Regresar

Analysis for possible tax evasions from the value added tax in Ecuador using an stochastic model with a non-parametric technique

Abstract:

This article establishes a methodology based on stochastic techniques, with the objective of creating a model that allows the identification of possible tax evaders of the value added tax (VAT) in Ecuador. A population of taxpayers that represents more than 90% in VAT collection during the fiscal periods between 2014 to 2017 is used, which represents a bid data mining base of approximately 86,400,000 records of the SRI authority, grouped by activity codes then being called economic conglomerates; Atypicalities are analyzed in the temporal behavior by creating a monthly sales index. The outliers are identified by Gertensgarbe and Werner’s nonparametric technique, determining thresholds that differentiate usual and unusual values, resulting in 32 contributors that show higher values of the index. Billing your customers and suppliers is compared. The results determine a high probability of sub declarations in the VAT and/or increasing accumulations of tax cbkp_redit, considering signs of tax evasion.