A Study on Financial Characteristics and Tax Avoidance Using Bayesian Network

This study analyzes the financial characteristics affecting the tax avoidance. This study would overcome the limitations of the analysis method of the multiple regression widely used for a long time in the existing studies. In other words, the analysis method by multiple regression can not reflects the complex real world, assuming the uniform linearity between dependent variables and explanatory variables. Thus, this study drew the causal relationship existing in various financial characteristics affecting the tax avoidance. This study could help various accounting information users surrounding the companies concerning what financial characteristics have impacts on the tax avoidance through a direct or indirect causal relationship with other financial characteristics. For this purpose, this study used Bayesian Network, and Markov Blanket drawn through this was proposed. This study analyzed companies listed on the Korea Exchange for 10 years from 2001 through 2010, and according to the result of the analysis, this study has contributions that all methods used in this study provided statistically significant results.