Entry into Innovation Areas

Abstract

This paper investigates the entry decisions of U.S. electronics firms across innovation areas. I apply topic modeling to over 630,000 patents to endogenously define innovation areas and merge those areas with market-implied patent valuations. Firms occupy several areas at once, and the number of firms differs widely across areas. An entry game, estimated by moment inequalities, finds that additional rivals reduce expected profit and that firms gain from occupying more areas at once. In the case study of semiconductor, a payment based on a quarter of predicted value shifts 0.1-0.5% into those areas and leads up to 0.8% of firms to add an extra area. Meanwhile, a 25% reduction in the occupancy cost shifts up to 7% into those areas and increases up to 10% of firms to add an extra area.