To What Extent Does Policy Uncertainty Affect Innovation? An Empirical Study of the Effect of Policy Uncertainty on Innovation

Abstract

Measuring innovation through the log of scaled patents, we find a significant negative relationship between policy uncertainty and innovation. This is robust to the use of raw patent counts. We use alternative measures and instrumental variables for the independent variable to address endogeneity and find the estimated relationship between innovation and policy uncertainty to be consistently negative for three of four measurements: the opposite relationship is reported when election data is used as an alternative measure. We decompose policy uncertainty into fiscal and monetary policy, and find that these have opposing effects, where fiscal policy uncertainty is associated with a decrease in innovation, monetary policy uncertainty has the opposite relationship. The impact of policy uncertainty is greater for innovation with more impactful patents and smaller for patents with a higher degree of novelty. Consistent with the precautionary motive to hold cash, we find that the innovation of firms with a lower level of cash holdings have a more adverse reaction to increased policy uncertainty. Moreover, we find the degree of financial constraint to be inconclusive, as it depends on the measure of financial constraint. For instance, we find that the debt to asset ratio indicates that innovation of financially constrained firms is impacted more by increased policy uncertainty compared to that of financially unconstrained firms, however, the interest coverage ratio suggests the opposite relationship. Finally, we find that the sensitivity of innovation to changes in policy uncertainty changes depending on the initial level of policy uncertainty. The sensitivity is bigger for large firms in both high and low policy uncertainty environments, consistent with previous findings that small firms generally have a more stable level of innovation and thrive under uncertainty relative to large firms.