FACULTÉ & RECHERCHE

The damage function used to assess the economic impact of secular changes in temperature is one of the most speculative components of integrated assessment models of climate change. Existing work informing this debate is based on pooled empirical models incorporating limited non-linearity and giving little regard to dynamics. We use aggregate and agricultural data for 151 countries over the past six decades to estimate dynamic heterogeneous models which (a) allow the weather-output nexus to differ freely across countries, (b) help distinguish short-run from long-run effects, and (c) account for unobserved time-varying heterogeneity. Overall, we find that, in low-income or high-temperature countries, a permanent 1 °C rise in temperature is associated with a fall in income per capita of about 1.3% in the short-run and 8.5% in the long run. These long-run effects are substantially larger than those commonly suggested in the literature.