developing countries [01:17:07] yeah thank you yeah hello thank you yep yeah i wanted to take on what my colleague just explained now with respect to uh the valid return of education especially territory institutions uh i think uh uh [01:17:21] patronus also spoke on this but it should be for about the attire has been debunked that i can remember in the year 2000 there was a conference head over there during the dental uh united nations ahead of kofi anna they [01:17:36] have a conference i mean summit in senegal with respect to that of the international monetary fund and the world bank and that of the uh the other sister organization the state would give the whole [01:17:49] then they have a conference specifically especially in africa most of the developing countries they they don't invest technically in education especially high in education like the other second rate and and and and other for the and other territory institutions [01:18:03] simply because while they believe that they return to education it must grow lesser especially for high school and uh and university or college in college education but then most often judging from the what have been experienced in [01:18:17] africa especially most of the african countries developing countries is that they invest most of things in the infrastructure development attack of social capital where rahima capital are left behind but in terms of purchasing other different mineral research i mean other [01:18:31] different materials that sort of constructions and maybe bringing expertise down to come and walk to come and educate the people over there it is much more like great and expensive more than that of developing nations so if at all quality education has no more i mean return in terms of [01:18:46] investment without an investment then technically speaking i'm not quite sure that there will be an advancement in what's in this african conscious because with all the progressions of human capital in a country it is difficult take this as a case study how you expect somebody to construct a [01:19:00] bridge or breaches tonight and other different heroes without having the expertise so without advance knowledge of education would be quite more difficult so the quality of education is one of the major problems that we have in most of the african countries [01:19:12] especially specifically most of the african country higher institutions they are embedded with political i mean infiltrations instead of institution to be independent of their own and crafting uh uh let's just say uh [01:19:26] institutional policy based on state-of-the-art technology or based on state or other educations so the major problem is the institutions so institutional failures are one of the major urban problems that we have in most of the african country thanks [01:19:39] thank you let me try to put two more mechanisms in and then i'll stop and take some questions i'm sorry i know i'm running a little late but i'll stay as late as anybody wants to ask me questions [01:19:53] there's two other ideas that float out there one is to let another investment matter for human capital production and one that's been shown to work really well this this dates back to manueli and sasha is to let people buy some sort of goods [01:20:07] to be used in the production of human capital and goods here you want to think about something that's produced so books chalkboards desks computers paper pens etc if you do that you get a problem that [01:20:21] looks like this now the budget constraint says that consumption over my life plus e that's my expenditures on the goods has to be equal to what i earn and on the right hand side the only thing that's changed is that my human capital [01:20:34] depends on how long i spend in school and how much i earn if you specialize to the power human capital production function which in practice everybody does then you can show very simply i've got the derivations in the back i'm not [01:20:49] going to do them today this does not fix the problem but how long you go to school is the same in all countries but people in richer countries or higher tfp countries will spend more on school and so the way [01:21:03] the literature thinks about this is that that could give you more education quality right you have more books chalkboards better teachers or something like that that will give you more expenditures and then you can show that we'll also raise human capital [01:21:17] okay so one trick you can do is just to say it's not only my time that i invest i also have to invest resources and the underlying intuition i'm not going to have time to go through all the math the underlying intuition here is simply that rich countries [01:21:30] computers are cheap and so if you can use computers to buy education quality people in which countries will naturally buy more computers and they will get more education quality and they'll get more human capital this is probably the most widely used mechanism right now [01:21:45] if you want a model that gets you cross-country human capital differences this is sort of what's most commonly used at the moment the last thing you can do is let the government matter through something that looks like public [01:21:59] education and the way the public education works is imagine that the government also provides an input to human capital there's many ways to write this down and i'm running out of time so i'll just say quickly the basic idea here is that the government can matter either if it [01:22:12] provides so much resources that it's basically you know getting families to invest more than they would have or if the government provides a complementary input to human capital production so there are a smaller set of papers [01:22:26] that do this there are a few other there are a few other ideas out there and the one thing i want to say before i conclude i think the big open thing here most of this and most of what i've talked about and really most of the human capital literature in terms of modeling has been [01:22:41] aggregate and so especially for a group like stag i think something that's really interesting is that these issues of skill bias technical change where what you the value of your human capital depends on the state of technology [01:22:55] or also structural trans structural transformation right maybe the value of your human capital depends on what sectors are growing versus shrinking right maybe human capital is more valuable in a non-agricultural society than an agricultural society [01:23:09] these are ideas that i think still could use to be explored more these are things that could be built into models of human capital across countries i think basically i should stop there i'm way over time as it is and i promised i would take questions so let me call it a conclusion those are ## Q&A (01:23:23 – 01:40:51) [01:23:23] the two themes somebody else talk uh i have a question about probably the [01:23:37] very last point uh on structural change so i i didn't see much on the demand side uh meaning if there are no for instance let's say firms large firms like in western countries demanding for skills [01:23:52] um i i see why it would be difficult for people in a developing country because there's no return to such investment you see so i wonder whether for this kind of [01:24:06] models this is something that has been implemented if it's more like general equilibrium kind of thing structural change and so on and so forth um so i've seen two related ideas this is a [01:24:22] great question i think it's something that's still under explored there's two related ideas that float out there one is you can write down a model where skilled and unskilled labor are imperfect substitutes like you would do in most labor [01:24:36] models at this point and then you can have something that's called skill bias technical change right so that that benefits the skilled labor relative to the unskilled labor and then in that model there is an idea that whether you want to become skilled [01:24:50] depends on how much skill bias technical change there has been and so i could point you to a few papers that are thinking about the world that way i think that's the most fully worked out model that i've seen that has these ideas [01:25:04] a related thing that's out there i think many people have known and shown empirically that the value of skills like the return to a year of schooling is much higher in parts of services like business services there's a really high return to skill [01:25:19] it's really low in agriculture so i think there's probably an interplay there which is that as an economy experiences structural transformation as agriculture shrinks and the skilled service sector grows there's probably a rising demand for for education and that might lead [01:25:34] workers to get more education um joe do you have the full model of that would you say i don't know he's probably not listening but joe has the closest model i've seen to that but i don't think he has all of that fully worked out i think there's still some room for that idea [01:25:48] yeah i agree that there's room i actually there was a related question in the q a and i answered i think that there's an area for research yeah i'm i'm really i mean i'm surprised i haven't done it i'm surprised nobody's [01:26:02] done it i'm surprised you haven't done i mean it's been kind of floating in the work but i have not seen anybody fully work it out i think it would be great thank you very much i mean i don't have a full model [01:26:16] part of my research is about liberalization of services and how this uh fosters human capital accumulation because this attracts uh new firms uh from foreign firms or private sector [01:26:31] firms and this raises the returns to education i mean i've not seen anybody even build that in i agree that seems like a natural starting point right like we know you can just go get i've done this you [01:26:45] can go get minster wage equations and just show that schooling is twice as valuable in some of the service sectors than it is in agriculture then it's obvious that i mean that seems like something you could just build into the model pretty easily that at least right that [01:26:58] as service sectors grow because of liberalization structures that's what you're talking about right yeah yeah that seems like a great starting point at least i i hope people take these ideas up alicia you had a question about degrees [01:27:11] versus years of school yes thank you professor and thank you for your lecture i have a question and in the first approach we treat the workers with and without a degree [01:27:24] basically the same but in real life with like because firms only if some workers with certain degree or higher which could cause their wage differences and their [01:27:38] human capital accumulation in the future so i think like the degree might matters in in a way i'm wondering do we have evidence in literature shows that [01:27:51] the years of schooling matters more than the certain degrees yep great that's a great question so i'll give you two answers one is if you want to just read about this a little bit it has the weirdest name in the world [01:28:04] the literature is called the literature on sheepskin effects i'm so sorry i don't know why they decided that was a good idea but if you search for the phrase sheepskin effects like literally skin of sheep you're going to find the papers you're looking for and that's the literature which asks [01:28:22] exactly the question you have in mind which is if i know you went to 16 years of school but you didn't complete college versus whether you did does that still make a difference for your wages and the answer is yes the answer is it makes a pretty big difference and so people have investigated that in a wide variety of [01:28:37] ways and then they've talked about whether that has something to do with selection or signaling there's a whole literature which is honestly totally different from what i talked about today at a practical level with people in macro development the [01:28:50] papers i'm talking about today including my own really in most countries data sets they don't ask how long you went to school they ask what degree you have like do you have a college degree and then what most people including me do is we say well college takes 16 years [01:29:04] and we just write 16 years of schooling and then we run the linear wage equation okay so the idea that you need a certain degree or that this degree is more valuable than that one that's just [01:29:18] ignored that's the that's the that's the best phrase i can give you for it thank you you're welcome the honest truth is signaling is really tough for human capital people right [01:29:33] yeah and when i say degree not only like the degrees also could be the graduate high school and the middle school like the all matters especially for developing countries yeah [01:29:48] yep yep i agree completely and like i said it's something that i don't think anybody's really taken into account yet i've not seen it todd there's another great question in [01:30:01] the q a q a a becker type question from emmanuel leclaire when we think about borrowing constraints is there also view that a firm may not want to invest in developing technology that supports high skilled individuals [01:30:14] because human capital is intangible so not collateralizable and imperfect financial markets may lead firms to invest in technology that promotes the use of collateralizable capital instead [01:30:27] yep that's a fantastic question um you it's such a good question it's tough to answer [01:30:39] my hunch is yes really when we think about these issues you know this is this is all right let me talk away from the literature now let's throw the literature away for a minute and i'll [01:30:53] come back in a second i think most of us would agree that the value of investing in specific technologies plants organizational structures depends very much on the types of skills workers have [01:31:07] the types of skills workers are willing to get depends on the plant organizational structure and technology that firms are going to invest in they're complements and if either one of those is constrained it's going to affect both margins that's kind of the idea of the question the way i would see [01:31:20] it i think that's exactly right ben jones has a paper where he kind of like talks about this stuff i don't think he has a full quantitative model of it though he kind of hints at it i don't think anybody's [01:31:34] really worked it out now coming back to the literature you'll notice my answer to a lot of these things are the same a lot of the really best questions come back to the same thing which is here once you write down this production [01:31:52] function that stuff can't show up so what would be really interesting if you wanted to do that i can kind of give you a road map of how to do it you would want to write down again the model where skilled and unskilled labor are like two distinct types [01:32:05] and they're imperfect substitutes then you want to have skill bias technical change but you want firms to invest in that and you want the workers to invest in the skills and then you constrain one of them with a borrowing constraint that's a model where you could talk about all of this [01:32:19] you'll notice that's not this model this model would not even let you begin to have that conversation so i think that's a fantastic idea that would be really interesting to read okay unfortunately we are um out of time [01:32:34] and um todd is willing to stay here but i know uh mandy and kirsty uh it's late in the evening for them over in the uk um so i want to thank todd i want to thank you all for participating [01:32:47] um again the todd's talk and his slides are available on the stegg website and tomorrow uh 4 p.m london time we have bertolt herrendorf who will start talking about um structural transformation at the [01:33:01] level of sectors um so let's uh let's i've already done it thank you very much todd and uh we'll see you all tomorrow hopefully thank you bye-bye um hello todd [01:33:14] hi yeah uh there is an interesting thank you very much for the lecture it's quite inspiring uh there's an interesting perspective here in in germany uh they they have [01:33:27] well structured um uh how do i put it now a kind of labor labor market kind of um the the year of schooling is structured in such a way that when [01:33:42] a child gets to a particular level it can choose to whether move for a degree or to go for what they call a kind of training what they call house building and um i'm not sure if there's any study on [01:33:54] these so far but there have been a lot of concern with respect to that because those who actually go for the house bidding system they tend to progress faster in their career because many of the organizations just need [01:34:08] the only job experience you guys know so after like a two or three year of training you go right on into the organization uh if it is a mechanical electrical or whatever depending on the industry [01:34:23] and then you get um on-the-job training now how do we factor in for this variation in in some of this model because at the end of the day the guy grows on the job [01:34:35] and we i think they reward more on years of experience and and that's why i particularly like that model the the uh mindsets wage equation because it captures both the [01:34:48] years of schooling and as well as the experience [01:35:00] yes and that's why i think you know we were excited to uh reincorporate experience back into this so at that level you know i think it's taken into account and this [01:35:13] what i mean by that is that if what we're interested in is just measuring human capital and making sure we measure experience yeah into human capital that i feel good about you know that [01:35:24] that is captured by by these papers there's a second effect which is around the world countries structure their educational systems very differently very differently yeah in some countries [01:35:39] people they call this tracking you know when you start to break students into different groups and some of them are labeled the college group and some are labeled the work group some countries do that very early some countries do it very late some countries set up exams so that [01:35:53] students study like crazy when they're 15 years old in america you don't do anything when you're 15 years old yeah at least i didn't um and i don't think anyone has yet figured out the importance of those institutions [01:36:07] you know i'm saying like the fact that it affects the experience profile we've measured that that's great i feel good about that or i don't know as good as i do about anything um but the underlying importance of the [01:36:21] institutions i don't think we have a good handle on that that's the truth and if you look by the way if you look at germany germany is a really weird country when you do these calculations barrow and lee tell us that germans have very low education levels [01:36:35] because people aren't in schooling very long they go into these vocational training i can't say the german word house whatever yeah but then if you look at their experience profiles they're extremely steep germany has the steepest experience profiles of [01:36:48] any country in the world so they're a very unusual country in this sense right very little education and education human capital but more experienced human capital it would be nice to have a theory of that but i don't think anyone does yeah absolutely and i think so to a very [01:37:02] large extent maybe and i saw one of the graphs you you presented where the return on education is actually higher even than that of america that of us and all of that and and i i think it's also partly because [01:37:17] they have this graded returns definitely if you have a bachelor's degree you don't have the same amount as someone who has a master's degree and then if you have a phd so they've got this graded [01:37:31] system that that is actually uh used in in in in how they uh calculate their thing but i don't know and but for you which do you think is the best model for a developing country to approach you [01:37:46] know uh adopt for example i think the truth is that there is so little research that's been done on this that i don't even know where to start okay i'm somebody who's really reluctant to make policy prescriptions [01:38:01] you know even in cases where i think there's been a lot of work done but here there's been almost no work i like i wouldn't even know what paper to recommend to you to read so that you could start this is this is this is totally uncharted area as far as i know [01:38:15] i think this is probably sorry i think this is probably a moment to make clear the distinction between the academic research and the policy conversation which is important and which we hope to inform but where [01:38:30] the decisions that get made are often um are often based on much more than the academic evidence right political political forces come to bear it's not clear that the policy choices [01:38:44] should be made entirely based on on what the academic evidence suggests because political processes are there for a reason but i think it's a really important and interesting set of questions that you [01:38:57] raise about what policy should do i'd love to continue that conversation but i i am conscious that here in in the uk as in other places we're keeping our colleagues at cepr [01:39:09] um past the end of their work day and i think we should wind things up i want to thank i want to thank todd i also want to thank the members of the panel the participants who've been here um you've [01:39:22] all done just a fantastic job it's great to have these conversations and um and i think one of the things we need to reflect on collectively is this relationship between what we do as researchers what we do as academic [01:39:37] economists and and the real deep challenges of making policy um one of the things that the stag program is supposed to do is to support research that speaks to policy [01:39:51] and we've all to be very honest we all struggle with what that means i think many of us like todd um feel very uncomfortable with the awareness that what we have to say is only [01:40:03] only begins to get to the edges of the big policy questions and i think what what you've raised are abdullahi are really the central questions that policymakers struggle with how should we design our systems in any [01:40:18] event i'd like to thank those of us who are still around i would encourage you if you can to unmute your your microphones and let's give todd an actual um old-fashioned out loud round of applause um i just think we want to thank you for [01:40:32] taking the time out of your day and for the effort that went into preparing the lecture todd um we'll see you all tomorrow tomorrow todd thanks thanks a ton thank you for the opportunity thanks for that [01:40:46] thanks a lot bye everybody bye