« Back to Results

Developments in Infrastructure and Transportation

Paper Session

Friday, Jan. 3, 2025 10:15 AM - 12:15 PM (PST)

Hilton San Francisco Union Square, Union Square 6
Hosted By: American Economic Association
  • Chair: Gabriel Kreindler, Harvard University

Ports, Disruptions, and Trade

Giulia Brancaccio
,
New York University
Myrto Kalouptsidi
,
Harvard University
Theodore Papageorgiou
,
Boston College

Abstract

Transportation infrastructure is vital for the smooth functioning of international trade. Ports are a crucial gateway to this system: with more than 80% of trade carried by ships, they shape trade costs, and it is critical that they operate efficiently. Yet ports are susceptible to disruptions, causing costly delays. With enormous budgets spent on infrastructure to alleviate these costs, a key policy question emerges: in a world with high volatility, what are the returns to investing in infrastructure? What are the best policies in battling disruptions (e.g. infrastructure investment, vs. congestion pricing)? And how do port disruptions affect trade, prices and inflation? We develop simple tools to address these questions.

A Method to Characterize Optimal Policies in High-Dimensional Settings

Gabriel Kreindler
,
Harvard University
Arya Gaduh
,
University of Arkansas
Tilman Graff
,
Harvard University
Rema Hanna
,
Harvard University
Ben Olken
,
Massachusetts Institute of Technology

Abstract

Spatial models often involve detailed geography, economic inefficiencies, complex substitution and complementarity forces, and high-dimensional policies, making social optimum analysis challenging. We introduce a general framework to characterize the properties of optimal policies. Our starting point is an estimated model that predicts welfare at counterfactual policies. We assume social welfare depends on model welfare and additional idiosyncratic factors that capture uncertainty about the model. This leads to a probability distribution of optimal policies over the space of policies. The expected welfare loss under this distribution relative to the global optimum is bounded by a simple expression. Markov Chain Monte Carlo algorithms can be used to sample from such distributions in an asymptotic sense. Repeated iid sampling from this distribution enables estimating the properties of optimal policies. At little additional computational cost, it is possible to compute local comparative statics with respect to model parameters, incorporate parameter uncertainty, and assess robustness to the size and correlation structure of idiosyncratic shocks. We illustrate the method with an optimal road transportation infrastructure investment problem based on Allen and Arkolakis (2022).

Sea Level Rise and Urban Infrastructure

Allan Hsiao
,
Stanford University

Abstract

Sea level rise threatens Jakarta with significant increases in coastal flooding by 2050. I quantify the exposure of Jakarta's current urban infrastructure to future coastal flooding, focusing on roads, schools, and hospitals, and I assess the scope for new protective infrastructure, focusing on sea walls and pumping stations. I discuss the trade-off between investing inland and adapting in place.

Optimizing Electric Vehicle Infrastructure

Costas Arkolakis
,
Yale University
Kenneth Gillingham
,
Yale University

Abstract

What are the welfare gains from upgrading electric vehicle infrastructure? We estimate a spatial model of electric vehicle charging using rich data on electric vehicle registrations, cell phone tracks, and electric vehicle charging locations and characteristics. Our estimates allow us to explore counterfactuals that adjust the charging station network, including increasing charging capacity at any given location and upgrading all level 2 chargers to level 3 chargers. We quantify the welfare benefits from upgrading charging infrastructure incorporating trip choice and allocation across links, providing important insights for policymakers.

Spatial Externalities, Inefficiency, and Sufficient Statistics

Gabriel Kreindler
,
Harvard University
Kartik Patekar
,
Poverty Action Lab

Abstract

How much economic inefficiency is generated by spatial externalities such as agglomeration and congestion? We show how to express deadweight loss in a spatial model, building on a formula from Harberger (1964). Our expression highlights the importance of two empirical objects, an externality matrix and an equilibrium elasticity matrix, and clarifies how specific model assumptions may constrain the magnitude of deadweight loss. Similar arguments apply to marginal effects from shocks and policies such as transportation improvements. We illustrate with several examples at different spatial scales.
JEL Classifications
  • L9 - Industry Studies: Transportation and Utilities
  • R4 - Transportation Economics