The carbon exposure of a stock is estimated by regressing its returns over the past 36 months on total US energy CO2 emissions data from the US Energy Information Administration. A hypothetical portfolio that goes long stocks with low CO2 emission betas and short the high emission beta stocks is found to earn average annual returns of 3.6-5.3%.
The authors argue that the negative relation between carbon exposure and future returns is consistent with the view that CO2 emissions are indicative of perceived deterioration of investment opportunities by investors. The results are inconsistent with the view that carbon exposure may be rewarded with a risk premium, as the sign is found to be the other way around.
We like this paper, but have some methodological concerns. In particular, the aggregate CO2 emissions data shows a very strong seasonal pattern for which the authors do not seem to make adjustments, but which might have a large effect on their estimated emission betas.