This thesis aims at assessing the effectiveness of carbon pricing policies in reducing carbon emissions using replication analysis, regression analyses and metaanalyses. Three carbon pricing regimes are chosen as the focuses of the analyses, including the carbon taxes in Sweden and Norway and the pilot emission trading schemes (ETS) in China.

The replication in Chapter 2 reproduces and confirms Andersson (2019a)’s findings that the Swedish carbon taxes significantly reduced per capita carbon emissions (by 6.3%) in the transport sector since its introduction in 1991. As I further extended Andersson’s approach to Norway’s transport sector, I found little effect (2.6%). In Chapter 3, I assessed the carbon mitigating effect of China’s pilot ETS program using advanced econometric techniques to control for the impact of confounding factors. The estimated effects vary from a statistically insignificant reduction of 12.5% in per capita CO2 emissions and to a significant 18.0% reduction.

The results in Chapter 2 and Chapter 3 indicates the heterogeneity between estimated effect sizes across different countries, sectors and estimators. Thus, this thesis further conducts meta-analyses in Chapter 4 and Chapter 5 to synthesize the literature estimating the effect of China’s pilot ETS program. Chapter 4 uses a “multiverse” approach to generate effect sizes from 8192 regressions, basing on the same emission data set but varying the dependent variables, the time of treatment, the estimators and the control variables included in the model. Based on these estimates, a single-paper meta-analysis is conducted. The resulted estimate of the overall mean effect is 18.0% reduction in carbon emissions for the ETS pilots.

In addition, a standard meta-analysis is conducted in Chapter 5, with 752 estimates from 80 studies drawn from the literature. To address inflation in estimated effects caused by publication bias, Chapter 5 employs procedures to adjust estimates for publication bias. The associated estimates of overall mean effect range between - 0.101 and -0.057 (using Fisher’s z scale). Several factors are identified as influential to the estimated effect sizes, including empirical methods such as “matching between the treated and control units before regression” and whether or not to control some covariates such as population.

Overall, I found strong evidence that the carbon pricing policies reduced carbon emissions in countries and regions with such policies. But the effect of carbon pricing policies also seems to be country/region/sector specific. Thus, researchers and policymakers should be careful when drawing conclusions about the effectiveness of carbon pricing policies.