The Utility of Neural Model in Predicting Tax Avoidance Behavior
| Autor: | Codruţa Mare, Coita Ioana-Florina |
|---|---|
| Rok vydání: | 2021 |
| Předmět: | |
| Zdroj: | Intelligent Decision Technologies ISBN: 9789811627644 KES-IDT |
| Popis: | The phenomenon of tax evasion and phenomenon of tax avoidance are two facets of similar behavior. Both affect societies across globe. It is a well-known fact that tax revenues account for the majority of public revenues and therefore the phenomenon is of great interest for public authorities as it is for other stakeholders. Designing effective tax policies aiming at maximizing public revenues can only be done through a deep and complete understanding of taxpayer behavior and its internal motivations. The present study aims at developing a model for detecting risk of tax fraud in taxpayer behavior by trying to predict propensity of individual showing intention for evading taxes. The purpose of this study is to model fiscal behavior through artificial intelligence, using MLP network model that was trained and tested on a real data set comprising behavioral or biometric data. Results were compared to that of a binary logistic regression. The empirical model tested was based on qualitative attributes, which relate to behavioral elements such as the attitudes and perceptions of taxpayers and also internal motivations, morality or personal values. The prediction efficiency of the MLP model relies on 70% which shows similar performance to comparable research. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|