From the literature review that was presented in chapter 2.2 it is apparent that no other study has expressively concentrated on incremental invention and innovation before this one. This might be because of a lack of theoretical understanding of what incremental innovation actually is and how it could be recognized. This study has tried to resolve both of the aforementioned conundrums: To establish a coherent theory around innovation typology, and to measure the industrial salience and the technological content of these kinds of inventions.
The theory behind incremental invention rests on three concepts: base principle (Arthur, 2007), system hierarchy (Arthur, 2007; Murmann & Frenken, 2006), and the dominant design (Murmann & Frenken, 2006). Incremental invention seems to be one that does not introduce a new base principle, i.e. it is not recombination of ideas (Weitzman, 1998) but a development of an existing idea. Further, incremental invention requires some kind of structure or unit of measure within a particular technology. Such a structure can be found in the interrelations or linkages of individual components and subsystems (Murmann
& Frenken, 2006). In this view, incremental inventions apply to peripheral components in terms of new knowledge that they bring to bear (Murmann &
Frenken, 2006). Relatedly, as incremental innovations take place in peripheral components, the number of these components destines the possibilities for improvement: The more complex a product is, the larger the possibility to
improve it incrementally (Murmann & Frenken, 2006). Lastly, the emergence of a dominant design within a product category ensures that the aforementioned requirements are filled and further developments will take place in peripheral components.
Overall, incremental innovations seem to have a special place in the gestalt of a generic technology's life cycle where the relative salience of incremental innovations tells that a particular technology has reached a certain kind of maturity. Convenient solutions have been found to design problems higher up the system hierarchy, and thus, future problems will be found and solved in peripheral parts of the system. In practice, this often means the emergence of a dominant design, and it is this concept that allows us to tie the discussion around incremental innovation to larger context.
The emergence of a dominant design can be seen as a watershed moment in an industry's evolution. According to the theoretical findings of this study, what seems to be happening is that the R&D procedure itself changes its form from stochastic into deterministic one with the emergence of a dominant design. The logic behind of this argument should be evident from the previous discussion: A dominant design essentially means the solidification of core components, and hence, further developments can be said to be the refinement of an existing design. What is important here is that the design already works.
The same can not be said about design trials that take place before this watershed moment.
This changes the relative incentives to partake a patent race, i.e. the game that is played changes from Gilbert and Newbery's (1982) one to Reinganum's (1983) one. Essentially, this means that incremental innovations can be associated with established firms and mature industries. Established firms have an incentive and the capability (Henderson, 1993) to protect their position in the market with improvements to their products, and these improvements are likely to be small (Reinganum, 1983), and this behavior is largely incentivized by the existence of patents (Gilbert & Newbery, 1982). If, then, incremental innovations follow the emergence of a dominant design, they are likely to be process innovations (Klepper, 1997). Essentially, this means a shift from product characteristics (quality) to production costs (productivity).
All this involves a remarkable similarity with the observed “inverted-U”
pattern of entry and exist in some industries (Murmann & Frenken, 2006). In the era of ferment (prior to the emergence of a dominant design) market leadership changes hands frequently and competition is fierce – more and more
firms enter the marketplace. Then, some designs prove better than others or some designs are selected because of network effects to be better than others.
Firms that have invented these designs see their market shares rise and the market becomes more oligopolistic. If imitation is possible and one design rules them all, a dominant design is likely to emerge. After this threshold, innovations are likely to be incremental. Therefore the following statement:
Technological change and market structure seems to be fundamentally and causally connected.
Theoretical findings of this study can also be thought of as a form of aggregated learning where a given technology's – base principle's – full potential is achieved, i.e. as the learning curve. Invention, on the other hand, can be though of as birth of a new technology, i.e. radical invention. As it so happens, both of these phenomena have been modeled in growth literature. Young (1993) and Stein (1997) present combinations of the aforementioned models. Young's (1993) model can be used to model the development of whole economies and Stein's (1997) model is more suitable for the examination of industry evolution.
More on these models in appendix B.
Yet, how can these developments in the product or the underlying technology be measured? By reviewing the prior empirical literature it was discovered that perhaps the simplest solution to this conundrum is the use of patent statistics. Utility models, especially, offer a novel method for the study of technological change. A growing evidence points to the fact that a growing number of patent applications might embody small or even trivial inventions (Guellec & van Pottelsberghe de la Potterie, 2000), while only a small percentage of patents prove to be valuable (Scherer & Harhoff 2000; Silverberg
& Verspagen, 2007). Utility models, on the other hand, are always likely to be small and incremental inventions.
Using this method it was found that technologies and industries that could be said to be in the era of incremental change, measured by the relative prevalence of utility models, were mostly supplier-dominated industries (Pavitt, 1984). However, it is also apparent that those industries that patent also use utility models – to a varying degree. Furthermore, it is clear that these results are at least to some extent driven by selection. A large firm that innovates incrementally always has an incentive to protect its invention as broadly as possible as it can spread the cost of doing so over many products. A small firm does not have this option, and hence, such a firm should be more careful in choosing the form of legal protection.