Kernel density decomposition with an application to the social cost of carbon
A kernel density is an aggregate of kernel functions, which are itself densities and could be kernel densities. This is used to decompose a kernel into its constituent parts. Pearson's test for equality of proportions is applied to quantiles to test whether the component distributions differ from one another. The proposed methods are illustrated with a meta-analysis of the social cost of carbon. Different discount rates lead to significantly different Pigou taxes, but not different growth rates. Estimates have not varied over time. Different authors have contributed different estimates, but these differences are insignificant. Kernel decomposition can be applied in many other fields with discrete explanatory variables.
READ FULL TEXT