Revisiting the Productivity Effects of Public Capital

We explore various production functions and estimators to examine the productivity effects of public capital using US state-level data from 1986 to 2005. We show how the estimated effects vary with several factors, including the use of flexible nonparametric and semiparametric production functions, the decomposition of the total effect into Hicks-neutral and non-neutral effects, the control of cross-sectional dependence using factor models, and the disaggregation of public capital into its components. In general, we find a positive overall average effect of public capital. However, considerable heterogeneity exists, and incorporating factor models into the production function leads to insignificant public capital effects. We complement our investigation by conducting additional robustness checks on the results.