Knowledge Diffusion and Knowledge Transfer: Two Sides of the Medal

The new growth theory considers knowledge to be a decisive engine of economic growth. More precisely, knowledge is not used solely to the benefit of its originator, but generates positive side effects also for others, provided they have the capability to understand the transferred knowledge potential. Knowledge generation has also welfare implications for a country or a region. On a macroeconomic level, the implication of knowledge diffusion for growth seems straightforward, whereas on the microeconomic level the effect of knowledge diffusion seems more complex. Before dealing with the question of how knowledge diffusion can be adequately modeled in a concise microeconomic approach, two important aspects have to be distinguished from each other: knowledge diffusion and knowledge transfer. Although these terminologies are well established in the relevant literature, it is not or rather inaccurately acknowledged within the knowledge diffusion modeling context. Knowledge transfer is associated with the exchange of knowledge within networks, which consists of innovators and imitators of knowledge. On the contrary, knowledge diffusion describes the diffusion of knowledge within the group of innovators and imitators. Apparently, knowledge transfer can accelerate but is not a necessary condition for knowledge diffusion. From this point of view, welfare implications are mainly expected from knowledge diffusion, which can be indirectly enforced by knowledge transfer. Therefore, the intensity of knowledge networks should also affect the diffusion pattern of knowledge. This paper proposes a model that can explain endogenously the knowledge diffusion patterns induced by network effects. In this way, to the best of my knowledge this is the first attempt to discuss both, knowledge diffusion and knowledge transfer in a comprehensive framework. A key result shows that unimodal diffusion patterns are generated by strong network effects, whereas bimodal diffusion patterns occur due to weaker network effects. Thus, the stronger network effects, the faster is knowledge diffusion. Furthermore, this model assumes that the knowledge diffusion process is embedded in a stochastic environment. Particularly, at the beginning and in the middle the uncertainty of adopting new knowledge is larger than at the end of the diffusion process. From an econometric point of view, this can be modelled via heteroscedastic errors in the error term. A further pleasant feature of this model is that it can be directly estimated with a seemingly unrelated regression (SUR) approach.

Klarl, Torben (2009), Knowledge Diffusion and Knowledge Transfer: Two Sides of the Medal, ZEW Discussion Paper No. 09-080, Mannheim.