Intro. [Recording date: October 10, 2016.]
Russ Roberts: So, let's start with an issue that is always challenging, which is how the labor market's doing, how workers are doing. And inevitably in public discourse people look at the unemployment rate. But you--and I'm sympathetic to this--you put a lot of emphasis on the employment rate. In particular, the ratio of employment to population. As many economists do. And in work you've done with Kerwin Charles and Matthew Notowidigdo, you've looked at what's happened to that ratio over the last 15 years--2000 to 2015. So, talk about what has happened, what do we know about that. And then we'll talk about why we think it has changed.
Erik Hurst: Yeah. So, the employment-to-population ratio--particularly for workers with less formal schooling, so, think a Bachelor's Degree or less--started falling well before the Recession started. Plummeted during the Recession. And has rebounded only slightly since then. So, this is a phenomenon that seems pretty pronounced. It's very different than the unemployment rate, as you said, because the unemployment rate was kind of low prior to the recession started; kind of spiked up during the Recession. Took a little while, but then it trended back down to pre-Recession levels. So, you know, we've been thinking about the extent to which the labor market was weak before the Recession even started. And we're not alone on this: David Autor, some of your listeners there--you know David; you've talked to him a few times--and us, have been focusing on the role of declining manufacturing and how that has affected certain types of workers. And, manufacturing, we lost about, I don't know, 4 million manufacturing jobs from 2000 to 2004, 2005. Which is a huge amount relative to historical periods of time, for such a short time period. And, so you might think to yourself, why wasn't the employment-to-population ratio falling sharply well before the Recession started, in 2004, 2005, 2006? And, the answer is: It was kind of falling some, but not as sharply; and our story was that the housing boom came along and lifted the labor market, particularly for workers who have lower degrees of accumulated schooling. So, you know, kind of the story that we can say is that the housing boom masked the structural problems in the labor market. And the way you think about it is, suppose there's a trend down in manufacturing employment and they are displacing workers; and that started in the early 2000s. It kind of had some periods where it declined fast, other times where it declined slow; but it's been kind of a downward decline since the early 2000s. And then you overlay that with a boom/bust in the housing sector: so, housing-related industries kind of boomed in the early 2000s, well above historical levels. And then kind of returned to those levels after 2007, 2008, 2009, during the Recession. And those sectors are going to be things like construction, particularly for men: once the housing boom was going on, we built a lot of houses. But it also affected mortgage brokers and real estate agents and--you know, if we believe in some sort of wealth effects associated with the housing market, that some people have talked about--restaurants were even a little bit higher than they were in 2003 relative to prior periods.
Russ Roberts: That income effect being that people who saw the value of their house go up felt flush, richer--
Erik Hurst: Yeah--
Russ Roberts: And therefore they can afford to eat out more often or go on vacation or whatever. I don't know if I believe that, but that's the argument.
Erik Hurst: Yeah, I don't know. I mean, there is work kind of saying that. But let me tell you, kind of, what we can see in the data. So at some point, how do we kind of, you know, kind of tease out some of these kind of stories. And I'll tell you, you know, how we do that in a second. I'm going to see restaurants kind of spiked in places that had big housing booms, relative to trends. So whether the causal story goes from a wealth effect to restaurants, or something else--you know, we can talk about that. But, you know, the way we kind of make progress on this agenda is we look at different regions of the country, some of which had big manufacturing declines and others didn't. So, think about Detroit relative to Orlando: you know, [?] into basically does Disney World. So it doesn't do a lot of manufacturing. Detroit does a lot of manufacturing. So there's a national decline in manufacturing production in the United States. It hits Detroit harder than Orlando. And in some places that had housing booms. Like Las Vegas. And other places didn't. Like Omaha. So you have this variation in the economy across the locations of the economy, where some places were hit hard with manufacturing, others not. Some hit hard with the housing boom and the others were not. In the early 2000s. And you could--[?] reminded, at least it helps me to think about like a 2x2 cell where we could look at all different types of--our area of location is like a metropolitan area. So think about a lot of them. But some of them had big housing booms and big manufacturing declines. Kind of like a Bakersfield. Others had neither. Some had one. Others had the other. And you can kind of use that variation. And Detroit looked like it was having declining employment-to-population ratios, sharp decline in employment-to-population ratios, far before the Recession ever started. So the Recession, if you kind of plotted the employment-to-population ratio in Detroit from 2000 to 2015, it looks like a much more continuous decline than the United States as a whole. And then you go to a place like Las Vegas, and you plot the employment-to-population ratio in Las Vegas. You see that the employment-to-population ratio increased in Las Vegas from 2000 to 2007--
Russ Roberts: Because their housing--
Erik Hurst: And then the employment-to-population ratio--
Russ Roberts: Because their housing [?] was so large.
Erik Hurst: Yep. Yep. And then the aggregate economy is just the sum of these two types of patterns. So, in some of our work, this is kind of what we're trying to do. We're trying to show that this masking--this is the word we use--the housing boom kind of masked this weak labor market due to the structural decline in manufacturing well before the Recession started. And again, disproportionately it hit workers with less than a Bachelor's degree than workers with more than a Bachelor's degree.
Russ Roberts: That's really interesting. I'm going to tell a slightly different story, and I want you to react to it. I'm playing classroom devil's advocate here, because, you know, it's always tempting to tell a Just-So story based on the data you have. So, you said, starting--from 2000 to 2015 is a big decline in manufacturing. Of course, that decline started in around 1950. Maybe 1945. It's a post-WWII combination of increasing trade and increasing productivity in manufacturing that--we're talking about employment in manufacturing, not manufacturing output, of course, which we often stress on this program--
Erik Hurst: Right. Yeah. Of course.
Russ Roberts: The United States increased--is a much larger producer of manufactured goods than it was in the past. The manufacturing sector as an output is doing pretty well. As a source of employment, it's not doing well.
Erik Hurst: During the 2000s, that's true as well. You know, manufacturing output is going up.
Russ Roberts: So, and maybe later we'll talk about the mix of trade and innovation that might be causing that. But, so, between 1950 and 2000, when the manufacturing sector was shrinking as a proportion of total employment, and at the same time the workforce was getting more educated and more opportunities were opening up for people with higher levels of education, people in manufacturing went and found new things to do. And that employment-to-population ratio didn't fall. It wasn't so--there might be a slower trend, but it wasn't as dramatic as what we've seen. And when you say things like, 'Well, it fell, and people couldn't find work with the housing sector masked it'--but usually they find work. They just find different work. Just not in manufacturing. We didn't see a long secular increase in unemployment, or a long secular--secular meaning 'over time,' for those of us who aren't economists. Sorry for using that technical word. It's a bad word because it doesn't mean in economics what it means in the English language. But there's a long time trend toward lower manufacturing employment, but not lower employment generally. Because people found other jobs. And it could be--and this is what I want you particularly to answer--it could be, this is a crazy idea, and I don't believe it, but this is the kind of thing you talk about in a seminar--it could be that the sudden drop in manufacturing, say, due to Chinese expansion--which took away employment opportunities, say, for some Americans--that that freed up a lot of workers to go be construction workers and helped fuel the housing boom. Now, the problem with that story is we know there are other things that helped fuel the housing boom. But it is possible that the housing boom is the result of this and not the cause of the other things that we are looking at. So, react to that.
Erik Hurst: Okay. So, there's three things that I want to react with. I'll react backwards. So, the first one is: Is there anything in the data we can see whether it was the displaced manufacturing workers that helped spur the construction boom and the housing boom? And again, looking spatially allows us to kind of get some sense on this. Now, on the downside, you know, we might not see tremendous amount of migration in the United States. We see some. But, the migration across regions has been falling. And that means that we get more pockets of inactivity in one place that takes time to kind of work its way through, if migration was faster. But the good news is for us, using locations as an experiment, to, you know, test different theories. The fact that people don't migrate makes it easier for us to isolate different effects.
Russ Roberts: Absolutely.
Erik Hurst: So, why do I bring that up? Because we saw the biggest decline in manufacturing, in places like Detroit. And there wasn't any housing boom in Detroit. And if you kind of correlate the places that had the biggest manufacturing declines and then the places that had the biggest housing booms or construction booms--or you can do it in housing prices, or construction activity--it's uncorrelated in the data during the early 2000s. So a story like you told, that there would be some kind of causal link
Russ Roberts: Reverse causation--
Erik Hurst: Declining--yeah, exactly. You would kind of expect to see the places that had the big manufacturing declines to maybe have a little bit bigger housing boom than other places. And you just don't see that in the data. And so we can kind of test that: that's something we can literally test.
Russ Roberts: Well, you can test it at the aggregate level, and I think, by the way, I think--
Erik Hurst: At the local level. That's what I was trying to test.
Russ Roberts: Yeah but--isn't it possible--again, I'm just speaking creatively here; not, I don't believe this story; but it's an interesting story. It's possible that even though migration is very low in the United States and physical mobility is very low, residential mobility, people moving to a new location--we are not talking about whether they can move around, the ability to move locations for work or home. It's possible that even though there's very little at the national level, that after the Detroit manufacturing trouble, after auto workers found many fewer opportunities there because both plants got more effective--meaning fewer workers were needed at any one plant; and many plants moved south or elsewhere--they could have moved to Las Vegas when they heard there was some opportunities there. And suddenly a boom started. Now, I think that's unlikely. But that's the--you have to--I find it remarkable at how little time economists--maybe sociologists have done it--have spent interviewing people in Las Vegas who worked on those houses and what happened to them. And what are they doing--are they [?]--
Erik Hurst: Yeah. We don't have that kind of narrative, that I would love to have. The only thing I could see in the data is I could ask: Of the people who moved into Las Vegas--so you could observe that, because some people didn't move in--
Russ Roberts: Yes--
Erik Hurst: Whether there are a disproportionate amount of them coming from places that had big manufacturing booms, relative to some sort of trend. And you see a slight amount. I just want to say it's not zero. You see a slight amount. But of workers over the age of 30, you don't see much of an effect. And that's not surprising because workers over the age of 30 just don't move that much. So, but over, you know, for young people in Detroit, they might have chosen to migrate to places that were other, you know, were booming a little bit more, in those construction industries. But you don't see it as much for the older--again, looking at the 30+-year-olds who migrated into Las Vegas. So, again, I think we don't have a complete way to rule out your conjecture. Just the first signs of, all, you know, some of these test that I would expect to see if your story was a first-order story, we just don't see in the data. So, again, it doesn't mean it's not there in some ways we are not seeing. But I think it's just--it's occurring in much more subtle ways, then.
Russ Roberts: I guess the other factor would be if what was driving Las Vegas was an increase in the demand for housing, rather, or a reduction in the cost of building a house unrelated to labor, you'd expect to see the demand for workers to increase; and workers', construction workers', salaries to go up rather than down--which is what you'd see if workers were flooding in from Michigan. And I suspect it was very lucrative to be a construction worker in Las Vegas in those times.
Erik Hurst: It was. Because we find wages in work were the same with what the employment rate says. Which, again, you know, if we are telling stories about labor demand, kind of the narrative we are saying is that manufacturing, you know, kind of stopped using workers as much as capital, potentially, in the production of its output. So the demand for workers are going down. If that's true, we should see something that wages fall. And we do. And if you go places like Michigan, wages were falling during the early 2000s. You go to Las Vegas, wages were rising in the early 2000s. Now, again, if we had a perfectly integrated labor market, those differences would be arbitraged away by people moving from Detroit to Las Vegas. And we just didn't see that in the short run. Doesn't mean it doesn't happen in the long run. I think that's what I want to get to with the response to your other conjectures about the long-run trends in manufacturing that have been going on. I think we agree--I think you and me agree; I don't know if everybody agrees, but you and me agree--that people adjust over time. So, you know, we used to be 100%, 80% agriculture workers in, you know, what? 1860. We are not doing that any more. People adjust from sectors. The difference with the decline in manufacturing in recent periods, relative to earlier periods, is, in earlier periods, manufacturing jobs were, you know, slowly going away. But at the same time, population was growing. So new young people could come into the market, and they adjust; and they don't see, there isn't enough jobs in manufacturing, and they kind of reallocate to other sectors. In the early 2000s, we lost, as I said before, about 4 million manufacturing jobs, in the early 2000, 2004, 2005 period. You know, from 1980 to 2000, we lost about 2 million manufacturing jobs over that 20-year period. And here we lost, you know, very quickly. And it might be that when shocks happen quickly, it just takes people longer to adjust. Some people get displaced; and then they have to work through, you know, accumulating new skills, moving to a different sector, moving to a different location. And if the young are more likely to do that than older workers, then we have to have enough young to kind of clear the market. And that might just take a little time. So, I don't think this is going to be a long-run problem. I just think it's--you know, we are seeing the medium run responses to this right now. And then, like, the third kind of component that I want to digest[?] at, is that we have seen employment rates for men falling--
Russ Roberts: Yeah--
Erik Hurst: over this time period as well. So, there is a correlation with just the share in manufacturing and the employment rate, that has been moving, you know, at least during the 1970s and 1980s and 1990s as well. I don't have good data, much before the 1950s and 1960s. So, but in the 1980s and 1990s we are seeing some relationship between manufacturing decline and employment rates.
Russ Roberts: You raised an interesting point I hadn't thought of carefully enough before. It's an issue we've talked about before on this program, which is: Is it possible that the dynamism and effectiveness of the labor market is not what it used to be, because--while it's always been true that there's creative destruction, some sectors are shrinking, others are rising--it could be that people without a college education in today's economic environment simply struggle to find alternatives to manufacturing and construction. And then, there's two ways to think about that--which is what you got me to see. One is: you know, you are a 50-year-old construction worker, manufacturing worker--and neither of those are doing very well, and there isn't an easy place to turn. And it may take a while. The more depressing possibility is it's actually the 25-year-old worker who can't find anything because that person isn't sufficiently educated. And that's a longer, bigger problem. And the future just looks tougher and tougher for people like that. Have any thoughts on that?
Erik Hurst: I concur again. I mean, your worldview isn't that different than mine. I mean, at some point--I'm trying to think about ways to test this, but me and Kerwin and Matt, the authors of the Masking paper, have been thinking about this, and trying to extend it in some sort of scientific way. To try to estimate exactly what you just said: What is the skill substitutability of different workers to different occupations, now, and has it been different in the past? Or has it always been the same, but just the mix of jobs has changed? Let me give you an example. Is it easier to move from agriculture to manufacturing than it is to move from manufacturing to services? That's--I don't know the answer to that.
Russ Roberts: That's [?] to think about.
Erik Hurst: And we know we went though a major transition in the U.S. economy from agricultural to manufacturing. Particularly for workers with less than a 4-year college degree--so a Bachelor's Degree, those were the primary sectors in earlier periods, you know, and still today. It's just the share of agriculture is small relative to manufacturing. Now, manufacturing is moving away. Are we seeing that it's maybe harder? And this is maybe kind of your conjecture--that maybe it's just harder to switch to other occupations? And then that would show up in the young as well as it would be in some of the older workers. And we are seeing now that the employment declines during the 2000s for young workers seem much more pronounced than they do for older workers. Conditional on skill.
Russ Roberts: Yeah, we're going--
Erik Hurst: That's something I've been kind of thinking about as well--is that these young workers that look different well into the recovery, compared to their older counterparts. Again, holding skill constant.
Russ Roberts: Although, that recovery--you know, you are talking about 2000, 2007, a time when the economy was doing pretty well--not great; there was a little recession at the beginning, but manufacturing jobs were doing poorly all the way through it. Then we had the Recession, which had an another tremendous shock to construction and manufacturing--again, two sectors where low educated folks find work. And those jobs may not be coming back. So, what I'm worried about--and its my next question--we're going to turn to younger workers in a second in specific. But if, just to take one example: If driverless cars and trucks become a reality in the next 5-10 years, which I think they probably will, that's another group of millions of workers who find decent jobs--sometimes pretty high-paying jobs--without a lot of education as truck drivers or Uber drivers or cab drivers. And those jobs might just totally disappear. What sector will those folks find work in? It's not going to be manufacturing. It's not going to be construction.
Erik Hurst: My conjecture is we're going to see, you know, a continued decline, then, in the employment-to-population ratio. I mean, this is, you know, just to put it in perspective, I was just doing some work this morning: that, relative to 2000, 31-55-year-olds with less than a Bachelor's Degree have reduced their hours' work collectively by 10% over this time period. That's through 2015.
Russ Roberts: Say it again.
Erik Hurst: So, annual hours worked for this group as a whole, 31-55, with less than a Bachelor's Degree--they worked about 2000 hours per year, in year 2000. And now they are working about 1750 a year, in 2015. That's a 10% decline, roughly, a little bit over--
Russ Roberts: For a short time period. Yeah.
Erik Hurst: Over. Yeah, exactly. A 15 year period. And it's not getting that better. So, it's not like you are seeing through the recovery that there's a big spike up back towards those 2000 levels. That is a big decline, for a group over a 15-year period. And again, it doesn't look as cyclical. And what I mean by that, you don't the cyclical stuff--I think usually when the economy starts to rebound, you see a rebound in hours. And, you know, since 2012, hours have been going up. But just by smaller rates than they fell during the early 2000s. So it's still to the point--you are about 10-11% decline in annual hours for this group as a whole, relative to 2000. That's a massive decline for a group who are prime age, that has nothing to do with retirement--
Russ Roberts: going to college, graduate school--
Erik Hurst: It doesn't go into graduate--this is, actually I've thrown out anybody going to graduate school. So--and the 30-year-olds just don't go to school as much as people in their 20s. So this is a rather pronounced fact that we're, you know, confronting. And I think your conjecture, that other technological advances might displace other jobs that they've migrated to, I think is--you know, on the table. It's something we might have to confront going forward.
Russ Roberts: And--normally, I use the example all the time about the blacksmith. The blacksmiths' jobs disappeared around 1910, 1915, 1925. And they went ahead to do other things. And if you had a specialized skill as a blacksmith, life was a lot harder and it wasn't easy, and you may have struggled to find work. This is a shrinking group--the group that you are talking about, the group that doesn't have a college degree. But it's still very large--is the problem. It's not like, 'Well, there's going to be a tough'--it's just a large portion of our population.
Erik Hurst: Yeah. 70%. Yeah. I know I tell my students, maybe many of your listeners. We tend to view the world like, you know, the people around us.
Russ Roberts: Yeah.
Erik Hurst: But it's 70% of men, 31-55, don't have a Bachelor's Degree. It might be 68%. But somewhere in that ballpark. So, think about it as a third do have a Bachelor's Degree. So, this is a very large group, still. And education rates have slowed down for this group, as well. While we've seen gains in, you know, college attendance for women, you haven't seen that much for men. And I want to stress that there's many ways to get skilled beyond going to college. You can--apprenticeship, and you know, you can learn a craft. But we have seen migration towards college over long periods of time, and that seems to have slown down for this group as well.
Russ Roberts: Yeah. And as I've learned from Bryan Caplan, and as you are pointing out, attending college is not that helpful if you do not graduate.
Erik Hurst: Exactly. Yes.
Russ Roberts: So, let's continue to go deeper into this. Let's talk about the other work that you've done. I think this is work with Mark Aguiar, Mark Bils, and Kerwin Charles on the behavior of what you call less-educated young men. And you found a rather striking set of behavior, both leisure and lifestyle, over this 2000-2015 period. So, first, tell us how you get information about these folks that you are going to be talking about in terms of their use of leisure. And, strangely enough, this is going to potentially revolve around video games. Which is kind of shocking, but really provocative. So, talk it.
Erik Hurst: Yeah. So, the first thing to stress, though, in all of this, is ongoing research even as we speak. It seems there is a tremendous amount of interest in this work. So even when we present once or twice at a conference, it's picked up a lot of interest. So I shared a version of the paper with you. But we even didn't even post it on our website. Yes. So we're still a few weeks away from you know, kind of putting this out for public consumption. But the trends, you know, that we are documenting, are of the data. So let me tell you about those. So, using data from household surveys, that measure labor market status--so that's going to be some of our work, the same data that I've been using to talk about employment rates over time--using data on, from the American Community Service, which is like the Census Long Form--I can get detailed data on cohabitation patterns. I'm going to use data on the American Time Use Survey, which allows me to track how people are allocating their time when they are not working. And then I'm going to use data from the general social survey to track happiness. So, think in your mind now, for to use young workers--21-30 is what I'm going to call 'young' right now. So, kind of, just after schooling for most people, even though in the 21-30 some people are accumulating some skill, and we'll deal with that as well. So, 21-30 year olds, with less than a Bachelor's Degree. Okay, so that's kind of the group that we're focusing on, now. And so I'm going to go through some employment stuff. I'll go through some cohabitation stuff. I'll go through some leisure stuff. And then I'll go through some happiness stuff. That is all just kind of background information. And then I'll try to tell you how we're trying to put structure on it after that. But this is just unconstrained data.
Russ Roberts: Just the facts.
Erik Hurst: Just the facts.
Russ Roberts: To the extent they are facts. When you have to be a little bit careful about--
Erik Hurst: To the extent they have to be--exactly.
Russ Roberts: What people say they spend their leisure time doing.
Erik Hurst: Exactly. Exactly. Exactly. So, subject to any types of reporting error in the surveys--that this is just describing what's in the surveys. And that's a better word than 'facts'--within the surveys. So, the first thing is what I alluded to earlier, that the employment-to-population ratio, or hours worked, as we were talking about before, have fallen much more sharply during the 2000s for young men relative to their older cohorts. Okay? So, what I'm going to give you now is conditioned on men 21-30 with less than a Bachelor's Degree; and then men 31-55 with less than a Bachelor's Degree. And, as I mentioned earlier, those 31-55-year-olds kind of had a 10% decline in their hours during the 2000s. For the 21-30-year olds, there's been about a 15% decline. So, you know, you might think it's a 10% versus 15%--but that is massive. A 15% decline in hours is a very large decline. This is not substitution to school. So, I'm already conditioning on people enrolled in full-time school. So these are of people who aren't working and who aren't going to school. And if we go one step further and ask: What fraction of this group didn't work during the prior year? In 2015 relative to 2000? So, in 2000, if you happened to be a 20-year-old man with less than a Bachelor's Degree--and when I say 20-year old, I mean in their 20s--21 to 30. About 8% reported not working at all during the prior year. That number, today: About 18%.
Russ Roberts: Stunning. Unbelievably depressing.
Erik Hurst: So--18% of this group is sitting idle, for the prior year. So, the way I get this is just, these surveys, to the extent they are accurate--ask, 'How many weeks did you work last year?' And if you wrote, if reported, '0,' that's what I'm kind of tabulating. There's 18% of this group that are reporting Zero in two-thousand-and--you know, 2015.
Russ Roberts: So, two things are striking about that. One is, it's a big number--to me, and to you, obviously as well. But it's also double what it was 15 years earlier.
Erik Hurst: Exactly. And that trend started--again, just like these other things we are talking about--prior to the Recession. So you saw this increasing 2001, 2002, 2003, 2004--you know, by 2007 it was like just under 12%, maybe about 11%. So, 8% to 11%. And then it went to 18% during the Recession. And it's remained relatively flat from 2010 onward. So, this is--you know, the labor market measured by, 'Did you work last year?' hasn't improved from the depths of the Recession. And, you know, we have an idleness-not in school, not working, about one fifth, just under one fifth, of this group. That is stunning to me. Okay, so that's a fact. So then you might ask yourself, where are these people living? How do they eat?
Russ Roberts: Yeah.
Erik Hurst: That's what we go to the cohabitation data, from the American Community Survey. And there, you could track, you know, households. Just like the Census does. This is exactly what it is by the Census. That's what they are designed to do, to try to create a census of, you know, how many Americans there are and, you know, where they are residing. And in this group you see a large--again, just mirrored their nonemployment rate--a large propensity, and increasing propensity to cohabitate with their parents. Roughly 70% of this, men in their 20s, 21-30, with less than a Bachelor's Degree, and who aren't working, live with their parent or close relative. Most of them parent. 70%. And that number was like 50% in the early 2000s. So, again, we're seeing a shift in lifestyle. You might ask: Are they married? Are they having kids? And the answer is, 'No.' This is not a group--if you are not working as a man in your 1920s with less than a Bachelor's Degree, you are pretty much single and childless. Not all. But 90-some percent of them aren't married.
Russ Roberts: Why do you focus on men? And what's the story for women?
Erik Hurst: You don't see the same patterns for women. So, you don't see the differential patterns for young women relative to young--for older women, conditional on skill. So, for women, 21-30, the decline in hours during the 2000s, from 2000-2015, was very similar to older women, 31-55. And again, all groups are experiencing a decline, particularly, those with less than a Bachelor's Degree, for the reasons you and I talked about earlier. The low labor demand for people with less [?] see the differential patterns for women relative to--young women relative to older women. And so that's why we kind of looked at these young men, where the pattern seemed to strike out more, in terms of idleness. And I think that's why we focused on the men. I don't know if--maybe we should go back and look at women, too, but we just didn't see the gap between young women and--
Russ Roberts: Talk about they are using their time.
Erik Hurst: Yeah. So, women, we go to the time diaries, and this is basically for your listeners to get a sense--questions about: Tell me what you did yesterday; and then you put in detailed categories of time use. So, activities like, 'I ate my dinner from this time to this time,' and 'I went to work from this time to this time.' And for young men during the 2000s, we see this big increase, particularly young low-educated men, in leisure time. Not surprising: their working is falling down. And it shows up in leisure activities, like watching TV, hanging out with friends, playing video games, etc. And then when you into the subcategories of leisure and ask, 'Which leisure category was increasing the most?' it was far and away video games. So, let me just give you a couple of numbers from our paper. I just want to make sure I get the numbers correctly right. That, for, again, early 2000 for this group of young men with less than a Bachelor's Degree, they used to play about 3 and a half hours per week on video games, and now it's about 6 and a half hours per week on video games. So almost 100% of their increase in leisure was playing video games and on the computer. What I gave you before was video games and computer time. I should have been more clear on that. Most of that was video game time. But the numbers I gave, 3 and a half to 6 and a half was all time on the computer--so anything you do on the computer: surfing email, going to web pages, and playing video games either on the computer or on a consol. So this is a big change, in terms of how people are allocating their leisure. And you don't see the same pattern for women. So, women don't have this shift. That's another reason why we focused on the men.
Russ Roberts: So, what do we make of that? You wrote a really nice essay, which we'll link to, where you talk about the fact that everybody from the ages of 8, maybe 6--depends on your house and what the rules are--but parents have noticed that it's kind of hard to get people off video games, off the computer. And this is a phenomenon that has nothing to do with the Recession; it has nothing to do with education. They are really fun for certain types of people, and they really like them, and it's hard for them to stop playing them. So, there's a certain addictive--I don't like that word, but in common parlance, addictive nature to them. So, it raises the question really on: what's the causation here? So give us your thoughts on that.
Erik Hurst: Yeah. So, nothing we ever have has addiction in it or not. So, let me give you an overview of the model that we're trying to think about, and then I'll tell how we come and test that. So, the overview of the model is just: something comes along that makes leisure time more attractive. So, when we're making our decisions to work, we compare the benefits to work relative to the cost of work; and the cost of work is you forgo the wage you could have earned if you were working; but you also get the marginal utility of leisure. And if leisure time is becoming more attractive, that should have an effect on people's work decisions. Now, how big of an effect--we're trying to answer that. But that's kind of the theory in the background. So, earlier we were talking about labor demand effects on people's work effort; this is more of a labor supply effect. So, how do you go out and test causation about whether people are playing more video games because they have nothing to do; or people who have nothing to do, who are choosing not to work because they want to play more video games? So, what we do is we try to estimate the demand system for leisure. So, what does this mean? It means, think about it in your mind about how much of a leisure--time spent on a given leisure activity on total leisure time, on one axis. So, the share of leisure time on TV-watching, the share of leisure time on hanging out with friends, the share of leisure time on playing video games--as a function of total leisure time. And suppose there is some relationship that we can try and estimate that say, 'Okay, if your leisure time doubles, you want to more than double--or less than double--your time on different activities.' So, we then go to the time diaries and try to figure out whether people are changing their time on given activities; whether that looks like a movement along one of these what we call leisure Engel curves, or whether it looks like a shift in one of these leisure Engel curves. And the Engel curve, again, it's just a mapping between the share of time on a given activity out of total leisure time on one axis, and then having total leisure time on the other axis. And so a whole new paper is trying to do exactly this. And we use the time diary data to try to see how you, Russ, would spend your leisure time if I shocked you out of work. And so that gives me the slope of this leisure curve. So we use regional variation just like I was talking about before: you're sitting out in California; people like you lost their jobs a little more in California more than people in Texas. And then I see how people in California allocated their leisure time after they lost their jobs, compared to people in Texas. And then I not only do it for California and Texas, I do it all states; and I use it, variation during the Great Recession, to isolate that. And that kind of gives me the slope of these lines. And then I go, in the time series data, the aggregate data, and see whether the changes through the 2000s for everybody puts us on that line, or are we off that line? And if we're off that line, it's a way we try to identify if there's been technology innovation in the leisure categories.
Russ Roberts: I just want to add--
Erik Hurst: I know that sounds a little bit confusing--
Russ Roberts: No, that's cool. That's really great--
Erik Hurst: But that's exactly what we're doing. And then with that we use the model. So there's no identification here. We use the model to figure out how much labor supply changed from a leisure shock. And so we find some effect, a small, non-zero effect of innovations in leisure computer activities for young men that explains about 20% of the decline in hours that I was telling you about earlier.
Russ Roberts: So, I just want to clarify one thing. You mention Engel Curves--that's 'E-n-g-e-l.' It's not Friedrich Engels. It's Ernst Engel. Which--I didn't know his first name but I cheated and looked him up on Wikipedia.
Erik Hurst: I didn't know his first name, either.
Russ Roberts: I used Wikipedia while we were talking.
Erik Hurst: It usually out of the demands. It usually is in money space--how much money you would spend on restaurants out of your total money spending. Or, how much money you would spend on, you know, vacations, out of your total money spending. We've kind of defined, now, Engel curves in time space, as opposed to money space.
Russ Roberts: So, I have no doubt that I spend more time on my smartphone as a form of leisure today than I did in 2000, because I'm pretty sure I didn't have a smartphone in 2000. And it's possible that I didn't just--that I don't just spend less time with friends and less time reading. I just spend more time fooling around--wasting, let's call it 'leisure' as we would in economics, unjudgmentally. But I concede it as sometimes wasting time. You are making the claim here--I mean, it's an incredibly interesting idea; I don't believe it--but it's a modest claim. It's a modest claim.
Erik Hurst: Which part don't you believe?
Russ Roberts: Well, it's hard to accept the idea that people in their 21-30s find so much fun from videogames that they've chosen to live at home and play on their parents' computers or consoles, as opposed to the idea that they are having trouble finding work, and while they are unemployed they may as well play video games, because that's the thing we do now when we have extra time. So, you concede, obviously, that it's both--
Erik Hurst: Yeah, but I've got a couple of questions for you.
Russ Roberts: Go ahead.
Erik Hurst: Again, I want you to follow your own logic for a few more steps.
Russ Roberts: Go ahead.
Erik Hurst: So, the first step is: You have a conjecture that these people would be less happy under their current lifestyle than they would have been before. So, I could go to the Happiness data and just download--
Russ Roberts: So clever[?], Erik--
Erik Hurst: in the Social Survey, and ask them how their life is. And despite the fact that the hours worked have fallen by 15%, and their propensity to live in their parents' basement has gone up tremendously, their reported life satisfaction has kind of gone up. Because that's one thing I want to say. Second, I want to ask you: Do you believe that innovations in home production affected women's work? I mean, we have a large literature in economics talking about that. Do you believe it, or not.
Russ Roberts: I'm not sure. I think a lot of people are skeptical about--I think the nature of it changed. I think laundry down by the river is hell for a modern woman--or man--and we're all grateful that we have washing machines. But we also--that tended to push our time some way--yes, women worked more, they worked more hours. But they also spent a lot of time doing other things around the house--they'll tell you.
Erik Hurst: Of course. Of course they did. And I'm saying, what we saw in the 1960-2013 period, -2015 period, is a large reduction in women's time spent on non-market work, on home production; and an increase in their time spent on market work. And, you know, going back to Becker, and before him Mincer, kind of showed that the existence of an outside option that women did, historically--it's gone down recently, but historically--has made women's labor supply more elastic than men's. So there's a large literature that women's labor supply is more elastic than men's.
Russ Roberts: More responsive to wages.
Erik Hurst: Yeah. More responsive to wage change. And that potentially, innovations in home production, have helped, not completely but on the margin, resulted in women working more. Our paper is kind of the opposite. It's basically how innovations in leisure technology has drawn people away. And some of these people tend to be a little bit more elastic than they would have before. And maybe--again, this is a conjecture--maybe people are taking more leisure on the front side of their life now than on the back side of their life--in retirement, because we're living longer, maybe it makes sense to take a little bit more leisure on the early part of their life.
Russ Roberts: Well, I think they take it everywhere. I think they take it on the job. There are people who play video games--I know right now, people are listening to this conversation on the job. They tell me so. And I will call that leisure. You can call it investment if you want to fool yourself, possibly, or console yourself--I don't know how to describe it. It's a fascinating idea. I'm pushing back on it. It happens.
Erik Hurst: Yeah. And [?], quantitatively is the real issue for me: how big is it? Theoretically, it kind of has to: When the price of apples goes down, do you buy more or less apples? You kind of buy more apples when the price of apples goes down. When the price of leisure goes down, in some sort of, you know, utility sense, do we take more or less leisure? Theory says we should take more; and then it's got to come from someplace. And now we're saying some part of it might be coming from market work, because there's a time budget constraint.
Russ Roberts: Yeah, but the margins are really subtle and complicated. There's a big difference between me saying--again, which I think is absolutely true--that my evenings are very different today because of the pleasure I get from what the Internet gives me, relative to my evenings 20 years ago. Twenty years ago I did a lot more reading of books. I still read books. I read them online. And I'm calling it a separate thing from[?] say, watching YouTube videos. I don't play videos but I have the equivalent of that: reading that's 73 articles written about the Patriots today--and I apologize because I know you are a Dolphins fan.
Erik Hurst: I am a Dolphins fan.
Russ Roberts: So, I spend a lot of leisure time doing that. And you're right: it has to come from somewhere. But I worked last year. So, to say that there's a big difference between saying at the margin of spend a little more time on the Internet and a little less time with my family--which I may not be happy ultimately about that tradeoff--or a little more time on the Internet, a little less time advancing my career--I get that. But to suggest I'm going to take a few years off so I can play with my video games is a little bit dramatic.
Erik Hurst: We get into the habit of viewing--
Russ Roberts: It's almost like saying--
Erik Hurst: viewing the world, our kind of situation, as we talk about in the paper, one thing these effects compounded: If you are close to your reservation wage.
Russ Roberts: Excellent point.
Erik Hurst: So, if you are on the margin of working or not, then movements in the value of leisure could have bigger effects. You and I are kind of earning well above our reservation wage, I would conjecture. And as a result, movements in the value of our reservation wage probably isn't going to have the same effect on us. So it's a confounding[?compounding?] factor that a weak labor demand and a more attractive outside option could potentially have the effects. Now, again, the [?] compounded--
Russ Roberts: Reservation wage?
Erik Hurst: Reservation wage--the value of outside options. The value of not working.
Russ Roberts: Yeah. The thing it would take me to get into the labor market. How much I'd have to earn.
Erik Hurst: Exactly. Exactly. And so it's a combination of the situation you and I were talking about before--weak labor demand--interacting with things that can be moving around--how happy you are not working. And I think parents allowing you to live in their basement helps with some of that as well. And the fact that you could do something kind of fun while you're in the basement kind of helps with that. So, if we strengthen the labor market--I think maybe your conjecture isn't too far from what we are saying. If we strengthen the labor market such that people's market wages were farther away from their reservation wage, as we defined it earlier, then you wouldn't have this problem. The problem is because it's close to the reservation wage--the market wage is close to the reservation wage--things that move around how enjoyable leisure is could have more of an effect for people who are on the margin of working to begin with.
Russ Roberts: So, I don't know how well you can throw a stone, Erik, but somewhere within a stone's throw by someone of you--it might be a Cub's pitcher, to throw the stone--but somewhere within a stone's throw of you right now, possibly, is Casey Mulligan--
Erik Hurst: Yes; I probably, back when my arm was a little stronger, I could throw that stone.
Russ Roberts: So, Casey has been arguing that a lot of the decline in employment opportunities is supply driven. It is choice-driven by the people we're talking about who, because they are receiving more generous benefits in all kinds of ways, and because the implicit tax rate--what they forfeit in benefits when they start to work--is so high for low-skilled workers, that's what's driving some of these effects. Talk about his arguments.
Erik Hurst: Yeah--I don't want to--I mean, I'll tell you how I view for our group of studies. So, I don't want to talk about, you know, older potential workers. For these young workers, they don't receive a huge amount of government transfers. So, what do I mean by that? There's not a welfare program designed for young, single, childless men. So, in the data you can actually see this. There's just none of these guys receiving any government program. So some of the stuff Casey talks about, like SNAP (Supplemental Nutrition Assistance Program) benefits--
Russ Roberts: Unemployment insurance--
Erik Hurst: Unemployment insurance--none of these guys have had a job, yet. So, none of these things--Worker's Compensation, Disability--none of them kind of accrue to a group that has no labor market experience and tend to be childless. So, for this group, I think public transfers probably aren't discouraging the work effort. Again, I can't speak to the older cohort. But what is there, is private transfers are extremely important, as we talked about before. The way they are able to survive is through cohabitation, with, usually a parent. If not a parent, a close relative like a grandparent or a sibling. But almost always a parent. And they are providing transfers. And that income effect, if you will--the fact that they can still eat and take leisure, has to have an effect on labor supply. Let me kind of flip it around. If they are out on the street, my guess is that would dramatically decrease their reservation wage. They would be much more willing to work even at low wages if it came up to a decision to starve or not.
Russ Roberts: And of course, because of--
Erik Hurst: I don't think. Again--so I think, you know, it's a similar kind of feeling to something Casey was talking about, but not so much through the public policy dimension, but more through interfamily relationships that are providing that insurance or subsidy. Again, I don't even know what's the right word to use. Is it insurance? Or is it a subsidy? I don't think I have anything in the data that distinguishes the two.
Russ Roberts: I was going to make a remark--we were talking about the housing boom earlier. And of course these folks get to live in their own bedroom. They are not really in the basement. They are in their bedroom.
Erik Hurst: Yeah.
Russ Roberts: In the old days, they'd be rooming with the younger sibling, which would take some of the fun out of it. But since houses have gotten larger, through both the fact that we are wealthier and also because we've subsidized housing artificially--and so, it's easier just to live at home. And I'm not surprised that happiness levels are high.
Russ Roberts: Let's turn now to a longer time trend, a longer time period. Really interesting work you've done on leisure over the 40 years between 1965 and 2005. And something that changed in 1985, which is--it ties in to some of this in a certain way. It's really interesting. So, talk about what you found there, in that study. And that was work with--
Erik Hurst: Mark Aguiar. And I have, you know, besides our leisure paper in recent time period, Mark and I have probably 5, 6, 7 papers together on exploring the interaction of how people use their time away from market work, with long run trends in market work, business cycle trends in market work, and life cycle trends and market work. And what you are kind of referring to here is some of our time series patterns. And, since 1980, we've seen the leisure-time increase--and we can say [?] orthogonal to what we were talking about earlier--is, leisure time increased for lower-educated workers has been much more dramatic than higher-educated workers. And this just--kind of the flip side of the side that market work hours have declined more for low-educated workers than they have for high-educated workers. And, so, when we started thinking from the fact that, when we measure kind of wellbeing measures, there is some value from leisure going on. Now, on the other side, we're right back to where we were before: How much of it is a constraint because labor markets are poor and how much of it is a choice because labor markets are poor? And, that's something Mark and I have been trying to think about. And you are starting to see that in our leisure/luxury paper that we were talking about earlier, about the video games. But it's something that him and I have been thinking about now for almost a decade.
Russ Roberts: And what changed in 1985? In the data. It's an interesting difference there between high- and low-skilled, high- and low-educated workers.
Erik Hurst: We know that the skill premium has been changing since that time. And, and, so what is that skill premium? The wages of those with a 4-year degree have been rising at a rate that's higher than those with less than a 4-year degree. And if the returns to work are lower, you know, traditionally [?] theory says, if a substitution effect is important, then people should be working less. And the substitution effect would be more important if there is, again, some sort of insurance mechanism, either from the family or the government, that insures people when they are not working. And I think that's kind of the patterns we are seeing in the data. The skill premium patterns, and the market work patterns, and as a result, conversely, the leisure patterns, are all highly correlated with each other. And just for people who are listening at home: The difference between market work and leisure is there are other uses of people's time. You can invest in schooling. You could take care of your children. You could clean the house. You can, you know, invest in your own health care. So, there's lots of different other uses of time. And we tend to measure 'leisure' as those activities like hanging out with your friends, watching TV, going to the movies--those types of leisure activities. Playing video games in recent periods. So, prior to 1985, leisure patterns were increasing for both higher-educated and lower-educated workers. So--for both men and women. So, men were taking more leisure, usually by working less, in the 1960s and 1970s. Women were taking more leisure by working more in the market but working much less in the home sector. Kind of like we were talking about before--the increases in home production technology might have freed up some time for women to both work more and take more leisure. So, higher skilled and lower skilled--at least measured by education--higher educated and lower educated men and women were tracking each other very closely in their leisure time, up through about 1985. After 1985, that breaks. And when it breaks, it's right around the same time period--people have tried to estimate it--that the skill premium started to change. And this isn't my work but work by Larry Katz and David Autor and Kevin Murphy--a whole bunch of people show that the earnings of the higher educated have been growing faster. So, Mark and I have been puzzled: Do we believe that income effects or substitution effects are important in determining labor supply? What do I mean? If your wage goes up, do you work more or less? Okay?
Russ Roberts: Because it could go either way. Usually you'd say, 'Well, I'll work more--I'm getting more per hour.' But then you are wealthier, so you might decide to take some of your added wealth in the form of leisure.
Erik Hurst: Exactly. Yeah. So, what I want to tell you now is just a puzzle Mark and I have been--maybe have identified--but have been trying to solve over a period of time. So the puzzle is, as we got richer in the 1960s and 1970s, both lower- and higher-skilled workers, leisure went down [up? --Econlib Ed.]. We got richer; we all took more leisure. Hmmm. Sounds like something that would be consistent with an income effect potentially being important. Then we go from 1985 on and the skill premium starts to change such that the price of leisure is really expensive for high-skilled relative to low-skilled. And during that time period you see this big divergence in amount of leisure time people are taking between these two groups. And you look at the data and you are saying, 'Maybe that suggests a substitution effect is important.' The fact that as we hire educated individuals got higher wages, it was more expensive to take leisure now. That's a substitution effect. So, how do you explain the time series patterns and the cross-group patterns within the same model is something Mark and I have been thinking about for a long time. And we have not yet solved that. So, we've documented patterns; we've postulated puzzles; and we've left it at that.
Russ Roberts: Well, one of the challenges , of course, is that, again, there are a lot of margins that people make these decisions on, and one of them is over their lifetime. And we don't have good data on lifetime labor supply and leisure. We have some, but it's not as good as our annual measure, say. And so it could be that people with high wages work a lot when their wages are high, when they're young-ish; and they retire earlier. There's some evidence of that, right?
Erik Hurst: Yes. Yes. I mean, that we could. I don't know if we're always retiring earlier, but we're living a lot longer. So, the amount of leisure we take late in our life, even if we retire around the same ages, or even if we work a year or two longer, is offset by the fact that people with higher education tend to live a few years longer. So, you are getting more leisure later in the life cycle.
Russ Roberts: But I think the other thing--and I say this sometimes in jest but I think it's quite serious, and it's not, I didn't think of it myself; I can't remember who told me--it might have been Larry Iannaccone, a colleague of mine from George Mason who years back suggested this to me. So, I work--my working day is really about 7:30 in the morning till about 1 o'clock at night. So, I work--that was humor. Thank you. I work about 18 hours a day. Now, because when I'm in the shower, I'm thinking about EconTalk sometimes. And when I'm doing the dishes, I'm something thinking about a video I want to create. So I've got work on my mind a lot; and I happen to really like it. And that's another phenomenon that's new, I think: the number of people who actually enjoy their work. At the same time, I'm taking leisure all the time during the day. I'm reading that story about the Patriots, I'm watching the YouTube video of Lin-Manuel Miranda that my son sent me last night.
Erik Hurst: From Saturday Night Live.
Russ Roberts: Yeah. I'm doing it all. I am--I'm--it's a very rich, multitasking, crazy world I live in. Compared to my dad, who came home at 5:30. We had dinner--and I have dinner with my family still, and we don't have any devices--I just want to say, for the record, I think it's a really good idea. But my dad would come home 5:30 and then he--we'd read a book most of the night. And we'd talk sometimes. But it's a different world. And so, I think our ability to measure leisure accurately is probably very challenging.
Erik Hurst: I do, and I think it's more challenging now than it used to be--it's exactly what you said. And particularly because things intermingle all the time. You know. And particularly, what I'm most concerned about, in terms of our work, Mark and I's work going forward, is we don't have really good measures of time use when people are at work. Our time-use surveys are pretty much: 'What time did you spend at work?' not accounting for the fact that you might be working, kind of, and listening to your podcast at the same time. So we don't have those types of metrics. And for some jobs, where there is that type of flexibility--leisure and work link together. I think outside of work, we have detailed categories. And sometimes we are eating dinner and watching TV; and those are kind of similar enough that it doesn't matter when I'm clumping them together. But at work, it really does miss that kind of interact. Now, on the other side, I may also be missing when you are working. Because you are working in the shower; you are working--so I think that kind of bleeding in is a difficulty that I think time-use measurement is going to have to review[?] it going forward.
Russ Roberts: So, we're almost out of time. Your work is very interesting; and you are a cheerful fellow--which I really appreciate. But it's kind of deeply depressing to me, some of things that you've documented. Let's put it aside, these issues of causation and how magnitudes--it does raise the question of whether the United States' labor market is increasingly something like Europe's. Maybe for different reasons; maybe for similar reasons. But I think Europe has some real issues, because they have a lot of young people who are not getting into the habit of working; they are not investing in human capital on the job. And I think the social consequences of that are not going to be pleasant. And they won't be pleasant here. And one view says, 'Well, it's just a short-run problem.' I am starting to believe that it's not a short-run problem. And that our education system needs to change very quickly, with some urgency, in terms of preparing for life either without a college degree or life in a world of driverless cars--just something a lot more flexible. Are you as concerned as I am? And, what would you think are important ways we might deal with this, if there's anything to be done at all? Maybe there's no policy to fix it: it's an emergent thing and people will figure out how to deal with it if we leave them alone.
Erik Hurst: I'm going to say two things. One is, I'm optimistic, in the long run, that I do know that things over long periods of time tend to work themselves out. People tend to adjust. Etc. Now, the question is: How long is that long run? And so I am pessimistic that we might be going through a period of time where change is occurring so quickly that it's hard to adjust with the change in real time. So there's going to be--if it's a slow atrophy of jobs and people slowly adjust, those kind of things match up together. But things are occurring so quickly that maybe that adjustment takes long; and then we are going to go through a period like we're seeing--that people are going to be not working. Particularly for those with lower levels of skill, broadly. Not necessarily schooling, but training or whatever you want to call. Whatever you want to kind of add those together. And I don't think that's changing any time quickly. And then when I start thinking about policies, as a person, myself: there's part of me that always wants to be, 'We need to help kind of manage the transition.' But on the other part, I realize that some of the things we do to manage the transition actually just reinforce the problem.
Russ Roberts: Yeah, for sure.
Erik Hurst: So--and, so, figuring out which policy levers to pull to fix this is hard. So, I'm like you, I think, where we have to figure out, in the human capital sector what it is; and then we have to find out what that friction is that's preventing easier human capital. Is it really people don't have money to get the schooling? Or is it people, they don't have access to the training that they want and need? They don't know about it? [?]
Russ Roberts: Or the schooling sector is not competitive enough, possibly--
Erik Hurst: Exactly. Yeah. Maybe when they get to the age of 12, or 14, or 16, they don't have this, yet--so I don't know what those are. But that's where I'd want our focus to be. As opposed to other types of programs, that are well intentioned but have disincentive effects on labor supply.