Tax Administrations have been strengthening their control actions by introducing significant changes involving regulations and access to local and international information.

Nevertheless, such progress is insufficient if not focused on more precisely identifying the tax risk behaviors and patterns for an adequate and timely selection of operations that should be subjected to control.

This objective may be achieved through the development of tax risk models based on adequately applied data mining techniques, according to the knowledge and experience of the technical staff which may focus all its efforts to obtain objective and optimum results that may entail sufficient tax collection.

The Internal Revenue Service- SRI, as the entity in charge of control has created a transfer pricing risk model whose implementation in Ecuador endeavors to continue strengthening its struggle against tax evasion and fraud.

In structuring the risk model, the SRI’s specialized technical working team involved in the project the executing units of the transfer pricing examination processes at the national level. It also took into consideration the experience of the Tax Administration Service – SAT of Mexico. Thereafter, it applied the CRISP-DM (Cross Industry Standard Process for Data Mining) methodology to determine the tax risk probability, based on the understanding and preparation of the information available at the SRI for developing the main risk variables and indicators in the application of transfer pricing.

This information was then used to model the data by means of artificial intelligence techniques (neuronal networks), analysis of the main components, Kohonen networks, K-means clusters, Two-step clusters, decision trees, optimum bands and text mining. It is worth noting that the population examined corresponds to the companies that carried out operations with related parties abroad and the information analyzed covers the 2012- 2017 fiscal periods.

To facilitate the analysis and selection of taxpayers to be controlled, the results obtained from the model were used to establish a risk score. That is, a value between zero and one thousand points calculated on the basis of the probabilities generated from the mathematical model, to finally assign each case a low, medium or high risk category.

Due to the applied technical specialty, we are convinced that the transfer pricing risk model will be a strategic tool for the objective selection of tax audit (examination) cases whose effectiveness may be maintained through time. This will be so, to the extent that the experiences resulting from its application are taken into consideration for its continuous improvement.

On behalf of the entire working team, I wish to thank the United Nations Organization, GIZ German Cooperation Agency (which financed the technical assistance); the Inter-American Center of Tax Administrations-CIAT; the Tax Administration Service-SAT of Mexico, as well as the staff of these important institutions, especially Gonzalo Arias, Irving Ojeda, Carlos Pérez Gómez Serrano, Miguel Gaeta, Eduardo Díaz and Mauricio Zavala, whose collaboration and support rendered possible this important project which has resulted in the creation of the transfer pricing risk model in Ecuador.

1,264 total views, 4 views today