Knowledge spillovers and their impact on innovation success: A new approach using patent backward citations

We propose a new patent-based measure of knowledge spillovers that calculates technological proximity based on firms that were identified via patent backward citations links. We argue that this measure has a couple of advantages as compared to the 'standard' measure proposed by Ja e: First, it reflects spillovers from both domestic and foreign technologically 'relevant' firms, second, it is more precise because it only takes into account knowledge relations with technologically 'relevant' firms. Our empirical results indeed show that the measure performs better than the standard measure in an innovation model. We find - for a representative sample of Swiss firms - that knowledge spillovers measured in this way have a positive and significant impact on innovation success. However, the knowledge spillovers appear to be localized: Spillovers from geographically distant areas such as the USA and Japan matter less than spillovers from near destinations such as Europe and particularly Switzerland itself. Moreover, the spillover effect on innovation performance decreases with increasing number of competitors on the main product market so that this effect would appear only in niche markets or oligopolistic market structures. However, an additional effect of competition can only be detected for more radical innovation success.