Cities and Growth with Gilles Duranton and Diego Puga

Episode Summary

Executive Summary: Gilles Duranton and Diego Puga discuss their theory-and-quantitative paper on urban growth, arguing that city productivity gains must be weighed against congestion, housing costs, and political barriers to expansion. Their model explains city size distributions, incumbents’ resistance to growth, and why relaxing land-use restrictions could raise aggregate output even if some current residents lose.

Main Topics: Origins of the paper and research program (Priority: 5/5): The authors explain that the paper grew out of a survey on city growth, where they saw the need to synthesize scattered urban economics findings into one transparent model disciplined by data. Agglomeration benefits versus urban costs (Priority: 5/5): They emphasize the core urban tradeoff: larger cities raise productivity through spillovers, but also increase housing costs, commuting time, and congestion. Political economy of land-use regulation (Priority: 5/5): A central innovation is modeling city size as a conflict between incumbents and newcomers, capturing why existing residents often block housing growth and new entry. City size distribution and churn (Priority: 4/5): The model is designed to match empirical regularities such as Zipf-like city size distributions and changing rankings of major cities over time. Commuting, congestion, and estimation strategy (Priority: 4/5): They describe how they quantified commuting costs using multiple data sources and multiple model layers, finding consistent estimates across methods. Counterfactuals and welfare effects of deregulation (Priority: 5/5): The authors discuss model simulations showing that lifting planning restrictions can substantially increase national income, though incumbents in big cities may lose.

Key Arguments: Cities matter for growth not just because they are productive places, but because their size and composition affect aggregate income and welfare. Purely empirical cross-country work is too weak to identify the macro effects of cities, so theory must be combined with disciplined calibration and estimation. Urban growth is constrained by a political tension: current residents internalize congestion and housing costs more than the broader national gains from allowing newcomers. City size is not purely efficient; productive cities can be held below socially optimal scale by land-use restrictions and local politics. Housing and commuting costs are essential to explain why all people do not end up in one megacity despite strong agglomeration economies. The model’s consistency across within-city and across-city estimates gives confidence that it captures real urban cost structures. Relaxing development restrictions would shift population toward more productive cities, raising aggregate output even if incumbent homeowners in those cities oppose change.

Data Points: Collaboration duration: about 25 years - Duranton says he and Puga started collaborating in grad school and have worked together for roughly 25 years. Elastecity of driving distance vs. distance from city center: 0.08 - They estimate that doubling distance from the city center increases driving distance by about 8%. U.S. metro population growth claim: 1980 to 2010: no metropolitan area shrank - Puga notes that all U.S. metropolitan areas grew over this period, even if some grew slowly. San Francisco growth under current constraints: less than 1% per year - Used to illustrate how restrictive land-use regulation limits growth in highly productive places. Potential growth absent barriers: 5% to 6% per year - They argue prosperous U.S. cities could grow much faster without restrictions. Counterfactual welfare gain from less restrictive city growth: 25.7% - In the discussed simulation, allowing New York to double toward 40 million produces a large average income gain. Incumbent New Yorker welfare effect: about 13% worse off - Current residents of the largest city lose under the counterfactual even as the nation gains overall. Boston incumbent welfare effect: about 1% loss - Smaller losses appear in less extreme large cities, illustrating that incumbents are often near their preferred city size. City size bound in counterfactuals: 30–40 million - They cap very large cities in simulations to avoid unrealistic extrapolation to megacities far beyond observed data. Observed city size range used for parameter estimation: 100,000 to 20 million - Their urban cost parameters are estimated from American metropolitan areas within this population range.

Pivotal Quotes: "the sum is worth more, the whole is worth more than the sum of the parts" — Gilles Duranton: Explaining agglomeration economies and why cities can raise productivity. "we cannot micromodel absolutely everything" — Gilles Duranton: Describing the need to simplify political and housing-market details while preserving the core mechanism of incumbents versus newcomers. "understanding this process and understanding what's taking us there and how we can move on" — Gilles Duranton: Summarizing the paper’s broader goal beyond any single numerical counterfactual result.

Implications: The conversation suggests that urban policy can materially affect national prosperity. Restrictive zoning may protect incumbents but reduce aggregate welfare by preventing workers from moving to more productive cities; the long-run gains from reform may be large.