Trade policy and learning by doing: the case of semiconductors

Abstract

The study examines the role of learning by doing for production of different types of semiconductors and how this affects the dynamics of market structure. For erasable programmable read only memories (EPROMs) the learning curve determines aggregate market dynamics, whereas for dynamic random access memories (DRAMs) economies of scale seem to be more important. These results are compared with some stylised facts. For EPROMs, persistence of leadership in product innovation could be observed, whereas for DRAMs, there was leap-frogging in product innovation. Moreover, scope for vertical product differentiation is more enhanced for EPROMs and price competition is less intense. These findings are then related to the problem of identifying the appropriate trade policies in oligopolistic industries. The analysis concerns the antidumping actions undertaken by the European Commission. It is shown how markets enjoying similar patterns of protection have a completely different evolution of market shares for European producers. The conclusion for strategic trade policy is that industry specific features matter crucially.