Spillovers, Innovation Difficulty, and the Dynamics of Productivity

Is the aggregate productivity slowdown in the U.S. driven by a decline in successful innovation? This paper addresses this question using a medium-scale DSGE model with endogenous technology growth. The model distinguishes between two innovation channels: a spillover channel, which governs the efficiency with which aggregate R&D advances the technological frontier, and a difficulty channel, which governs the probability that sectoral R&D efforts successfully generate innovation. We estimate the model on U.S. macroeconomic and R&D data over the period 1984-2019, using macroeconomic observables and incorporating a patent-text-based measure of technological creativity that is informative about innovation probability. The results show that spillover shocks are the main drivers of short and medium run fluctuations in TFP growth, while R&D difficulty shocks mainly explain the probability of successful innovation. Once creativity data are included, the estimated difficulty shock becomes less volatile and more persistent, suggesting that innovation difficulty is a slow moving force shaping successful innovation. However, its quantitative contribution to TFP fluctuations remains substantially smaller than that of spillover shocks, although it matters in specific episodes.