Studies that test the effect of economic outcomes on political transitions using weather variations as instruments have generally overlooked findings from climate science that economic output is a hill-shaped, rather than linear, function of temperature and precipitation levels. We design an improved set of instruments for growth based on this fact, and find that growth-maximizing temperatures coincide with levels that maximize energy sector output in the climate response literature. Previous studies significantly overestimate the increase in the probabiliTY of democratic transitions resulting from negative growth shocks, although we find leadership transition frequencies rise significantly following transitions to democracy.