Predictive Analytics as an Instrument to Prevent Bankruptcy

Abstract

As of today there are a lot of well-known bankruptcy prediction models. Scientists have been paying much attention to the development of bankruptcy prediction models since 1970. However, most of them are unable to predict bankruptcy, thereby making it impossible for firms to prevent it today. The paper researches predictive ability of existing bankruptcy prediction models suitable for small business. The primary goal of this paper is to examine methods of predictive analytics on empirical data and use obtained results to prevent bankruptcy of firms. Combination of predictive analytic methodology with bankruptcy prediction models’ testing made it possible to identify the models having high predictive ability. The study was carried out by using data bases of accounting reports of Russian’s firms. The study is based on the data from fifty small enterprises, divided into two types: failed and non-failed firms. Common-size and index analysis, financial ratios method and multidimensional statistical analysis were used to achieve the solution of the study.

This paper makes these contributions: 1) summarizes methods of predictive analytics that indicate approaching bankruptcy; 2) evaluates the accuracy of bankruptcy prediction models one, two or three years before bankruptcy; 3) identifies models showing high predictive ability among small firms and provides a suitable model for small businesses. The results provided in the paper would be useful to many users such as scholars, financial analysts, board of small enterprises, lenders, auditors and tax inspectors.

Keywords: predictive analytics; big date; model; discriminant analysis; bankruptcy; small business

JEL Classification: C51, C53, C58, C65, C83, G32, G33,O16