To internalize pollution externalities of municipal waste, Pay-As-You-Throw (PAYT) pricing schemes for unsorted waste are adopted worldwide. This paper estimates causal effects of PAYT on unsorted, recycling and total waste generation, and analyzes the determinants of policy adoption and compliance. The empirical strategy uses a multiple random forest approach on a unique six-year panel data on a large set of covariates for about all municipalities in Italy. Policy adoption is found to be most likely for municipalities with high recycling, and heterogeneous in socio-economic and political variables. Results show that PAYT is on average effective, decreasing per capita unsorted waste up to -72%. Effects are mostly driven by an increase in recycling (up to 40%), and less by waste reduction (up to -19\%). Analyzing heterogeneity reveals that total waste increases (rebound effects) are possible when households have low opportunity costs of recycling, and low education. Thereby, targeting these municipalities with awareness-raising campaigns could increase PAYT effectiveness.