Some like it cold: Heterogeneity in the temperature-economy relationships of Europe

Econometric analysis of the global impact of temperature fluctuations on Gross Domestic Product (GDP) suggests a generalised coupled-climate-economy relationship in which higher temperatures harm warm countries, benefit cooler ones and an "optimal" temperature exists in between1–3. However, aggregate temperature-GDP relationships reflect an average across different spatial scales and sectors of the economy, thereby masking any underlying heterogeneity. Ignoring such heterogeneity could misrepresent the overall costs or benefits of temperature change and provide misleading guidance for mitigation and adaptation policies. Focusing on Europe, we use administrative district level data on the growth rate of Gross-Value Added (GVA) and GDP to estimate the impact of temperature fluctuations on economic growth at the national level, the district level, and by industry. Unlike previous studies with a global focus, for Europe we find negative effects of warmer-than-average years on GVA and GDP in relatively cold districts (annual mean temperature -1 to 15 degrees Celsius). In warmer regions (15 to 25 degrees Celsius) warmer-than-average years tend to increase economic output. Warmer temperatures only benefit extremely cold (<-1C) districts and are only costly to extremely warm (>25C) districts. Across Europe this robust U-shaped temperature-GDP relationship implies a -0.22 (95% CI: [-0.19, -0.25]) percentage point reduction in economic growth, rather than a benefit of +0.18 ([0.15, 0.21]) percentage points implied by the inverted-U-shaped relationship from previous studies. Disaggregating by economic sector, the negative temperature effect in cold countries stems primarily from the production of goods (e.g. agriculture and construction) not services. Disaggregating by district, the marginal temperature effects are highly heterogeneous, even within countries. Overall, our results invert the narrative that warmer than average temperature will be economically advantageous to colder regions, point to regional vulnerabilities stemming from specialisation, and suggest local temperature optima not a global one.