Much ink has been spilled on how Covid-19 will impact the urban geography of the United States. Early in the pandemic, some were even forecasting the death of the nation’s superstar cities as some urban dwellers fled for the suburbs.
As the year went on, demand for suburban homes fueled questions about whether these moves would be permanent. A June National Bureau of Economic Research paper by researchers from the University of Chicago estimated that 37 percent of jobs can be performed entirely remotely. It emphasized that jobs that can be remote tend to pay more than those that cannot, highlighting yet another disparity in how Covid-19 has affected the labor market.
But there’s a number of reasons so many humans and firms clustered into cities to begin with. Understanding why that is, and the pre-Covid-19 geography of employment, undercuts the likelihood that a significant amount of the American workforce will work remote in the long run.
To understand the economics behind why people cluster in these high-cost-of-living regions and how the pandemic could change that, I turned to Enrico Moretti.
Moretti is an economist and preeminent researcher in the fields of labor and urban economics at the University of California Berkeley. His 2013 book The New Geography of Jobs details the forces shaping where people live, where people work, and how those outcomes are inextricably linked.
In this interview, Moretti explains why high-productivity workers cluster in a handful of cities and why the strength of those forces means it’s unlikely that very many of us will be working fully remotely in the long run. We also discuss why such a small slice of the American labor force can determine so much about which cities dominate.
“I think everything that we know from the economic geography before Covid tells us that these forces of agglomeration are quite powerful. And there’s no reason to think that the same tendency to cluster will be all that different in a post-Covid world,” says Moretti.
The following transcript has been edited for length and clarity.
Something that a lot of urban economists discuss is this concept of agglomeration economies. Can you explain what that is and why it’s so important to the US economy?
Agglomeration economies is one of the most important concepts to understand the geography of employment in the US and the geography of wealth in the US.
Agglomeration economies exist in all sectors, but they’re pretty pronounced in the newer industries, in the innovative industries. It’s the tendency of employers and workers to cluster geographically in a handful of locations. So it is the tendency, for example, of an industry like biotech to cluster geographically in three or four key cities. It’s the same whether you’re talking about social media or pharmaceutical or finance.
I have a new paper where I’m looking at high-tech clusters and I find a staggering amount of clustering when you look at a very narrow level of specialization. So, for example, if you look at all the inventors in computer science, the top 10 metro areas in the US account for 70 percent of all inventors in computer science.
And that number is even larger if you look [at people who work with] semiconductors — 79 percent. If you look at biology and chemistry, that number is more like 56 percent; it’s still incredibly high. So what this is telling us is that there’s a deep-seated tendency of some sectors to cluster geographically. In some of my work and in some other people’s work, it’s emerged that the main reason is productivity.
I think this is one of the key defining features of the economic geography of the US of the past 20, 30 years — in fact, of the economic geography of most industrialized countries — because they all exhibit characteristics of agglomeration.
And can you explain a little bit more about the mechanism by which agglomeration economies form? Is it that a large company, let’s say making semiconductors, forms and then someone who works for that company goes off and makes a startup that does the same thing and he’s already living in the same city?
Or is that all of these companies are consciously moving to be near one another? Or some other mechanism?
Historically, the [first] pattern you described is the correct one. That’s what we’ve seen, for example, in Seattle, which is Microsoft. It is the same for Austin; in Austin there’s a different cluster with some people linked to Michael Dell. It’s the same for the research triangle, you know, Raleigh-Durham.
Now, you’re asking why, why do we see that there is this increased concentration? What attracts people and companies to that cluster? Yes, the channel that you describe is certainly one important one, whereby the alumni of a certain company leave that company and then open their own startup. There are studies that point to how many startups are in Seattle created by Microsoft alumni. But I think there are even deeper reasons; it’s not just that people are leaving the company and sticking around and opening another company.
One microeconomic reason is the matching between labor demand and labor supply, between workers and firms, especially when we’re thinking about very specialized firms and very specialized workers. In larger labor markets, in labor markets which are thicker, where there are many companies for employees and many employees looking for companies — there’s a growing body of evidence that points out that there’s better matching between an employee and a company.
So just to give you an example: If you are a biotech engineer who specializes in a certain branch of biotech and you move to Silicon Valley, where at any moment in time there’s a thousand biotech firms looking for biotech engineers, you might be able to find biotech firms that really value your branch of biotech. That same person moves to Chicago, when at that moment in time there’s a handful of firms looking for employees in biotech; well, you might have to settle for a less good match, a biotech firm that is not really looking for your area of specialization. Notice that it really favors both the firm and the worker. Firms move to the Bay Area and they’re really looking for somebody that is specialized in a certain branch of biotech; and vice versa, it’s much harder for them in Chicago.
And also notice this advantage is not there for unskilled or non-specialized labor. If you are a janitor or a secretary or a welder, the advantages of agglomeration don’t really mean much for you — but if you are a specialized scientist or mathematician or engineer or an innovator, that market thickness will provide a better match. So that’s one important channel that has been documented to improve the productivity both of the firm and the work.
When people talk about high wages in cities, people often think of, you know, tech workers or other people who are working in high-wage industries — can you talk a little bit about the benefits that have been conferred to people not in high-wage industries, but that are still living in these metro regions?
Sure, the vast majority of the US labor force in any city does not work in tech or innovation-intensive industries. Even the San Francisco Bay area, which is arguably the one that has the highest concentration of tech jobs, even here that accounts for a minority of jobs. Typically, in the average US city, about two-thirds of the workers are employed in local services. Whether you’re an Uber driver or a doctor, whether you’re a lawyer or a construction worker, what these jobs have in common is that they reflect local demand.
So they sell a service within the confines of that metro area. And so what you see historically is that when jobs in the innovation sector grow, you see a strong growth in the much broader group of jobs that are in the local service sector, a very large multiplier effect. Because those innovation sector salaries get spent on the local economy and therefore generate jobs for this much broader, much larger, and also much more diverse set of workers.
Covid-19 changed a lot about how and where people work. Industries that thought they could not work from home are working from home. Is it your belief that it’s possible to get the benefits from agglomeration economies, in some industries at least, remotely?
Personally, I don’t think so. I don’t think the economic geography of the US will be profoundly different in the long run, and I think the reason is that I don’t think that we can access those particular advantages that come from agglomeration remotely. When we talk about the long run — I don’t mean, like, next fall; I think about the next few years — I think that once we feel safe, once enough time has passed to give firms and employees time to readjust to the new normal, I do believe that the new normal will look a lot like the old normal.
Right now, if you look at San Francisco, for example, 89 percent of office workers are working remotely. So right now people are claiming that going forward, what you define as “superstar cities,” or high-cost cities, are doomed. I’m skeptical of that, I think everything that we know from the economic geography before Covid tells us that these forces of agglomeration are quite powerful.
So I don’t mean that nothing will be changed. I think that the share of work from home will be higher.
How much higher do you think?
Well I think we can probably agree that it will be higher than before Covid and will be lower than the 89 percent [that we’re seeing in San Francisco]. I think it’s going to be closer to the former — most likely, for the typical employer it’s going to take the form of one work-from-home day a week, or at most two days of work from home a week. And if that’s the case, then what that means is that the economic geography of employment after Covid will look a lot like before Covid.
If you have to show up at the office three or four days a week, you still need to live in the metro area where your office is. The link between place of work and place of residence will be restored and people will flock back to places like the Bay Area or Seattle or New York or Boston for the same reason that they were flocking to these places before Covid.
But since the geography of American cities, as you’ve described, is dependent on a very small slice of individuals — these high-wage workers who are driving demand in a lot of these cities — isn’t what’s most relevant how these individuals will be able to behave?
Before Covid, it didn’t seem possible for me to bargain down my wages and up my ability to work full-time remote, because it was such a cultural taboo. But now that’s no longer the case, so some workers are able to bargain. Is that something that could affect the economic geography of the country even if only a small slice of workers are able to take advantage of it?
My impression is that there’s going to be cases like the one that you described but the main question is that they’re not going to be the modal cases; they’re not going to be the majority of cases, for two reasons.
First of all, for the innovation sector broadly defined, I think they’re going to see quantifiable losses in productivity as measured by quantifiable losses in the amount of innovation these types of workers will be able to create. A lot of the existing research points to the fact that by clustering geographically, these inventors, before Covid, were significantly more productive in quantifiable ways. I have a paper where I quantify the number of patents that an inventor could gain by moving to a tech cluster and the quality of those patents as measured by patent citations. So we’re talking about quantifiable causal effect on productivity and creativity; the moment you start losing that creativity and productivity, that’s when both the employer and employee have something to lose from this decentralized application.
I think the notion of less productivity, less creativity, less innovation, and lower wages is not going to be so appealing for most of them.
And when you say “for most,” you don’t just mean “most of the whole labor force,” but also most of the highest-wage workers?
Correct. That said, I agree with you that some occupations can be probably managed in the long run remotely without huge losses in productivity. Probably that depends, from industry to industry and employer to employer. But I would also point out the second reason for why we saw such a growth in the concentration of high-skilled professionals in the decades before Covid.
So we’ve been talking a lot about labor demand — people moving to superstar cities to get these good jobs. There’s another facet, which is labor supply. A lot of young people actually want to live in these places — a lot of young people were attracted by the urban amenities. Right now it’s not too surprising that places like San Francisco and New York are deserted by a lot of these same people, because right now a lot of these urban amenities are shut down.
Assuming that we can go back to feel safe around each other and the vaccines can manage our safety effectively, I think it’s fair to assume that urban amenities will come back pretty much at the same level that existed before, so [the] labor supply of well-educated workers will keep flowing to these places.
And you mentioned that if there is widespread work from home in these sectors, that it would take the form of a day or two off a week. If that happens, that could significantly reduce commuting time for some people, which may end up pushing people out into the suburbs or the exurbs. Is that what you think will happen, or it’s not enough of a change in commuting time to justify significant moves in that area?
I think it’s a good question. I think if we’re thinking about the health of superstar cities, in particular in the urban core, I think there are two countervailing factors. One is the one that you just said, that it makes it easier for people to live farther away, and on the other hand, if the average worker works from home one day a week, that means 20 percent fewer workers on the freeway or on the subways and less congestion in the city streets. So that means increased attractiveness of the urban core.
So I think that both forces will be at play — one pushing people out and one making the core more attractive in the long run — and I think it’s really way too early. It’s going to take us years to see which one of these two forces prevail. So we will see, maybe in five years, what the data will tell us.