Quantifying sunk costs and learning effects in R&D persistence

This paper analyzes and quantifies the fundamental factors that are likely to cause persistence in performing R&D activities: the existence of sunk costs associated with R&D activities and the process of learning that characterizes this type of activity. We estimate our model with Spanish manufacturing firms for the period 1991-2014. By decomposing the effects of sunk costs and learning effects, we find that both are important determinants of R&D persistence, and that failing to allow for learning systematically overestimates sunk cost effects. Both large firms and SMEs benefit from direct and indirect (via productivity) effects of R&D experience, but in large firms this is more likely to be manifest through productivity improvements while in smaller firms the effect is more skewed towards a direct effect on R&D likelihood. Further, our results suggest that whereas the impact of sunk costs in R&D persistence is greater for large firms than for SMEs, the scope for direct learning from continuous R&D engagement is greater for SMEs than for larger firms.