Governança corporativa e conservadorismo contábil: um estudo em empresas da B3

Abstract

This study aimed to analyze the advisory value (expertise and sociopolitical influence) of the board members of companies listed on B3, considering them as relevant parameters for impacting accounting conservatism. The main independent variables were Expertise (education, experiences, skills, and competencies) and Sociopolitical Relations (political exposure and socio-community ties) as more specific mechanisms to support good corporate governance. The dependent variable used was Basu's Accounting Conservatism. The data for this study were secondary, with a sample consisting of 50 firms from the Brazilian Stock Exchange (B3), categorized into cyclical and non-cyclical entities, observed between 2017 and 2019. The analytical method used in this study was the balanced panel data analysis, using E-Views 12. The results indicated that the analyzed companies had board members with more expertise characteristics compared to sociopolitical ties. This parameter (expertise) had a positive and significant effect on accounting conservatism, suggesting that board members with more experience and technical-professional skills tend to increase the level of conservatism, and therefore, corporate governance, helping companies to report losses beforehand. The variable defining sociopolitical characteristics did not show statistical significance, although it had a negative effect, as defined methodologically. The combined model of both variables (hybrid) showed a moderate, significant, and positive effect, indicating that the advisory value of board members impacts the level of accounting conservatism. These results support hypotheses 1 and 3. However, other governance variables added to improve control did not show significance, while the financial variables such as company size and debt showed a significant effect on the early reporting of undesirable accounting results. The models presented a moderate adequacy of the adjusted coefficient of determination (adjusted R²), with 53.87% for Model 1 and 41.36% for Mathematical Representation 2, both above the figures reported in previous literature for this estimator (conservatism), distinguishing this study from others in its field. The study suggests that using larger samples and expanding the methodology, especially by evaluating other professional characteristics as proxies for community-political influence, may yield more robust results with panel data. Furthermore, considering other variables to control for the effects of strict Brazilian regulations could be beneficial.