190. Climate change, floods, and the future of auto supply chains

Episode Summary

Executive Summary: The episode examines how climate-driven floods disrupt global auto production and why carmakers respond by diversifying plants and maintaining spare capacity. Using global production and flood data, Juan MacCastro-Vincenzi shows floods reduce plant output persistently, firms shift some production to unaffected plants, and climate change will make resilience more valuable but also more expensive for consumers.

Main Topics: Climate shocks and the auto industry (Priority: 5/5): The episode opens with examples from Thailand, Detroit, South Africa, and Mexico to show that floods already disrupt automaking worldwide. How car supply chains are organized (Priority: 5/5): Cars rely on many bulky, geographically concentrated inputs, with final assembly plants near demand and many components sourced locally. Evidence from floods and plant-level production (Priority: 5/5): Using a global dataset of over 1,000 plants, the guest estimates how severe floods affect production and finds large, persistent output declines. Firm adaptation through multi-plant networks (Priority: 4/5): Multinational automakers reallocate some production across plants after floods, but recovery is partial and slow because spare capacity is limited. Climate change and future plant location choices (Priority: 5/5): Climate projections imply floods will become more frequent in major auto hubs, pushing firms toward more diversified and resilient production footprints. Costs to consumers and policy trade-offs (Priority: 5/5): Resilience lowers vulnerability to shocks but raises costs through smaller plants, lost scale economies, spare capacity, and potentially fewer car varieties. Broader relevance beyond autos (Priority: 3/5): The discussion extends the resilience trade-off to vaccines and semiconductors, where geographic concentration also creates vulnerability.

Key Arguments: Car production is highly complex and geographically concentrated, so local floods can disrupt both assembly plants and upstream suppliers. Automakers are multinational and multi-plant, which gives them flexibility to shift some production when one plant is hit. Historical flood events show that disruption is not just immediate; plant output can remain depressed for years after flooding. Flood impacts are often indirect, affecting infrastructure and local supplier networks rather than only the plant itself. Firms reallocate production to unaffected plants, especially those already making the same model, but only partially because spare capacity is limited. Climate change increases the frequency of extreme precipitation events in major auto-producing regions, making diversification more attractive. Resilience is economically costly because it reduces scale economies and requires maintaining unused capacity. Consumers may face higher car prices and fewer model/trim choices as firms pay for resilience and pass costs through. Policy should treat resilience like insurance: valuable, but with an ongoing premium that must be budgeted. Similar trade-offs appear in vaccines and semiconductors, where concentration boosts efficiency but increases shock exposure.

Data Points: Plants in the global dataset: More than 1,000 plants - Global car production dataset used in the research Countries covered: 54 countries - Geographic scope of the production dataset Average models per plant: About 5 to 6 models - Typical product variety at a car assembly plant Average plants per model: Around 3 plants - Typical number of plants producing a given model Highly common models: About 15 plants - Examples include Ford F-150 and Toyota Corolla Local inputs share: Around 60% - Share of car inputs sourced locally to the final assembly plant Median distance for engine/transmission input: 170 kilometers - Average distance of certain key parts from assembly plants Plants exposed to floods: Around 3 out of 4 plants - Share of plants experiencing a major nearby flood over 20 years Flood exposure window: 2000 to 2019 - Time span of the flood-production analysis Production decline after one year: Down by 20% - Average plant output one year after a severe flood Production decline after 10 years: Down by 32% - Long-run effect of a flood on average plant output Flood threshold in main analysis: 100 kilometers or less - Definition of a plant being affected by a nearby flood Alternative distance checks: 25 km, 50 km, 100 km - Robustness checks showing similar flood effects across thresholds Honda Celaya plant closure: 4 months - Honda plant in Mexico shut after 2018 flood Celaya plant output: About 50,000 cars per quarter - Scale of production interrupted by the flood Production shifted to another plant: Around 10,000 cars - Honda’s recovered HR-V production at another Mexican plant Recovered lost production: About 20% - Share of Celaya’s lost output recovered elsewhere Indiana disruption: 1 month - Honda engine-component disruption affected a U.S. plant Climate change flood frequency example: 1-in-10 years to 1-in-8 years - Illustrative change in flood probability for the Mexico plant location 100-year flood change: From 1% to 2% probability - Event frequency roughly doubling under climate projections Tennessee flood probability example: 2.2% - Projected probability for a 1% event in Tennessee Michigan flood probability example: 2.3% - Projected probability for a 1% event in Michigan Tokyo flood probability example: 1.4% - Projected probability for a 1% event in Tokyo Production at unaffected plants after flood: About 25% larger after five years - Measured reallocation to other plants over time

Pivotal Quotes: "these firms are going to hold more plants in the future, which allows them to diversify their production to these shocks. However, these plants are going to become smaller, right? Which is bad because they are losing scale." — Juan MacCastro-Vincenzi: Explaining the core resilience-versus-efficiency trade-off "By one year after, production on the average plant is down by 20%. And what I found really surprising is that it keeps declining. So 10 years after the flood happens, production is still down by around 32%." — Juan MacCastro-Vincenzi: Summarizing the persistent effect of floods on plant output "This is like an insurance policy, right? And we need to balance how much we value kind of goods and having low prices in both circumstances." — Juan MacCastro-Vincenzi: Describing the policy logic behind resilience investments

Implications: Resilience against climate shocks will likely mean more plants, more spare capacity, and higher car prices. Policymakers must weigh lower vulnerability during disasters against permanently higher costs in normal times.