International, intersectoral, or unobservable? Measuring R&D spillovers under weak and strong cross-sectional dependence

In the theoretical and empirical growth literature, private and social returns to R&D have been identified as the key drivers of productivity gains and economic development. However, recently the debate on the relative importance of private vis-à-vis social returns has been reinforced by contributions in two emerging fields of the applied econometric literature, namely spatial panel modeling and the common factor approach, which stress the role of cross-sectional dependence as a source for a potential estimation bias linked to the measurement of returns to R&D. In this paper, we account for these methodical advances when estimating sectoral knowledge production functions for OECD countries under weak and strong cross-sectional dependence. By doing so, we are able to uncover technology- and trade-related R&D spillover channels associated with social returns to R&D, while effectively controlling for other types of productivity spillovers and latent macroeconomic shocks. Our results highlight the role played by international-intersectoral R&D spillovers for the social rate of return to R&D, while we get limited evidence for private returns to R&D once cross-sectional dependence is properly accounted for.