Learning and Persistence of Leadership in Product Innovation: Theory and Some Facts from the Semiconductor Industry

Introduction and summary

This paper shows how diffusion of quality innovation is an equilibrium outcome of firms competing a’ 1a Cournot and what are the conditions for having always the same firm to innovate first, i.e what generates persistence of leadership. The analysis is carried out within a (pure) horizontal product differentiation model in an infinitely repeated game form. Firms are identical and have perfect information. Although I will deal only with the three firms case, diffusion should be an equilibrium outcome for any market structure which is not monopoly. Furthermore, first mover advantages (e.g. due to learning) are crucial in avoiding leapfrogging, and innovation is always carried out first by the same firm and the other firms follow in the same sequence. The model implies clearly defined patterns of market shares which allow to identify different strategies for firms. These are contrasted with empirical evidence from the semiconductor industry. This industry has been chosen because the typical product, the memory chip, satisfies closely the basic assumption of the model.

The paper is organized as follows. The section 2 lays out the vertical product differentiation model in its static form. In the section 3 a three firms framework is extended to a dynamic game in order to analyse the innovation process. The equilibrium outcome implies different strategies for otherwise identical firms whereby three different strategies can be identified. Firms adopt new qualities sequentially whereby the leader remains leader in innovating and followers remain followers in the same order. This model has clear implications for time profiles of market shares of different firms, evolution of prices, sales and concentration indices. In the section 4 this model is contrasted with empirical evidence from the semiconductor industry. Memory chips, as EPROMs, have the ideal properties of vertically differentiated products. It is shown that the world market shares of the US firm Intel, the inventor of memory chips, have a pattern which is strikingly similar to that implied by the leader in the model. Similarly, the world market shares of the US firm AMD, known as “second source” producer, follow closely the patterns implied by the imitator in the model. The market shares of the bulk of the firms show a pattern which lies between these polar cases.