The most common approaches in favor of liberalizing international trade and international finances suggest that opening policies will have a positive impact in reducing corruption. Despite of the complexity of studying these relationships, most research in this field is limited to correlational or deterministic studies. In this paper we applied a predictive model of classification based on neural networks called Multilayer Perceptron (MLP) that meets a set of desired statistics qualities, with the purpose of estimating the characteristics or symptoms present in countries categorized as more or less corrupt. Of the variables used, the levels of human development (HDI) and levels of economic openness are the common characteristics shared by countries with similar levels of corruption. We found evidence that shows us that if the economic openness level of a country is lower, the chance of being classified at a higher level of corruption increases.