Firehose #168: 🏙 Urban flight. 🏙

Will urbanization reverse? Plus: a backup plan for universities, new experiments in TV, the problem with gluons, and dating in the time of COVID-19.

One Big Thought

File:City-of-the-future.jpg - Wikimedia Commons

I’ve lived in large cities my whole life — first NYC, then Boston, now San Francisco. COVID-19 has given me a new perspective on urban life. This week, I want to share a bit of a thought experiment. Could COVID-19 be enough to blunt the centuries-long trend of urbanization?

Over 4 billion people live in cities today, or 54% of the global population. 2007 was the first year in human history more people lived in an urban environment than a rural one.

To date, urbanization has been synonymous with development. Cities drive significant economic output for countries. Countries with a greater share of urban population tend to have a higher GDP per capita:

The trend is true within countries as well. In the U.S., the GDP per capita of each state also correlates with levels of urbanization:

States with higher levels of urbanization tend to have higher standards of living. For example, average hourly earnings are 30%+ higher in urbanized states like CT, MA, and NY compared to relatively rural states like VT and ME:

More urbanization (somehow) leads to greater economic output. Greater economic output means more demand for employment. More demand for employment leads to higher wages and standards of living. Higher standards of living attract more people into cities and, thus, the virtuous cycle of urbanization continues. Obviously, this model is overly simplistic, but the general linkage is hard to dispute.

The magic happens in the above, parenthetical “somehow.” Why do cities generate more economic output per capita?

One of my favorite theories is the so-called “economies of agglomeration”:

Agglomeration economies are the positive benefits of economic activities that firms obtain from being located in close proximity with those engaged in similar businesses or interests (i.e. agglomerating). It refers to the reduction of business cost as more efficiency and productivity occur because of positive technological and pecuniary externalities arising from the interaction of economic agents located in close spatial proximity due to economies of scale and knowledge spillovers. Certain degree of density is necessary to efficiently share some basic infrastructure and to reduce the user unit costs, particularly for specialized inputs of production including labor skills. Proximity may also facilitate the sharing of new ideas and dissemination of production techniques. Certain spillovers may only be available in large cities such as those caused by diversity or access to international human capital. The incubation functions come more naturally to agglomerative cities. Diversity and specialization of resources and supplier functions are the major externalities of agglomeration.

The logic emphasizing the role of diversity and specialization in enhancing economic efficiency suggests that national growth is enhanced by the combined forces of the heterogeneous features of modern cities and the level of specialization.

There are two types of economies which can be described as external economies of scale: localization economies and urbanization economies. “Economies of localization” arise from many different firms in the same industry located close to each other. The main benefits are: 1) labour pooling allows firms to have access to a variety of skilled labour forces; 2) the development of industries due to the increasing return to scale in intermediate inputs of production; 3) the enhanced interaction, exchanges of ideas, supplies, and labour market due to the proximity.

“Economies of urbanization” arise if economic activities benefit mainly from diversification of industries and/ or coexistence of different industries. Division of labour (i.e. specialization of functions between firms) can increase with the scale of the city. The larger the city, the more specialized operations and services can be formed. Providers of specialized services in large cities can achieve the economies of scale. But cities can not grow unlimited. If cities grow too big, other negative externalities increase such as congestion and pollution.”

In summary, dense cities agglomerate critical, often scarce resources that can be shared collectively. Utilizing these shared resources, firms are free to specialize and innovate on their small slice of the pie. This combination of shared resources and specialization is the magic that makes cities so productive.

Silicon Valley is the canonical example of economic agglomeration. The knowledge of how to build, program, and operate computer components was a shared, scarce resource in the early 1960’s. Programmers and technicians often moved between companies, transferring best practices. This collaboration caused companies to release products with complementary interfaces, allowing them to focus their efforts on innovating in specific layers of the computing stack. Those specialized areas spawned startup companies, some of which created their own categories and ecosystems. Rinse, and repeat.

The resulting explosion of startups in the late 1960’s was chronicled in this genealogy from Walter Isaacson’s The Innovators and continues through today.

A simplified version of this eye chart shows nearly a dozen offspring from Fairchild, including Intel in 1968.

the Fairchild family However, Stanford University, together with local computer chip and microprocessor companies, had already developed another initiative by founding a new R&D centre in the field of Very Large Scale Integrated (VLSI) computer chips, the Center for Integrated Systems (CIS), founded in 1981. Initially, CIS/VLSI was funded by 18 industrial sponsors, who together provided $ 40.5 million, to which the Department of Defense added another $ 8 million for additional research contracts. Stanford University committed to the project by taking on the management of CIS and by training highly qualified electro-technical engineers (100 Master students and 30 PhD's), who would be the first available staff for the sponsors.  

Is physical proximity necessary for economic agglomeration? Clearly, it was in the 1960’s, when Silicon Valley first caught fire. Companies manufacturing servers, chips, and routers needed workers in close physical proximity to build these complex devices. Expensive capital equipment was often required in centralized facilities. Apple, which was founded in 1976 and funded by a Fairchild alum (Doug Valentine of Sequoia Capital), notably still designs its hardware products entirely in Cupertino. Decades later, when much of hardware development was outsourced to overseas supply chains, most developers still preferred to sit in small offices and collaborate face to face.

Two factors may be responsible for breaking the linkage between physical proximity and economic agglomeration. The first is the shift from fixed cost to variable cost development. The vast majority of software startups (with a few major exceptions like Zoom and Cloudflare) build on top of rented infrastructure provided by Amazon*, Google*, or Microsoft. They purchase business tools from Salesforce*, Adobe,* and Slack. The products they build are increasingly distributed bottoms-up, through viral features or simple word-of-mouth in communities. The result is that we’ve now seen multiple companies achieve billion dollar valuations (e.g. Instagram, WhatsApp, Notion, etc) with a few dozen employees and nearly zero physical assets.

The second is the massive improvement in tools for distributed work. My partner Merci Grace, who previously led growth product at Slack, wrote extensively about the market for these products. Nearly all of these companies started within the last 5 years:

These tools make it possible, for the first time, to effectively get work done without physical co-location — at scale. The last truly significant fixed cost for many startups today is an office lease. Multiple scaled startups (Automattic, Zapier, Invision, Gitlab) have demonstrated the fully remote model, negating the need for an expensive office lease. Gitlab even open-sourced its handbook for remote operations.

These two factors together make it possible, for the first time, to agglomerate economic activity virtually. Could the next big tech region be built on the internet? All it needs is one big push.

That push could be COVID-19.

This virus makes cities comparatively unattractive to rural environments. In addition to increased physical density, cities are also socially denser places. Studies show that social connectivity, as measured by the number of phone contacts and total communication activity (K), is related to the population of a city (N) by a superlinear function (K ~ N^1.12). A city like Lisbon, which is 134x bigger than the rural Portuguese town Lixa, has almost 83% higher social connectivity for the average resident.

Figure 1.

Much of the evidence showing how COVID-19 spreads has been focused on person-to-person interactions in social gatherings (e.g. restaurants, churches, workplaces, indoor sports venues). While urban density cannot solely explain the spread of the virus, cities are naturally more dangerous places without active testing, tracing, and isolation.

If the fact that cities create more occasions to catch the virus wasn’t enough, a huge swarth of “non-essential” workers have been placed in a forced Work from Home (WFH) experiment for the last month or so. Many of them are finding it surprisingly effective. According to a survey by Zapier, 65% of respondents felt their productivity had increased. 80% said they could better manage interruptions by coworkers. 80% enjoy seeing their families during the day. Only 42% say they miss their co-workers!

Before the COVID-19 crisis, only 3-4% of Americans did WFH at least half the time. Now nearly 60% are doing so! If a small fraction of those workers are allowed to continue, many of them will choose to move out of urban areas to cheaper, safer suburbs and rural communities. The longer COVID-19 ravages communities, and the longer we swing in and out of “shelter in place,” the higher the odds that workers will seek an alternative to urban environments.

Lastly, I realize that not every company is a technology company. Centralization is necessary for some hardline and service industries, and that won’t change because of better remote work tooling. However, the technology industry is subsuming many legacy industries (e.g. big box retail, movie theaters, financial services). The way it works will be adopted by those who survive — perhaps tautologically. There’s a reason that the 5 largest companies in the S&P 500 today are Apple, Google, Facebook, Microsoft, and Amazon, and that those companies are 20% of the market cap of the index.

market concentration goldman.PNG

In short, COVID-19 has the potential to accelerate the adoption of remote work and put reverse pressure on the previously unstoppable macro trend of urbanization. Hopefully, the infrastructure built by the technology industry will allow the productivity gains of economic agglomeration to continue, as we build new “cities” on the internet.


Tweet of the Week

Links I Enjoy


Reimagining learning online. →

Few industries move slower to adopt technology than education. The need is great during the time of COVID-19, however, with 91% of global students affected. With a likely “second wave” in the fall, the next school year will likely have a false start.

This article in Harvard Business Review argues that digital transformation for universities is now risk mitigation. Those without a technology backup plan may not open at all in the fall.

King of social media. →

Big brands struggle to look “cool” on social media, but every now and then they pull off something genius. Check out this breakdown of a unique strategy Burger King employed to get nearly every social star on Twitter to mention them.


Travis Scott's 'Astronomical' Fortnite Event Was… Overwhelming ...

TV’s brave new world. →

With the vast majority of the country sheltering in place over the last few months, we’ve seen real innovation in media formats tested on much larger audiences. Concerts on Fortnite*, “The Bachelor” on Zoom, and now Twitch is developing reality TV shows for gamers. Its focus will be on natively interactive formats for multiplayer entertainment.

Escape room, anyone?


Image of CLEAR airport security checkpoint

Health ID.

Companies have pivoted their products rapidly to solve problems caused by COVID-19. I’ve used CLEAR for years at the airport. Imagine if these devices could link your identity (confirmed by biometrics) to your COVID-19 health status in a secure manner. I think it would confer a lot more confidence among employees as they return to work. That does assume that we get testing infrastructure to an adequate place, however.


Three particles moving closer together then farther apart. The region between them appears in green.

Stuck like gluons. →

One of the great paradoxes of physics is that we have some very powerful theories that admit very few non-trivial, analytical solutions. The so-called strong and weak forces that hold sub-atomic particles together in atoms are so complicated that you don’t even learn their equations in most college physics curricula.

“The strong force is weird for two main reasons. First, whereas the electromagnetic force involves just one variety of charge (electric charge), the strong force involves three: “color” charges nicknamed red, green and blue. Weirder still, the carrier of the strong force, dubbed the gluon, itself bears color charge. So while the (electrically neutral) photons that comprise electromagnetic fields don’t interact with each other, collections of colorful gluons draw together into strings. “That really drives the differences we see,” Lancaster said. The ability of gluons to trip over themselves, together with the three charges, makes the strong force strong — so strong that quarks can’t escape each other’s company.”

One of the biggest prizes in physics will go to the scientist who can solve those tangled quantum equations that govern the formation of massive entities like protons.


No sex in the Zoom room. →

Shelter-in-place is actually forcing couples to know each other better. Dating apps like Tinder and Bumble are building video chat right into their respective apps. One silver lining of lockdown could be more substantive dating relationships, enabled by technology.

Enjoyed this newsletter?

Getting Drinking from the Firehose in your inbox via Substack is easy. Click below to subscribe:

Disclaimer: * indicates a Lightspeed portfolio company, or other company in which I have economic interest. I also own stock directly in AAPL, ADBE, AMZN, CRM, FB, GOOG/GOOGL, SHOP, SNAP, SPOT, and SQ.

Header image credit: