The empirical results reported in this paper suggest that only about 20% of the aggregate effects of public investment in highways in the US are captured by the direct effects on each state output of public investment in the state itself. The remaining 80% correspond to the spillover effects from public investment in highways in other states. This result may provide an answer to the paradox in the literature that the findings of large effects at the aggregate level have not been matched at the regional level. This is because regional analysis has typically ignored the possible existence of spillover effects.

The impact of public capital on private sector performance has been a matter of great interest in academic and policy circles in the past decade