The empirically inscrutable climate-economy relationship

Empirical models of the macro-level climate-economy relationship project climate change impacts ranging from trivial to catastrophic. This wide range suggests identification challenges. Here, we show that such models indeed face identification challenges that are axiomatic and irreducible. The climate-economy relationship has qualitatively important variation across space and time, due to complex regional heterogeneity and affluence-driven adaptation, among other factors. Empirical models must assume away some of this variation to preserve degrees of freedom. Influential observations caused by economic disasters and miracles, along with low signal-to-noise ratios, create additional challenges. We illustrate these challenges theoretically and then empirically by replicating and sensitivity testing prominent climate-econometric studies. The estimation challenges create economically meaningful sensitivities and fragilities. We recommend a cautious and uncertainty-emphasizing approach to applying climate econometrics to public and private decision-making. Importantly, however, our analysis does not dismiss the existence of a negative carbon externality nor economically beneficial efforts to reduce greenhouse gas emissions.