When economic data appear in the media, they are generally discussed as national statistics; for instance, a given number of jobs were added across the country or the economy as a whole grew by a certain percentage during the past quarter. Those data can yield useful information, but they can also mask important regional variations and trends, argues Erik Hurst, an economist at the University of Chicago’s Booth School of Business.
For instance, during the housing boom of the previous decade, employment was particularly strong in certain areas of the country, helping overall employment numbers and obscuring structural decline in employment in other areas of the country. As a result, when the Great Recession ended, it shouldn’t necessarily have been a surprise that job growth would be relatively mild, because those areas that had boomed went back to their pre-boom trends while those that were in decline continued to struggle.
Hurst has used regional data in a series of papers to look at other macroeconomic phenomena that would be hard to examine using national data alone. He also has done important work on household financial behavior — including consumption and time use over people's life cycles — and on labor markets. Business startups have been another interest of his: Much has been written about the importance of entrepreneurship to the U.S. economy, but what, he has asked, actually motivates people to open their own businesses? In addition, in a recent paper, he and co-authors have attempted to quantify how much the decline in barriers to employment of women and minorities has contributed to economic growth. Among his current research interests is explaining the decline in labor force participation among prime working age males.
Hurst joined the Chicago faculty in 1999 after completing his Ph.D. He is currently co-editor of the Journal of Political Economy and serves on the board of editors of the American Economic Review. He is also a research associate at the National Bureau of Economic Research, where he is a member of the public economics, economics of aging, and the economic fluctuations and growth programs. Aaron Steelman interviewed Hurst in his office at Chicago in April 2016.
EF: Many people have an image of the typical entrepreneur in their head and it often includes a significant taste for risk and large long-term aspirations. What does your work on entrepreneurship suggest about that profile?
Hurst: In some ways, I'm perceived as the anti-John Haltiwanger when it comes to entrepreneurship. He and I are often on the same panels, and he's definitely seen as the glass half full guy and I'm definitely the glass half empty guy. But I think we both are right. His line is there's a huge amount of job growth that comes from these new small entrepreneurial businesses and that's very important to the dynamics of the U.S. economy, and I agree 100 percent. My line has always been essentially that most small businesses simply don't grow. But that doesn't mean those statements are necessarily inconsistent with each other.
Most small businesses are plumbers and dry cleaners and local shopkeepers and house painters. These are great and important occupations, but empirically essentially none of them grow. They start small and stay small well into their life cycle. A plumber often starts out by himself and then hires just one or two people. And when you ask them if they want to be big over time, they say no. That's not their ambition. This is important because a lot of our models assume businesses want to grow. Thinking most small businesses are like Google is not even close to being accurate. They are a tiny fraction.
My work with Ben Pugsley has been emphasizing the importance of nonpecuniary benefits to small-business formation. Because when you ask small-business people what their favorite part of their job is, it's not making a lot of money. They do earn an income and they're very happy with it, but they get even more satisfaction from being their own boss and having flexibility and all of those other nonpecuniary benefits that come with being the median entrepreneur in the United States.
In our culture we seem to want to subsidize small businesses because it's the American dream. I think that could be fine, if you believe that there's a friction out there preventing some small businesses from starting or growing. But you might want to target that friction directly as opposed to targeting all small businesses generically. Ben and I have a recent paper in which we show in a simple model that if you subsidize all small businesses, in a world with nonpecuniary benefits being the big driver of small-business entry, the policy is highly regressive. Why? High-wealth people are already small-business owners because they can afford the nonpecuniary benefits that come with owning a small business. So in that world, we're basically just transferring money to high-wealth people when we subsidize small businesses overall, and that's something we need to consider.
EF: Income inequality has been a widely and hotly discussed issue recently. You have looked at a related issue that seems relevant to this discussion: the relationship between the wealth of parents and those of their children.
Hurst: There is a high correlation of parental wealth with child wealth — and it is very highly correlated at the tails of the distribution. That is, children of very low-wealth or very high-wealth parents rarely end up with wealth substantially different from their parents' wealth. This alone may not be surprising to a lot of people, but what may be surprising is that it is true above and beyond the correlation in their earned income and before bequests are received. We demonstrated that with data from the Panel Study on Income Dynamics. Subsequently, many people have used other and often better datasets to look at the issue, and they are coming up with roughly the same findings that we did.
So where is that residual coming from? Part of it is that people often marry people who are like them and they join their wealth. I also believe that saving propensities between parents and children are correlated and that they tend to allocate their portfolios similarly. That may be because children often have similar preferences as their parents, or it may be due to mimicking of behavior unrelated to preferences. I'm not certain which of the two is more important but it's an interesting question.
EF: If the marital sorting story is true and the importance of education to earnings is increasing, this may suggest that we could expect additional stratification.
Hurst: I think there is a good chance of that. Many people find their mates at college or shortly afterward. And if people who are going to college still sort there or in the jobs they move into after college, there is going to be more segmentation, with relatively higher-skilled people finding mates in one pool and relatively lower-skilled people finding mates in another. That is naturally going to propagate inequality, probably even more so than we have seen in the past.
EF: I would like to talk about your paper "The Aggregate Implications of Regional Business Cycles" with Martin Beraja and Juan Ospina. What was the motivation for that paper, and what were the principal findings?
Hurst: We came at this paper from two different perspectives and kind of merged them together. One is my work exploiting regional variation to try to learn about the drivers in the aggregate macroeconomy. Local elasticities need not be the same as aggregate elasticities and without acknowledging that I think, as macroeconomists, we can get some misleading results. Second, I was very interested in what was causing the Great Recession.
So what we did in this paper was to construct a bunch of different regional economies that aggregate up to the overall economy. We embed in that model three different shocks. We have one that looks like a household discount rate shock. We then put in a shock to firm productivity/markup. And then we also have a shock to people’s taste for leisure. All three of those shocks have local and aggregate components to them, and we use local data and aggregate data to tease out the relative importance of these local and aggregate shocks.
We know aggregate wages didn’t move that much during the Great Recession. We know aggregate prices didn't move that much either. In the paper, we construct wage series at the local level using household surveys. More importantly, we construct price data at the local level using scanner data. And with those data we come up with some new facts.
The facts are real wages moved very strongly with employment across regions. Nevada was hit very hard by the recession, for example, while Texas was hit much less hard. Wage growth, both nominal and real, was about 5 percent higher in Texas than it was in Nevada during the Great Recession. So if you're going to just correlate employment movements and wage movements, both real and nominal, at the local level, you see a pretty strong reduced form correlation.In the aggregate time series, you don't. Wages hardly moved at all despite employment falling pretty sharply. So there are some differences in the correlations between wages and employment at the local level and the aggregate level during this recession, and that's the fact on which the paper is based.
In the second part of the paper, we have optimizing consumers, we have optimizing firms, and we have a monetary authority in the background that is following a Taylor rule. We also have one nominal rigidity in our model: sticky wages. And we have those three shocks I mentioned earlier: one to the household discount rate, one to firm productivity/markup, and one to people's taste for leisure. Then we ask what would be the effect of a local shock — for instance, the discount rate shock — on local employment and ask how that compares to a model-implied aggregate discount rate shock on aggregate employment. Are they the same? The answer is that it's not even close. The local elasticity in our calibration is about three times higher than the aggregate elasticity of the same shock.
EF: From there, you should be able to use the local data to estimate the aggregate shock.
Hurst: That's right. And instead of doing reduced form elasticities like others have, we try to estimate a structural parameter, wage stickiness, and apply it to the aggregate economy. So we use our model to estimate the extent of wage stickiness at the local level and imply that parameter of wage stickiness in our aggregate decomposition of the shocks. What we find is that wage stickiness at the local level is modest. If wages were really as modestly sticky as we estimate from the local level, wages should have fallen at the aggregate level given the large decline in employment. But we don’t see wages falling, so that tells us there must be a labor supply shock in the background.
We also have decomposition for prices. If it were only a demand shock, prices would have fallen. But prices didn't, so that tells us there’s some productivity shock in the background. So we use our model in this estimated parameter to quantify the relative contributions of the discount rate shock, the productivity/markup shock, and the leisure shock in the aggregate economy. And we find that the discount rate shock could explain a pretty significant share of the decline in employment in the United States during 2008 and 2009. But it explains close to zero of the persistence — why employment remained low from 2010 through 2012. Instead, you have to look at the productivity/markup and labor supply shocks to explain that.
EF: This suggests perhaps a different narrative about the lack of wage growth than we sometimes hear.
Hurst: When people say the reason we haven't seen real robust wage growth in the recovery is because wages were so sticky in the beginning period, I just don't think that holds water with the flexibility of wages that we see at the local level. In addition, real wages have been roughly constant in the United States since 2000. So even though the economy was growing between 2000 and 2007, we still didn't see significant wage growth.
EF: I think that gets us to your work on structural change in the labor market, including your paper on how the housing bubble masked the decline in manufacturing employment.
Hurst: The way I usually describe the paper, which I wrote with Kerwin Charles and Matt Notowidigdo, is to take two regions, Detroit and Las Vegas. Las Vegas has very little manufacturing relative to Detroit. Detroit didn’t have a big housing boom but Las Vegas did. Now there are a whole bunch of different kind of stories about what caused that boom: low interest rates, extension of credit to subprime borrowers, and potentially bubble-like behavior in some places. When you look at this early 2000s period, if you focus only on Detroit, you see employment rates going down, particularly among prime-age workers. It looks like there was a structural decline in employment well before the recession ever started. When you look at Las Vegas during the boom, the employment rate was well above long-run local averages. Normally, most people in their 20s and 30s work, but some of them don’t. During this period in Las Vegas, among lower-skilled workers in their 20s and 30s, nearly everybody was working. So when you put aggregate statistics together, when you sum together Detroit and Las Vegas, it looks like employment rates were relatively constant over this time period. But one was really low compared to historical levels, and one was really high relative to historical levels.
In this paper, we show that the decline in manufacturing that occurred during this period nationally — when you add in the Detroits, the Worcesters, and the Youngstowns — was masked by the aggregate housing boom in places like Las Vegas, Phoenix, south Florida, and some places in California that were growing well above average. Now one of these was temporary and one isn’t. The housing boom we know busted and then employment in Las Vegas plummeted. If you look at 2010 or 2011, the employment rate in Las Vegas is roughly the same as it was in 2000, meaning it increased and went back to trend, where the old manufacturing centers just continued declining relative to their 2000 level. You have a very temporary boom-bust cycle overlaid with a structural decline, and what you get is kind of a hockey stick pattern for the aggregate.
So for macroeconomists looking at the Great Recession, this is important for understanding why the employment rate hasn’t bounced back to its 2007 level. It shouldn’t have because 2007 wasn’t a steady state. In terms of policy implications, this means that monetary policy arguably is not an especially effective tool for strengthening the labor market. Instead, I believe you need to focus on retraining workers or investing in skills in some form. You might also want to look at disability and some other government programs that might act as a drag on unemployment.
EF: In addition to decline in total manufacturing employment, the skills required to work in the modern manufacturing sector have changed a lot as well.
Hurst: They have. The new hiring that’s going on in manufacturing is moving up the skill ladder more than it was before. Much more precision work is required now. My father was in manufacturing, and he would not be able to find a job in manufacturing as it is now. So I think the structural change in manufacturing has left some of these workers on the sideline. But I think this is true not only of manufacturing -- it’s also true of many administrative and service jobs. In these areas, people have moved down the labor supply curve, wages have been relatively stagnant, and people have chosen to leave the labor force.
Something that I like to stress is that it is not demographics. The aging of the population explains a little bit of it. But all my work is looking at people in their prime age — say early 20s through mid-50s. There is a structural problem for prime-age, lower-skilled workers in the economy. If you take a look at people with a four-year college degree, you can barely see the effects of the recession any longer. There’s been no lasting effect on their employment rate. Almost all of the effect is concentrated among people with less than four-year college degrees.
EF: Given the wage premium associated with a four-year degree and the availability of education financing, it seems like a real puzzle why more people are not obtaining degrees.
Hurst: I have been thinking a lot about that. What is it that’s causing so many young people, particularly young males, to not obtain skills required to be successful in today’s workforce? I have been working with Mark Aguiar and Kerwin Charles and Mark Bils to try to understand what these people’s lives look like. There’s a budget constraint that still has to hold. They have to eat. What you’re finding is that a lot of them are living in their parents’ basements or their cousin’s basement. So many are relying on family support. And a lot of them just aren’t even working at all. So when you go and take a look at the fraction of people in their 20s who haven’t worked in the prior 12 months in 2015, it’s 20 percent for men with less than a four-year college degree. In 1990, that number was 4 percent. So the first thing we are doing is documenting these facts and trying to find out what their lives look like: how they’re eating, what their living situations are like, what attachment they have to the labor force.
The second part we’re trying to think about is why. What we are considering is whether it’s possible that a leisure lifestyle is easier now in your 20s than it was in the past. In 1980, if you were in your 20s and you weren’t working, you were pretty isolated. You were sitting by yourself. You could watch a few channels on TV but no one else was out there. Now if you’re not working, you could be online on social media or you could be playing videogames in an interactive way, things that make not working more attractive than before. And those videogames and leisure goods generally are relatively cheap compared to what they were in 1980. So when you’re making your choice of working relative to your reservation wage, your reservation wage has gone up some because the outside option of not working is a lot more attractive. So that’s what we’re thinking but I don’t know how we’re going to test it.
Also, eventually these people will get older, of course, and many will have a spouse or kids. When that happens, their income requirements go up and they need jobs, but they probably haven’t been building the type of skills required to get a job. So that’s hard to understand. I have never written a paper before where people were myopic, but the behavior of a lot of people in their 20s now seems myopic.
EF: Getting back to the housing boom, you have looked at its effects on college attendance. What did you find?
Hurst: We’ve talked about the housing boom and how it masked some important trends in the labor market. I want to see if it is possible that big housing booms could actually alter productivity going forward. One of the first steps in that is my paper, with Kerwin Charles and Matt Notowidigdo, on the effect of the housing boom on college attainment. College attainment rates for both men and women were relatively constant during the housing boom, which is different from the previous 50 years when they had been increasing. We argue that the housing boom may have caused some people not to accumulate human capital because of the labor market effects that we have talked about — meaning, in places like Las Vegas the housing boom created a lot of employment opportunities for lower-skilled people and raised the opportunity cost of attaining skills. Now, did those people go back to school when the housing bubble burst and the labor market weakened?
We can follow those people over time and ask: Did people who were in the labor market in their late teens and early 20s in Las Vegas in 2000 go back to college in 2010, when they were now about 30 instead of about 20? They didn’t. It’s not too surprising because 30-year-olds hardly ever go to college. So what the housing boom did was not only mask the secular decline in manufacturing and routine jobs in the economy, it has also somewhat altered the college attainment profile of people a decade later. There are now a lot of 30-year-olds with less schooling than they would have had otherwise, and we are going to be carrying around the legacy of the housing boom on formal college education going forward. To the extent that we believe a little bit of human capital makes people a little bit more productive, particularly in the labor market we’re in now, this is going to be a cost.
EF: When economists analyze time use, something that doesn’t seem quite clear is whether spending time with kids should be counted as home production or leisure.
Hurst: That’s a hard question. If I have you raise my kids for me, I’m going to save some time. But there are certain kinds of utility flows that I could only get by being around my kids. What Jon Guryan, Melissa Kearney, and I show empirically in a paper from a few years ago is that if you look at the income gradient of how we spend our time, the richer you are, the less home production you do. But the richer you are, the more childcare you do. So that income gradient between home production and childcare has opposite signs, which tells me it’s not exactly the same good. Whether that’s coming from the utility you get from being with your kids or whether it’s from investing in their human capital, that’s hard to say. We know people from high-income families go to school more, go to the doctor more, and spend more time with their families. So how much of it is investment, how much of it is home production, how much is leisure, I don’t know.
I have always advocated that you should have four uses of time — market work, home production, taking care of kids, and leisure — and then treat kids as somewhere between leisure and home production. So we tend not to put it in home production when we measure it, but we don’t put it in leisure either. We try to treat it as its own category.
EF: Could you describe the idea of endogenous gentrification you have developed with Veronica Guerrieri and Dan Hartley?
Hurst: I’m interested in housing markets, which are inherently an urban phenomenon. My work with Veronica and Dan has been to try to understand housing price variation within a city. Many urban models historically assumed that agglomeration benefits usually came from the firm side. Someone might want to be close to the center city, for instance, because most firms are located in the center city. So the spillover for the household was the commuting time to where the firms were, and the firms chose to locate near each other because of agglomeration benefits.
I have always been interested in it from another angle. When we all come together as individuals, we may create agglomeration forces that produce positive or negative consumption amenities. Thinking about it this way, when a lot of high-income people live together, maybe there are better schools because of peer effects or higher taxes. Or maybe there are more restaurants because restaurants are generally a luxury good. Or maybe there’s less crime because there is an inverse relationship between neighborhood income and crime, which empirically seems to hold. So, while we value proximity to firms, that’s not the only thing we value.
How important are these consumption amenities? And more importantly, how do these consumption amenities evolve over time, because usually in our urban models we assume that there’s an amenity in a place and that amenity is relatively fixed? For instance, we like nice weather in Southern California and we hate bad weather in Rochester. Those amenities are relatively fixed. But in a world where amenities evolve over time, the composition of people in the neighborhood could then affect amenities, which then affects house prices and further affects the movement of people into and out of those neighborhoods. What we asked is: If a city experiences a housing demand shock, what types of neighborhoods appreciate the most? We found, as our model predicts, that gentrification spreads out from neighborhoods that are already gentrified — that poor neighborhoods on the border of rich neighborhoods experience the largest increase in house prices.
EF: As we know, women and blacks have faced substantial barriers to employment in the United States. In many ways, those barriers have become less significant. How much did they impede aggregate economic performance?
Hurst: In 1960, very few women and very few blacks were in higher-skilled professions. If we believe that at birth the propensity to be a good doctor in terms of our talent draw is equally distributed by gender and race, we should see relatively equal propensities, but we haven’t, although those propensities have converged dramatically between 1960 and today. So there are two questions we want to ask. One, what factors might have caused a wedge between people’s occupational choices in 1960 and how might they have changed over time? Two, can those changes actually affect aggregate growth?
In a paper with Chang-Tai Hsieh, Chad Jones, and Pete Klenow, we use a Roy model to get at these issues. People are going to draw talent in a whole bunch of different occupations and, for the most part, the talent draws are going to be roughly similar between men and women and blacks and whites. We do allow for brawn-type occupations, where men might have a comparative advantage. Men might have been better construction workers in 1960, and that might have changed by 2008 because men and women can equally drive a forklift. But our assumption is that for most occupations, people’s talent draws are the same.
So why would men and women and blacks and whites differ from each other in their occupational choice? We have a few types of wedges. One is discrimination in the labor market. Women and blacks were discriminated against being, say, doctors in 1960, and that discrimination has changed over time. Partners in a medical practice, as well as their customers, are now less likely to see women and blacks as being unable to provide identical services as men and whites. Second are barriers to human capital accumulation among women and blacks. Those explicit and implicit barriers are things like segregation or underinvestment in schools in black neighborhoods, prohibitions on entry of women to certain professional schools, or social norms that steer women toward some occupations and away from others. Third are preferences. Perhaps women and blacks opted out of going into certain professions because of social norms, and they were willing to take a utility loss to not run up against those norms. Fourth are factors that affect home production and have increased labor market flexibility for women over time. This would include labor-saving devices such as dishwashers and washing machines as well as improved methods of birth control that permit greater control over fertility decisions.
So we have these barriers and we want to distinguish among those to see how much of the occupation choice differences in 1960, 1970, 1980, and so on are due to these barriers — and as a result, how much does the decline of those barriers contribute to economic growth.
We find that about 30 percent of growth in the United States between 1960 and 2010 is due to declining labor market barriers for blacks and women across those four areas. That’s a big chunk of growth, and most of it is due to increasing participation of women simply because there are many more women than blacks in the population and the labor force. And for women, almost all of it is due to changing barriers to human capital. For black men, it’s about half discrimination in the labor market and half human capital barriers. For black women, human capital barriers play a larger role than discrimination.
Now, there is still progress to be made in these areas, which will yield gains. But there is good reason to think that many of those wedges have lessened substantially, and if that’s the case, the United States is not going to experience the type of growth that it did since 1960.
EF: What are you working on now?
Hurst: I think the question about why labor force participation is so low among people who are of prime working age is really interesting and I want to investigate that. Another issue I am examining is wage stickiness. I’m going to try to get administrative data from a payroll-processing company to try to measure wages and hours for a large section of the economy and then ask: How sticky are wages? Or how sticky were wages during the recession? I sort of want to do for wages what Mark Bils and Pete Klenow did for prices. Part of the reason this has been hard to answer is we don’t have really good administrative data on wages; we have good administrative data on earnings, but all the fluctuations in earnings could be due to hours. So we need to isolate how much people’s earnings are due to firms changing wages and how much of it is due to people or firms changing hours. We don’t know the answer to that. But having actual payroll data would allow us to separate the wage and the earnings parts. From there, you could ask a lot of interesting regional questions, such as: Were wages stickier in Las Vegas, where there was a big labor boom, or were they just as sticky in Dallas?
EF: Which economists have influenced you the most?
Hurst: The three people I try to keep in my mind all the time are Gary Becker, Kevin Murphy, and Bob Hall. If I could be a fraction of those three combined as I get older, that would be wonderful. They always start with economic theory, pretty straightforward price theory, and then go to the data, Kevin and Bob particularly, and test their theories without using overly fancy tools. They are true empiricists in the best sense. Plus, they all just love economics and that comes through in their work.
Bob will strongly believe in X today, and if tomorrow somebody shows him a piece of evidence that it’s not quite X but it’s X prime instead, he moves with that as well. I love that fluidity as a scientist. It’s not like he has a stake in the race; he just wants to know the truth. He’s also well into his life cycle and still producing at the frontier, just as Gary did. I’m with Bob sometimes and he still gets irritated when he has a paper rejected. I want to be 72 and still get irritated when I’m rejected.
Also, all three of them were so generous with their time with me. When I got to Chicago, I was working on discrimination issues, such as wealth gaps between blacks and whites. I had been on campus probably just a month or two and emailed Gary and spent an hour and 15 minutes talking about a paper of mine. I hope to be that generous with junior faculty as I move along in my career as well.
V. Duane Rath Professor of Economics, Booth School of Business, University of Chicago; Research Associate, National Bureau of Economic Research
B.S. (1993), Clarkson University; M.A. (1995) and Ph.D. (1999), University of Michigan
“The Masking of the Decline in Manufacturing Employment by the Housing Bubble,” Journal of Economic Perspectives, 2016 (with Kerwin Charles and Matthew Notowidigdo); “Deconstructing Life Cycle Expenditure,” Journal of Political Economy, 2013 (with Mark Aguiar); “Endogenous Gentrification and Housing Price Dynamics,” Journal of Public Economics, 2013 (with Veronica Guerrieri and Daniel Hartley); “Marital Sorting and Parental Wealth,” Demography, 2013 (with Kerwin Charles and Alexandra Killewald); “What Do Small Businesses Do?” Brookings Papers on Economic Activity, 2011 (with Benjamin Pugsley)