MACRO: On (Un)Congested Roads: A Quantitative Analysis of Infrastructure Investment Efficiency using Truck GPS Data; Professor Michael SONG (Chinese University of Hong Kong (CUHK))

Abstract

This study aims to quantify the gains from investments in a transportation network, where the elasticity of driving time to traffic (``congestion elasticity'') may differ across roads.  We first use high-frequency GPS data from half a million Chinese trucks to uncover the congestion elasticity heterogeneity in China's city-to-city road links.  We find that one-third of the links are uncongested and that no more than 40% are associated with a large congestion elasticity comparable to the recent estimates for developed economies.  In contrast, using similar real-time traffic data for interregional highways in England, we find that almost all roads are associated with a large congestion elasticity.  We next incorporate congestion elasticity heterogeneity into a general equilibrium trade model with optimal route choices developed by Allen and Arkolakis (2019) and structurally estimate the model. To calculate the returns on investment in each link, we infer the benefit from the estimated model and calculate the construction cost and the opportunity cost of land directly from the data.  We find the returns to be highly unequal in China, and the heterogeneity in congestion elasticity can account for more than half of the dispersion. Numerical simulations show that dispersion is a robust indicator of misallocation and that optimized investments with a reasonable budget generate sizable welfare gains. Moreover, the optimal investment allocation turns out to be orthogonal to the actual allocation in the most developed provinces.  Our findings suggest a severe misallocation of road investments in China.