EASI Drugs in the Streets of Colombia: Modeling Heterogeneous and Endogenous Drug Preferences

The response of illicit drug consumers to large-scale policy changes, such as legalization, is heavily mediated by their demand behavior. Since individual drug use is driven by many unobservable factors, accounting for unobserved heterogeneity is crucial for modeling demand and designing targeted public policies. This paper introduces a finite Gaussian mixture of Exact Affine Stone Index (EASI) demand systems to estimate the joint demand for marijuana, cocaine, and basuco (cocaine residual or "crack") in Colombia, accounting for corner solutions and endogenous price variation. Our results highlight the importance of unobserved heterogeneity in identifying reliable price elasticities. The method reveals two regular consumer subpopulations: "safe" (recreational) and "addict" users, with the majority falling into the first group. For the "safe" group, whose estimates are precise and nationally representative, all three drugs exhibit unitary price elasticities, with cocaine being complementary to marijuana and basuco an inferior substitute to cocaine. Given the low production cost of marijuana in Colombia, legalization is likely to drive prices down significantly. Our counterfactual analysis suggests that a 50% price decrease would result in a \$363 USD gain in utility-equivalent expenditure per representative consumer, \$120 million USD in government tax revenue, and a \$127 million USD revenue loss for drug dealers. Legalization, therefore, has the potential to reduce the incentive for drug-related criminal activity, the current largest source of violent crime in Colombia.