A machine learning approach to find the determinants of Peruvian cocaine local price

Abstract

The coca leaf has many uses in the Peruvian culture. Although there are legal usages, people employ coca for illicit business. The most infamous illegal use is cocaine production. The cocaine business is highly profitable, but it harms human health. Then, what are the determinants of cocaine price? The current analysis aims to get the variables with the capability to explain the cocaine prices in Peru. The period analyzed is 2003-2019. The study gathered variables from DEVIDA and UNDOC databases. The Lasso technique selected the variables with the best capability to predict cocaine price. Those variables were: ENACO acquisition, coca seizures, and coca crops. OLS, VAR, and Granger analyses employed those variables to analyze the relationship between them. According to the OLS analysis, both ENACO acquisition and coca crops had adverse effects on cocaine prices, while coca seizures were positively related to the cocaine price. VAR analysis showed that only ENACO acquisition had a short-term relationship with the dependent variable. Moreover, it showed that the whole set of variables influenced the dependent variable. The Granger analysis proved that there was a cause-effect relationship between ENACO acquisition and cocaine price. Hence, the ENACO purchases expansion can rest the attractiveness of illegal groups to farmers. However, low-ering cocaine prices might attract more users. Therefore, educational activities are also required.