The results achieved by the project are exposed in five articles, some already published or submitted to leading economic journals, and others still working papers.
The first significant result is the development of a new methodology which allows to analyze the optimal design of infrastructure networks in general equilibrium spatial models. The results have been published in “Optimal Transport Networks in Spatial Equilibrium” (Econometrica, 2020), joint with P. Fajgelbaum. The theory builds on a quantitative trade model which nests a wide range of international trade, economic geography models and urban models. Locations are arranged on a graph, and goods or people can only travel along the links of the network. The optimal infrastructure planning problem is analyzed through the lens of a planner who chooses how much infrastructure to build in each link, taking into account the private sector’s decisions (where to live/produce, how much/what to produce, optimal shipping/commuting routes and general equilibrium). This novel theory allows the authors to describe the optimal placement of infrastructure taking into account a) the distribution of productivity, resources and amenities, b) the particular geography of a location (availability of land, obstacles like rivers or mountains), c) the characteristics of the transport technology (degree of congestion, economies of scale in transportation). The theory is applied to Europe and evaluates the efficiency of its existing road network. The results broadly suggest a higher degree of infrastructure misallocation and underinvestment in Eastern European countries. An application to US states is also available on the PI’s website. A MATLAB toolkit is available on the PI’s website for the use of other researchers and policymakers.
In the article “Optimal Lockdown in a Commuting Network” (American Economic Review: Insights, 2021) joint with P. Fajgelbaum, A. Khandelwal, W. Kim and C. Mantovani, the project’s optimal network design methodology is applied to an urban environment with an application to COVID-19. The optimal network approach is used to study spatially targeted lockdowns at the level of a large metropolitan area. The theory builds on a standard urban framework in which workers choose to live in various districts of a city and must commute to work. The network is designed to control population flows in a commuting network and achieves a trade-off between saving lives and the economic costs implied by lockdown measures. The article applies its methodology to the metropolitan areas of New York and Seoul. The findings suggests great welfare and economic gains from spatially-targeted lockdowns as opposed to uniform lockdowns and recommends generally stricter lockdowns in those locations compared to reality.
A third paper analyzes how political institutions affect the design of real-world infrastructure project. “Political Preferences and Transport Infrastructure: Evidence from California’s High-Speed Rail” joint with P. Fajgelbaum, C. Gaubert, N. Gorton and E. Morales, uses the particular example of the California High-Speed Rail project to shed light on the role of political economy distortions. Using a quantitative spatial urban model with multiple modes of transport, the PI and his coauthors leverage highly-disaggregated voting data to analyze the voting behavior of Californian citizens and estimate the potential biases of the infrastructure planning authorities. The results suggest an important political bias and a significant overinvestment of resources towards pivotal counties in the election. In a fourth new paper, currently in progress, the PI and his coauthors P. Fajgelbaum and A. Khandelwal, apply the optimal network methodology to analyze China’s Belt and Road Initiative and identifies the potential inefficiencies and political biases present in its design.
A fifth working paper, “Zipf’s Law and Fractal City Networks” develops a new theory to explain the pervasive observation that the city-size distribution in many countries follows a Power law distribution with a coefficient close to -1. The theory first establishes that fractal city networks satisfy a Zipf’s law with a coefficient equal to minus the fractal dimension of the network. The theory goes on to establish conditions under which fractal networks are optimal and feature a fractal dimension close to 1. An empirical study is currently underway to test the theory’s predictions.
