Can knowledge industries escape superstar cities?
Here's what would be required. It's a tall order.
I am a digital nomad. Bloomberg allows me to work from anywhere; I get much of my best writing done when I’m in Japan, working out of cafes or at a friend’s co-working space. In fact, observing my improved productivity, Bloomberg has repeatedly suggested I move there, and I’m probably going to take them up on that soon.
In the wake of the pandemic, the big question on lots of people’s minds is: How many people will follow the Noah Smith lifestyle? How many engineers, VPs of marketing, HR managers, sales associates, etc. will work out of Tokyo cafes, cabins in Aspen, lounges in Prague, or ranch houses in the Austin exurbs? How many companies will exist merely as a collection of Slack channels, Zoom meetings, and email lists? Will the amazingly powerful clustering economies that have put tech hubs like San Francisco on top of the world be weakened, or will the inevitable economic logic of place reassert itself?
Right now, many people and companies are saying that they’re going to relocate. For example, in a recent post, Kim-Mai Cutler of San Francisco-based VC firm Initialized Capital reveals that her firm’s survey of their founders showed a marked shift between 2020 and 2021:
But Initialized’s founders might not be representative. Cutler notes that her firm has been shifting investment out of the Bay Area for a few years, due to high housing costs. As my colleague Justin Fox has noted, however, SF and Silicon Valley continued to dominate the VC landscape as of 2018:
So far, clustering has just conquered everything. Insane rents and house prices, crazy numbers of car break-ins, massive homelessness, dysfunctional social services, deep social tensions between transplants and natives…nothing could knock SF off its perch. NYC, Seattle, Los Angeles, and other Superstar Cities were less dysfunctional, but they’ve all had their own affordability crises over the past decade, and yet their positions have also seemed unassailable.
Essentially, this has made knowledge industry workers, and the companies that employ them, prisoners of superstar cities. Forced to locate in cluster hubs in order to remain competitive, they have to endure whatever costs the other people in those cities choose to subject them to. Sometimes those costs are reasonable things like taxes, but often they’re things that hurt the city in general, like high rents.
So the question is: Can knowledge industries escape the prison of superstar cities? In order to do so, they’ll have to be able to replace four basic benefits of physical proximity: Office productivity, knowledge spillovers, thick markets, and urban amenities. Call them the Four Jailers.
I like working in an office. But it doesn’t have to be my own company’s office. Here in San Francisco, when there’s no pandemic, I usually work in the office of Bloomberg Beta, the company’s venture capital arm. I can communicate perfectly well with my editors, other Bloomberg writers, and other colleagues electronically. This was true even before Zoom, Slack, and other remote collaboration tools exploded in popularity.
For people whose work is not as independent as mine, the question is whether Zoom and Slack are game-changers. If teams collaborate remotely, will they be 100% as effective as when they occupy the same building? Or only 90%? How important are those whiteboard brainstorming sessions and casual lunch discussions? How important is the ability to wander over and casually ask your colleague something in physical space, rather than messaging them on Slack?
This is a hard question to answer. Lots of studies and business experiments show that flexible work policies — i.e., letting people work from home some of the time — tend to boost productivity. But fully remote work is different than partially remote work; in order to replace offices with distributed workforces, companies will have to get by on 100% remote work, not 30% or even 80%.
The pandemic, of course, has caused many companies to move to 100% remote work. But its effects are confounded by all sorts of stuff — the stress of the disease, the fact that companies had to go remote with no warning or planning, the lack of availability of co-working spaces, etc. So far the results are inconclusive, with various surveys giving conflicting answers.
But it’s also important to remember that companies don’t have to go fully distributed in order to mostly escape the pull of the superstar cities. They just have to shrink their main offices, and manage the rest of their workers either remotely or at branch offices in cheaper places. If 50% of knowledge industry employees — the Noah Smiths of the world — are fine working remote, and the rest need onsite collaboration or tech support or whatever, that will still make a big difference. So between that fact, the amazing progress of tools like Zoom and Slack, and all the studies showing productivity boosts from partial remote work, I’m actually pretty optimistic that office productivity won’t be a big obstacle to the distributed future.
If you’re an engineer in San Francisco and you lose your job, it’s easy to find another one in the same area without moving. If you’re a company in San Francisco and you lose your employee, it’s easy to find another good one in the area. This is called the “thick markets” effect, and it’s another reason industrial clusters exist.
In order to escape the cluster cities, both businesses and workers will have to learn to do job search and hiring mostly remotely (except for flyouts for a few very high-value workers). You’d think that if remote work doesn’t damage productivity much, then remote hiring won’t either, but we should at least allow for the possibility that these are different in some way.
Another type of thick market effect is finance. Venture capital firms are still way too obsessed with being physically near to their portfolio companies. I don’t know of any study on this, but it seems like mostly or entirely a cultural thing. In fact, this is a negative feature of the VC industry with or without distributed workforces, since it robs second-tier cities of needed financing and contributes to a winner-take-all effect among American regions. So hopefully more VCs will follow the example of Initialized, and learn to invest remotely, whether or not companies become distributed.
Anyway, I’m fairly optimistic about this one too.
This one is going to be harder. One reason knowledge industries have such strong clustering effects is the existence of knowledge spillovers. If you’re an engineer in San Francisco, you’re going to meet a lot of other engineers, including some very smart ones, in the course of your daily life — whether that’s at coworking spaces, karaoke, house parties, talks at universities, or wherever. If you are an engineer living in the frozen wastes of Kamchatka, you will not meet other engineers, except online.
It’s hard to confirm this effect with data. Economist Enrico Moretti found that inventors in tech hubs produce substantially more patents than similar inventors elsewhere. Other research finds much the same. if patents are a reasonable proxy for innovation then this means that being in a superstar city yields an innovation boost. But some researchers argue that patent boosts are not a good measure of knowledge spillovers — for example, they may simply indicate that when companies are close to each other, they spend more effort trying to use IP law to block the competition from copying their ideas. So it’s hard to say.
But if knowledge spillovers are a big deal, then it will be very hard for businesses to overcome this hurdle and go remote. The reason is that knowledge spillovers are an externality — they often happen outside of a company, so they’re not something a company can control. It might be that in the age of Stack Overflow, knowledge spillovers are already shifting online, so maybe this isn’t a problem. But if Stack Overflow isn’t enough — if better online interaction forums are needed in order to take knowledge spillovers completely virtual — who will make those forums? What if it’s technically feasible but just not monetizable? This is a public goods problem, and it’s not clear if private companies will be motivated to solve it. Maybe Stanford University can create some worldwide virtual version of the networking events it used to hold for Silicon Valley engineers, back in the early days of Silicon Valley.
This is the final hurdle, and — I predict — the hardest one to solve. That doesn’t mean it’s impossible, though.
Cities are not awesome for everyone. If you want to have a McMansion with tons of space, or quiet empty leafy streets, or a big driveway with three cars, or a forest right outside your back yard, a city is not for you. But for many people, the amenities of cities are hard to resist. For young people, this means bars, music venues, fun social events, lots of potential friends in their age group, and — probably the most important piece — opportunities to meet romantic partners. For older people this means a large variety of good restaurants and cultural events like musicals.
For this reason, distributed workforces might just mean that knowledge workers live in different superstar cities. For example, when I started my current remote-work job, I moved from New York City to…San Francisco. And now I’m probably going to move to…Tokyo. All big knowledge industry hubs. They even all have Bloomberg offices!
For knowledge industries to actually escape the superstar cities, some of their workers are going to have to move to second-tier cities. They’re going to have to go live in Dallas, in Tulsa, in Sacramento. They’re going to have to live in Sendai instead of in Tokyo. Can they be as happy there?
They can certainly be richer. Cutler’s post argues that companies will not force employees to take salary cuts if they move. A $100,000 salary goes a lot farther in Tulsa than it does in San Francisco!
But that still leaves the question of all those juicy urban amenities. Technology and new business models — along with local government initiatives — will have to make living in Tulsa tolerably fun and exciting for young and old workers alike.
Dating might be the easiest piece to solve. Meeting partners is just a matter of matching, and apps are getting pretty good at this. You can meet as many people on Hinge or Bumble or Tinder in a suburb these days as you could going out to bars back in the 80s. Maybe more.
Friends will be a bit harder. Social networks all try to be social discovery tools, and with the odd exception of Twitter, I think they largely fall short. So social apps will have to get better at connecting groups of people with shared interests and compatible personalities. And remember that when these apps get better, it also makes life in superstar cities more attractive as well.
Restaurants will also be hard. Apps for sourcing of ingredients, digital sharing of menus, etc. will allow any city of reasonable population to have a lot of good places to eat. And if knowledge workers spread out to second-tier cities, their discriminating palettes will create demand for innovative eateries. The real problem with restaurants is density; you can spend a lifetime trying and failing to go to all the ramen shops in Tokyo, while a couple years of going out to eat in College Station, Texas will familiarize you with all the good places in town. But for cities with relatively large populations, this is probably less of a problem.
As for cultural events, theater was never really my thing, so here I can’t say.
The great jailbreak
Superstar cities have become a one-stop shop for so many different things that people want out of society — productive work, job opportunities, friends, romance, fun. We’re definitely not going to go to a workforce of hermits living in cabins in Kamchatka. The real question is whether distributed work will allow second-tier cities like Tulsa to take enough knowledge workers to bring them some of that superstar energy, thus relieving the pressure on places like San Francisco. Even that is a tall order, since the top cities exert many different kinds of gravitational pulls all at once. Escaping these cities isn’t like building a rocket — it’s like building four different rockets and riding them all at once.
So…maybe it can happen. It’s always worth trying. But it’s going to be tough.
You ask: what will replace the organized meetups at Stanford in the early Si Valley days? Well, most technically complex scientific communities depend on annual conferences with annual attendances of 5000, 10000 people and beyond. Conference venues for that size group require 2-4 years advance booking and have hefty fees for cancellations in the last 6-12 months. So 2020 was a disastrous year as all had to go virtual, and some had little time to redo arrangements. The result was a big drawdown of "rainy day" funds. The one I am involved with, NeurIPS (which covers from neuroscience to machine learning, AI, robotics etc, and is a premiere recruiting opportunity for corporate sponsors), was already starting to distribute its satellite activities by adding virtual meetups. A year of being fully online for a week with a schedule each day that was designed to be accessible from Beijing to Moscow has greatly accelerated the transition. The results for the December 2021 meeting are still under debate, but even under a better than expected world recovery from pandemic and populist insanity, I expect that the hybrid conference design will evolve further. A critical question is whether this will serve the recruiting and image-boosting needs of major sponsors, whose support has provided the cushion that got several conferences through the shutdown.
One factor that's not mentioned is tolerance. The nativist attitudes of many in low-density areas of the country are a barrier to their regional development (even if the central cities are tolerant, they are often small islands in MAGAland). A friend of mine did a series of interviews with business leaders in NYC. One finding that jumped out of his research was that execs felt that they could get any talent, from anywhere in the world, to relocate to NYC if they needed them and that was a prime reason they intended to keep their operations there.