The Distribution, Diffusion and Persistence of Technologies



Complexity Economics has intrigued me for long, and I am an avid reader of its creator, Brian Arthur (an economist credited with developing the modern approach to increasing returns). To summarize, complexity economics is the “radical” notion that the economy is not in equilibrium as conventional models dictate. Instead, this model views the economy as a dynamic and perpetually evolving system. As part of Brian Arthur’s book called Complexity and the Economy, he has a chapter entitled “Competing Technologies, Increasing Returns and Lock-in Historical Events''. In it, he reveals the mechanics behind why and how certain technologies get adopted and ultimately persist. The knowledge that he exposes can further be applied to understanding the distribution of technologies in our economy, which I will demonstrate.



Arthur’s paper on increasing returns to technology dates back to 1983, but it is now more relevant than ever before with the incredible pace of innovation in today’s digital age. Here is an excerpt from the chapter mentioned above that will help to explain things. It is especially wonderful that he uses simple language to allow anyone to understand the concept:

Modern, complex technologies often display increasing returns to adoption in that the more they are adopted, the more experience is gained with them, and the more they are improved. When two or more increasing-return technologies “compete” then, for a “market” of potential adopters, insignificant events may by chance give one of them an initial advantage in adoptions. This technology may then improve more than the others, so it may appeal to a wider proportion of potential adopters. It may therefore become further adopted and further improved. Thus a technology that by chance gains an early lead in adoption may eventually “corner the market” of potential adopters, with the other technologies becoming locked out. Of course, under different “insignificant events”—unexpected successes in the performance of prototypes, whims of early developers, political circumstances—a different technology might achieve sufficient adoption and improvement to come to dominate. Competitions between technologies may have multiple potential outcomes.”


This insight has been quantitatively proven in Arthur’s work, and I encourage you to learn more about it in his book, Complexity and the Economy. Now, however, I want to actually apply this knowledge (like a true economist). There is a fantastic paper via the Institute for New Economic Thinking called “The Geography of New Technologies,” which at times implicitly calls upon Arthur’s framework to discuss the distribution of emerging technologies. The authors use textual analysis of earnings conference calls, newspapers, announcements, and patents to examine the adoption of 20 technologies across organizations and labor markets in the United States.


The key results of the paper’s analysis are shown below:

  • Earnings call mentions and hiring announcements linked to the new technologies rise in parallel over time

  • While initial hiring is focused on high-skilled jobs, over time the mean skill level in new positions associated with the technologies declines sharply, which we term a “skill- broadening” effect;

  • New hiring in new technologies increases its geographic footprint over time, becoming less concentrated, which we dub “region broadening”;

  • The initial geographic hub retains an important advantage that persists over time. This pattern is particularly pronounced among high skill jobs; and

  • Hubs are most likely to arise around universities and areas with more educated populations.

Now, I will apply Arthur’s Framework to some of these findings as well as my own economic intuition.


Earnings call mentions and hiring announcements linked to the new technologies rise in parallel over time


This finding makes complete sense. With the widespread adoption of technology by firms, the demand for skilled workers in these technologies increases. This means that hiring announcements increase, leading to higher levels of employment. More workers translates to larger adoption of the technology, and the more the tech gets integrated, the more workers it calls for. This is evidently a self-reinforcing pattern (which Arthur talks about). Eventually, firms reach a status with the technology that enables them to go public, therefore increasing the amount of earnings calls.


While initial hiring is focused on high-skilled jobs, over time the mean skill level in new positions associated with the technologies declines sharply, which we term a “skill- broadening” effect;


When a new technology is adopted, the skills required for its implementation are high. This means that the labor pool reflects a strong bias towards a specific subset of highly-skilled workers. The initial adoption of the new technology is a high-skill, labor intensive process. However, as it becomes “less novel” over time (as it matures in adoption), new roles are created in association with the technology for purposes of maintenance and related domains. This is also an “Arthurian” self-reinforcing pattern. Because these associate positions require less skills, hiring for low-skilled workers soars. This occurs when the maturity of the technology exceeds the level of skill needed for initial adoption.


New hiring in new technologies increases its geographic footprint over time, becoming less concentrated, which we dub “region broadening”;


When a new technology enters the mainstream, it usually floods the market. For example, even though many technologies come from Silicon Valley, they eventually find their way across the U.S. As a result, the demand for employees who can work with these technologies increases. This is a rather straightforward point.


The initial geographic hub retains an important advantage that persists over time. This pattern is particularly pronounced among high skill jobs; and


We can think about this finding as a “Lock-in phenomenon” whereby a hub that happens to be the first to innovate a technology will find itself exponentially building upon it in a cyclical pattern. Because of how rapid the uptake of the technology occurs, it gains an absolute advantage over other hubs that persists over time. Similarly, if we have State A and State B, both of which are simultaneously innovating the same technology, due to chance gains or insignificant events (as Arthur alludes to), State A may outpace State B or vice versa in adoption.


Hubs are most likely to arise around universities and areas with more educated populations.


This is also a rather straightforward point. Populations that are better educated are a result of higher government expenditures towards R&D for research institutions or educational programs, as well as historical trends (some areas may have historically evolved into centers of knowledge/academic excellence). A more educated population is likely to have more skilled people capable of accelerating adoption of new technologies faster and better. Therefore, it is natural that “advantageous hubs” may arise around universities and more education populations.


To Conclude…

The distribution, diffusion and persistence of technologies is an interesting mathematical, economical and scientific discussion that is becoming ever more relevant in the age of AI, and is worth exploring in greater depth.



24 views0 comments