A reliability model for assessing corporate governance using machine learning techniques

Corporate governance assesses the efficiency and effectiveness of companies’ operations and decisions to ensure value creation for shareholders and optimal risk taking. As investors’ decision making process largely depends on financial information and corporate reports, transparency is capital for the stability of a company, or even the stability of a country via the corporate sector. This research introduces the system reliability theory to properly model the behaviour of companies regarding their corporate governance mechanisms. We propose the assessment of the corporate governance framework by mapping its inputs as components (either in operating or failed state) along with firm characteristics to determine an approximate Structure Function that enables alternatively modeling the functioning of the system, quantifying its reliability and detecting critical components. The advantage of the proposed mapping approach is illustrated using a sample of 1109U.S. listed companies during the period 2002–2014, reporting financial and non-financial information as components of the corporate governance system and the return on assets as the system output. The proposed approach is also useful for modelling other non-engineering sub-systems; companies, financial markets or even economies would be exposed to significant risk if these systems do not function properly.