Can America Reduce Highway Construction Costs? Evidence from the States
This chapter is a preliminary draft unless otherwise noted. It may not have been subjected to the formal review process of the NBER. This page will be updated as the chapter is revised.
Chapter in forthcoming NBER book Economics of Infrastructure Investment, Edward L. Glaeser and James M. Poterba, editors
Although infrastructure is a key input into economic growth, systematic evidence on its cost across time or place is very limited. In this paper, motivated in part by the difficulties in international comparisons, we focus on infrastructure for which we can consistently measure cost over time and space: the US Interstate system. Looking over the period of the build-out of the system from 1956 to 1993, we find that the 75th percentile state spent $8.8 million more per mile than the 25th percentile state, relative to mean spending per mile of $11.5 million (all dollar figures are 2016 dollars). If states spending over the median had limited their expenditure per mile to that of the median state, the Interstate system would have cost about 40 percent less to build. Even when we limit to costs within policymaker discretion, netting out pre-determined characteristics such as the slope of the terrain, a $3.3 million per mile interquartile range persists. We then show that this cross-state variation exceeds that in other related public and private spending, and examine patterns of correlation that provide evidence of common cost drivers. We review the evidence that does exist on the root causes of infrastructure costs and test some of these hypotheses in our cross-state setting. Our empirical tests find limited evidence for any single driver of cross-sectional variation.
Comment, Clifford Winston