ECONOMIC OUTLOOK FROM CLIMATE CONDITIONS: PREDICTIVE DENSITIES THROUGH MULTILEVEL FACTOR MODELS

C. VLADIMIR CABALLERO-RODRÍGUEZ

Assessing macroeconomic risk means characterising the full predictive distribution of the economic outlook, not just its conditional mean.
In this talk, I present a current extension of the macro risk program: conditioning macroeconomic outlook directly on climate. Fitting a multilevel dynamic factor model by EM/Kalman filtering to monthly centre and log-range temperatures at 880 US locations (1931–2024), we extract global and regional
temperature-level and temperature-dispersion factors and feed them into factor-augmented quantile regressions for US growth, inflation, and unemployment. Temperature dispersion,
not its level, carries independent information about downside growth risk; the effect is horizon-dependent and concentrated in the lower tail, while inflation and unemployment respond more weakly and selectively. I close with the natural next step: extending the framework to a copula-based joint predictive
density for growth and inflation, and eventually to climate-stress scenarios, an ongoing project, Outlook-at-Risk.