Decriminalize, Then Recriminalize: Evidence from Colorado's Fentanyl Policy Reversal

Abstract

We study Colorado's unique two-stage policy experiment on fentanyl possession: reclassification of felony possession to misdemeanor in 2019 (HB 19-1263, effective May 28, 2019) followed by partial refelonization in 2022 (HB 22-1326, effective May 25, 2022). Using difference-in-differences with wild cluster bootstrap inference and synthetic control methods comparing Colorado to other U.S. states, we find statistically inconclusive effects of either policy change on overdose mortality. Colorado's fentanyl deaths increased by 811% from 2018 to 2023 (130 to 1,184), but this trajectory closely mirrors national trends driven by illicit fentanyl supply. Point estimates suggest the 2019 reclassification increased total overdose deaths by approximately 17%, with analytical confidence intervals spanning -31% to +96%. Permutation inference shows this effect is unremarkable compared to regional trends (p=0.72), and the synthetic control analysis confirms these inconclusive findings (permutation p = 0.96). Our results highlight the difficulty of identifying demand-side policy effects amid massive supply-side shocks and underscore the importance of appropriate small-sample inference methods in drug policy evaluation.