Abstract:
Döbbeling-Hildebrandt et al. (2024, DH2024) conduct a meta-analysis of the effectiveness of carbon pricing. DH2024's abstract concludes that 17 of 21 schemes evaluated in the literature produced substantial emissions reductions. A subsequent press release was headed: "Carbon pricing works". This comment revisits the meta-analysis and examines whether its empirical evidence supports the claims made in DH2024's abstract and, notably, the press release. We use DH2024's own approach of accounting for statistical power and potentially biased causal inference in the underlying studies. We show that when these criteria are applied simultaneously and conservatively - which we argue they should be - only nine effective schemes remain, eight in China and one regional US scheme. We emphasize that statistical power is a major issue in most carbon pricing evaluations, because most carbon prices are very low, leading to weak signal-to-noise ratios. We conclude that DH2024's policy implications and its press release therefore cannot be squared with its evidence base.
