The Internet and Access to Distant Ideas

Local Knowledge Spillovers in the Internet Age

Remote collaboration in innovation is on the rise. The average distance between all the inventors listed on a patent has tripled between 1975 and 2015.

Of course, the number of inventors on a patent has also risen. But we see the same thing if we restrict attention to patents with just two inventors.

What’s driving this rise? A big part of it is an increase in the share of patents co-authored by distant authors.

(Note though, the number of 100km-500km collaborations dropped after 2008, and the growth of 500km+ collaborations slowed - I don’t know why)

(All this data is pulled from patentsview’s inventor location data, with dates by patent application year)

What’s going on? Probably the internet.

Forman and Zeebroeck (2012) and Forman and Zeebroeck (2019) both look at how internet access changes the collaboration patterns of firms with geographically dispersed establishments. To measure the impact of internet access, the papers have to reach back a long way, to the 1992-1998 era, when internet access was first beginning to roll across America. FZ12 show that after two establishments are connected to the internet, inventors in the connected establishments are more likely to be jointly listed on patents. In contrast, getting internet access doesn’t seem to have any impact on the number of solo-inventor or geographically clustered inventor team patents, which suggests the internet’s main advantage was in facilitating collaboration, not merely in increasing access to knowledge.

But it does that too. Studying the same era, FZ19 show that when two establishments are connected to the internet, patents by inventors from one establishment are more likely to cite patents by inventors from the other. 

What’s the upshot? Forman, Goldfarb, and Greenstein (2014) show the internet helps reduce the trend towards geographic concentration of innovative activity. They compare the growth of regional patenting over 2000-2005 to patent levels over 1990-1995. Overall, they see a significant increase in the concentration of inventive activity: the counties with the most patents over 1990-1995 also had the fastest growth of patents over 2000-2005. What is novel, however, is that they show this effect is reduced by greater internet access. 

For example, consider two counties with low internet connectivity in the 1990s. Specifically, among all US counties, suppose they are at the 25th percentile for internet adoption. If one of these counties was also in the 90th percentile of 1990s patents, it experienced annual patent growth in the 2000s that was 5.4% faster than a comparable county in the 25th percentile for 1990s patents.

Now consider two counties with high internet connectivity in the 1990s (the 90th percentile for internet adoption). If one of these counties was in the 90th percentile of 1990s patents, it experienced annual patent growth in the 2000s that was just 0.4% faster than a county in the 25th percentile. The innovation gap between counties grew much more quickly for counties without much internet access.

What’s more, the difference between high-internet adopting and low-internet adopting counties is largest when we restrict attention to patents featuring distant collaboration among inventors. Now; we know from elsewhere that collaboration is increasingly important for innovation. One interpretation of these results therefore is that people living in innovative counties in the 1990s didn’t really need the internet to find potential collaborators, so it’s presence or absence didn’t matter that much. But people living in less innovative regions benefited a lot from internet access, because it allowed them to find good collaborators and participate in the innovation economy.

Taken together, another reason we’re seeing a decline in the strength of local knowledge spillovers is probably the internet.