Using MLT To Estimate Corruption Patterns From The BEEPS

Abstract

The paper describes the opportunity for using machine learning technologies (MLT) for estimating corruption by clustering. We used the enhanced BEEPS data (The Business Environment and Enterprise Performance Survey (European Bank for Reconstruction and Development, 2014)). It contains 1672 variables and 59619 observations produced by well-respected agencies like Nielsen for the European Bank of Reconstruction and Development. The analysis of different indicators with the MLT allows us to cluster the countries by the types of potential corruption patterns. We suggested this method could overcome the shortcomings of the classical survey surveillance approach because we can estimate countries with some distortion or insufficiencies in the data (for example, when the business units may want to lie about the corruption due to some reasons). This gives us an additional measurement that can be used for analyzing the true corruption field. This can be useful for business units, scientific people, and policymakers for analyzing the patterns of corruption in different countries.