While I was in Mexico City, the most populous city in North America, I couldn’t help but think about network effects. As a technologist, I’m quite familiar with network effects on technology platforms like Facebook. But what about cities? Do cities produce positive network effects?
Let’s start with a definition. A network effect means each additional user increases the value of a service. For example, each additional user on a social network (like Facebook) makes it more valuable to the other users. Simply put: if your family is on Facebook, you’re that much more likely to use it and see ads.
Ben Thompson has a bunch of great articles about network effects, for example - The Moat Map.
One of the best metrics to measure whether a platform takes advantage of network effects is the ratio of Lifetime Value (LTV) and Customer Acquisition Cost (CAC). LTV is an estimate of the average value a user will create throughout the whole time they’re going to use the platform. CAC - how much it costs on average to acquire each new user. If a platform takes advantage of network effects, LTV/CAC ratio should increase.
To understand network effects in cities, let’s start with finding a metric. It would be simpler if we could measure LTV for a city. Alas, in a world with distributed companies, that’s quite difficult.
Instead, we can use the average salary. It indirectly captures the economic growth of a city and the average value each of the citizens produces. It’s not flawless as metrics go by any means. Salary depends on the competitive environment for talent and has no direct relationship with created value. But the average salary is as close as we can get to the average value created.
So, is there a relationship between city size and salaries?
Luckily, I found that there’s a bunch of research on this topic. I’ve examined several research papers. They’ve ranged from 1978 to 2011: Income Distribution, City Size, and Urban Growth, Urban Wages: Does City Size Matter?. In the ones I’ve examined, the findings were clear: there’s a positive relationship between city size and wages. Quote from the 2011 research paper:
every additional 100 000 inhabitants in the local labour market raises individual hourly wages by 0.12 per cent. Moreover, a doubling of the human capital density in a metropolitan area results in approximately a 2 per cent increase in average individual hourly wages.
The next question that came to me: does city specialization strengthen network effects in a city? Logically, it should. There are immediate examples that come to mind. Inflation-adjusted income in San Francisco, which specializes in tech, grew 20% in the last 15 years. While there are other tech cities, salaries in San Francisco are exceptionally high.
More people in an area skilled or knowledgeable on a single topic means better knowledge sharing and allows them to invest in better infrastructure for a specific speciality. If a city is specialized in shipping, it makes sense to invest in a great shipyard to continue taking advantage of the network effect.
There’s some research on this topic: Urban Specialization in the World-System. It supports my hunch that city specialization means good things for a city. But there are too few research papers for me to feel fully confident.
Negative network effects also exist. The current COVID-19 pandemic is excellent, and at the same time, a harrowing example of how cities create adverse network effects. If you live in a city, you’re more likely to get the virus. More people in a single location means more traffic and worse air quality. The inability of the already-mentioned San Francisco to invest in housing is another example and drove a well-known housing shortage. More people in a single location is not all good. Investments in the right infrastructure are required to keep the negative effects at bay.
People can also become stuck in a city due to a moat created by network effects. Let’s say that we have a city which specializes in car manufacturing. Great for a while. But then, car manufacturing becomes less important, and there are fewer jobs available. People might not want to leave a city due to network effects - for example, because their whole family and friends are there. Even if, moving away from the location for a big group would be beneficial.
It’s similar to how technology platforms create defensible and enduring moats. There might potentially be better search products than Google out there. But Google takes advantage of more people searching to create the best search results. To give another search product the best chance to compete, a lot of us would have to leave Google to allow another product to strengthens its network effects. We’ll be trapped in Google’s city longer than we’ll need to be.
Knowing all this, what could we do?
Currently, there are more than 30k technologists in Lithuania. 70% of them work in Vilnius. Most of the rest of the 30% - in Kaunas.
Now, before I write the next paragraph, a couple of disclaimers. I love Kaunas. I enjoy visiting, walking down the Freedom Alley and hopping into “Spurginė”. I root for “Žalgiris”.
Considering the network effects, it could be better for Lithuania in the long-term if we’d focus investments into Vilnius. That’s not what currently happening. For example, Invest Lithuania wants to bring more investments into Kaunas. That’s good and better than those investments not arriving at all. But it might not be the right choice long-term. Lithuania might be better off if we’d focus on making one city great, instead of two cities good. Though, emotionally (including for me) that would be a tough choice to make.
Now when everyone in Kaunas hates me… Let’s not forget that network effects are not everything. But they are a lot. Network effects make cities better problem solvers. And as Edward Glaeser wrote, they make cities humanity’s greatest invention. As long as we invest in the right infrastructure.