Esteban Rossi-Hansberg, University of Chicago
Join group to receive calendar invite: https://groups.google.com/a/modelingtalks.org/g/talks
Abstract:
We evaluate how anticipation and adaptation shape the aggregate and local costs of climate change. We develop a dynamic spatial model of the U.S. economy and its 3,143 counties that features costly forward-looking migration and capital investment decisions. Recent methodological advances that leverage the ‘Master Equation’ representation of the economy make the model tractable. We estimate the county-level impact of severe storms and heat waves over the 20th century on local income, population, and investment. The estimated impact of storms matches that of capital depreciation shocks in the model, while heat waves resemble combined amenity and productivity shocks. We then estimate migration and investment elasticities, as well as the structural damage functions, by matching these reduced-form results in our framework. Our findings show, first, that the impact of climate on capital depreciation magnifies the U.S. aggregate welfare costs of climate change twofold to nearly 5% in 2023 under a business-as-usual warming scenario. Second, anticipation of future climate damages amplifies climate-induced worker and investment mobility, as workers and capitalists foresee the slow build-up of climate change. Third, migration reduces substantially the spatial variance in the welfare impact of climate change. Although both anticipation and migration are important for local impacts, their effect on aggregate U.S. losses from climate change is small.
Bio:
Esteban Rossi-Hansberg is the Glen A. Lloyd Distinguished Service Professor in the Kenneth C. Griffin Department of Economics at the University of Chicago (since 2021). Previously, he was a Professor of Economics at Princeton University from 2005 to 2021 and at Stanford University from 2002 to 2005. He earned a Ph.D. from the University of Chicago in 2002. He is a research associate at the National Bureau of Economic Research and a research fellow at the Center for Economic Policy Research. He is the co-director of the International Economics and Economic Geography Initiative at the Becker Friedman Institute. In 2021 he became an Editor of the Journal of Political Economy.
Rossi-Hansberg’s research specializes in international trade, regional and urban economics, as well as growth and organizational economics. Among other topics, his work has studied the internal structure of cities; the implications of offshoring and changes in firm organization on economic outcomes; and the impact of spatial frictions and agglomeration and congestion forces on the gains from migration and the cost of climate change. He has published extensively in all major journals in economics.
In 2007 he received the prestigious Alfred Sloan Research Fellowship and in 2010 he received the August Lösch Prize and the Geoffrey Hewings Award. He is an elected fellow of the Econometric Society since 2017 and won the Robert E. Lucas Jr. Prize in 2019. In 2022 he was elected to the American Academy of Arts and Sciences.
Summary:
Context:
The planet is warming
Impact varies across locations
Current assessment frameworks do not account for
Location of extreme weather events
Anticipatory actions: investment and migration
Question: how do anticipation and adaptation affect costs of climate change-induced heat waves and storms?
Approach:
Dynamic Spatial General Equilibrium model of 3143 US counties
Local extreme events
Anticipation: investment, migration
Master Equation approach
Estimate damages from extreme events using 120 years of county-level weather data
Storm=17% capital depreciation, heat wave=5% productivity + 7% amenity shock
Social costs of climate change are much larger than previously thought
Framework:
Workers:
Live in some county but can move
Choose housing, consumption, county of residence
Utility = max (consumption + amenities) + (continuation value from aggregate changes) + (continuation value from migration)
Constraint: wage = (consumption cost) + (cost of housing in each location)
Capitalists
Located in some country, immobile, risk-neutral
Decide whether to invest in capital in location
Utility = max (flow utility) + (continuation value from net investment) + (continuation value from aggregate changes) + (continuation value from net savings)
Stom affect investment value
State-dependent depreciation rate
Access to capital not explicitly modeled: assume access to national bond market to fund local investment
Production (in each county):
Capital stock: housing, commercial structures
Labor: production labor, building construction
Buildings
Final good
Climate damages:
Global mean temperature: baseline + change
Change in temperature linearly affects amenities, productivity and depreciation rate
(could also down-scale global models to local temperature changes and use those)
Solution method:
General Equilibrium environment
Aggregate shocks
Distribution of wages, rental rates at state level (6284 values)
Solve via Master Equation approach
State-space analytic perturbation around steady state
Based on mean-field game literature
1st order solver (2nd order solver coming in future)
Data
Economic: 1960-2019
Investment - 5 year census of manufacturers
Wage and population from census and BEA
Historical climate data: ISIMIP (1900-2019)
Distributed lag specification of storms
Log wage, population, investment = lagged past values of variable + lagged past values of storm indicators
Estimating impact of event h time periods ago on today’s state
Effect of storms on: (coastal counties only)
Wages: decline by 2% within 2 years
Population: decline by 4%
Investment: increases by 15% over the following few years, followed by a small drop
Thus, storms are depreciation shocks in coastal counties
Heat waves (warm counties only):
Wages: no effect
Population: 2% decline
Investment: no effect
Thus, heat waves are productivity + amenity shocks
Cold waves have little effect
Estimation strategy:
Inversion of steady-state fundamentals
Conditional on: investment data, population, matrix of inter-county mobilities, etc.
Invert the model
Solve for the characteristics of the individual counties: amenities, productivities, available land, cost of investment, inter-county migration cost
Infer key elasticities:
Sample the model
Find model elasticity parameters that match the data
Warming makes extreme events more frequent
Used inferred elasticities (damage impacts) to infer damage from these increases
2%-8% increase in storms
5%-20% increase in heatwaves
Inferred 4% increase in capital depreciation due to storms on coastal Atlantic counties
Significant heatwave impacts on Southeast US
Prediction: Climate damages are twice as large as previously thought
Impacts are applied to the simulated economy and propagated through time
Southeast US strongly affected due to heat, Atlantic storms
Model predicts people moving further north
Workers in LA, TX, FL, SC lose over 10% ($6k/year)
Florida loses half its population by 2100
The role of anticipation
Suppose workers don’t move as a result of climate change because they don’t expect continued climate change
Don’t move away due to storms
Welfare in most affected locations declines more than if they move away
Exceeds 25% ($15k/year) on Atlantic coast
Capitalists benefit from lack of mobility due to higher returns in exposed counties
Global impact:
In US most affected locations are economically average, so migration has no strong large-scale impact
In less developed, smaller countries people will likely need to move out of the country
Globally this is good because these places are not that economically valuable
But, cross-border migration is politically challenging, can cause conflicts