Abstract

It has been empirically demonstrated that the average level of disclosure in Latin American universities is low and very heterogeneous (Abello 2018; Abello et al. 2018). Attributes of university corporate governance may explain the level of disclosure. However, only the direction of the effect of these variables on disclosure is currently understood. In order to advance in the understanding of disclosure processes and the incidence of corporate governance attributes, a typification of universities can be made. Using a quantitative approach, this paper aims to typify 219 Latin American universities according to their level of disclosure of information, considering the attributes of their university corporate governance. A non-parametric technique called regression trees is introduced to achieve the objective, which is used in many disciplines and is part of machine learning techniques. The results indicate that there are four nodes (or clusters) of universities whose levels of disclosure are different, and the attributes of their university corporate governance are closely related. These findings allow us to propose public policies to increase information disclosure in a more targeted way.