This paper tests the out of sample predictive ability of reduced form models found in the literature on forecasting CO2 emissions. We show that for a newly available panel data set covering the fifty U.S. States and Washington D.C. during the years 1960 - 2000, the benchmark models found in the literature are outperformed by over 3000 of the considered models. A search over a large universe of models selects a ”best” performing model using the mean square forecast error from a simulated out of sample forecast experiment. Forecasts of aggregate CO2 emissions for the United States are provided.

This paper provides comparisons of a variety of time series methods for short run forecasts of the main greenhouse gas, carbon dioxide, for the United States, using a recently released state level data set from 1960-2001. We test the out-of-sample performance of univariate and multivariate forecasting models by aggregating state level forecasts versus forecasting the aggregate directly. We find evidence that forecasting the disaggregate series and accounting for spatial effects drastically improves forecasting performance under Root Mean Squared Forecast Error Loss. Based on the in-sample observations we attempt to explain the emergence of voluntary efforts by states to reduce greenhouse gas emissions. We find evidence that states with decreasing per capita emissions and a "greener" median voter are more likely to push towards voluntary cutbacks in emissions.