On a macro level, knowledge creation relates to innovation and how it influences economic growth. Knowledge creation, however, originates at the micro level. Microeconomic evidence for knowledge creation and innovation at the level of individual inventors is however lacking. One interesting question is whether, and how, collaborations between individual scientists benefit knowledge creation. Such studies are important as they can inform policies. In context of innovation and economic growth, fostering collaborations and promoting collaborative spirit should trump scientific research isolationism.
For the collaboration to be of mutual benefit to the collaborators, they need first to bring something to the table. How much of this “differentiated knowledge” there is in a collaboration seems to influence the average collaborative output. Researchers more actively looking for collaborators that would maximize their differentiated knowledge reap a larger reward in the form of substantially increased output. Experience is a factor here. More experienced researchers are in general able to attract more collaborators because their own knowledge stock is larger and therefore more suitable to a broader set of potential collaborators.
However, one may also argue that this would lead to the emergence of “Jack of all trades” researchers. In principle, one would expect that the knowledge stock of researchers doing more in-depth studies is much more solid, although it is hard to differentiate this based on patents alone.
Mori and Sakaguchi  from the Institute of Economic Research in Kyoto and the Research Institute of Economy in Tokyo under the Japan’s Ministry of Economy, respectively, present a detailed study on the effect of collaboration on knowledge creation. They investigate an impressive set of 107,724 inventors and their research output in the form of patents over a period of 14 years. Although using published patent applications reflects innovation at the time of application, if the patent is later rejected it somewhat complicates the whole story because it may have been deemed not innovative, paradoxically even if it has received citations.
Their data shows how, due to the constant innovation forces, inventors constantly need to come up with new ideas as their old achievements become obsolete. Collaborations can avoid the trap of becoming irrelevant by seeking collaborators who offer substantially different knowledge.
(…), the data indicate that the set of collaborators evolves for each agent over time. To maintain their productivity, inventors appear to keep shifting their technological expertise to unexplored niches by meeting new agents with different backgrounds from theirs.
It is then not surprising that the more collaborations a researcher tries, the more productive and novel the output becomes. At the same time, collaborators with a small knowledge stock tend to seek more collaborations while collaborators with high knowledge stock prefer to have less collaborations.
(…) firms, cities, regions, and countries that promote encounters and collaboration among individual inventors across organizations and institutions, despite the possibility of imitations and undesired diffusions, may have better chances to foster innovation there. While lower organizational and institutional barriers for research collaboration are not incompatible with the protection of intellectual property by patents, our finding supports more active coordination than divisions among researchers to encourage innovation.
 Collaborative knowledge creation: evidence from Japanese patent data, Tomoya Mori, Shosei Sakaguchi, Aug 6 2019, arXiv:1908.01256v1
This paper presents micro-econometric evidence for collaborative knowledge creation at the level of individual researchers. The key determinant for developing new ideas is the exchange of differentiated knowledge among collaborators. To stay creative, inventors seek opportunities to shift their technological expertise to unexplored niches by utilizing the differentiated knowledge of new collaborators. Furthermore, a more active recombination of collaborators by an inventor facilitates the selection of collaborators to raise the amount of differentiated knowledge from their collaborators.