The inability of a simple real business cycle model to predict a rise in consumption in response to increasing government spending has stimulated the development of alternative models, which have all been used to evaluate the effects of government spending shocks. We quantitatively investigate transmission mechanisms for government spending shocks proposed in the literature and use a Bayesian approach in order to identify the one that fits the data best. We find that the mechanism featuring deep habits outperforms all others considered, while the mechanism relying on non-Ricardian “rule-of-thumb” consumers provides the poorest fit.