Financial ratios are often used in cluster analysis to classify firms according to the similarity of their financial structures. Besides the dependence of distances on ratio choice, ratios themselves have a number of serious problems when subject to a cluster analysis such as skewed distributions, outliers, and redundancy. Some solutions to overcome those drawbacks have been proposed in the literature, but have proven problematic. In this work we put forward an alternative financial statement analysis method for classifying firms which aims at solving the above mentioned shortcomings and draws from compositional data analysis. The method is based on the use of existent clustering methods with standard software on transformed data by means of the so-called isometric logarithms of ratios. The method saves analysis steps (outlier treatment and data reduction) while defining distances among firms in a meaningful way which does not depend on the particular ratios selected. We show examples of application to two different industries and compare the results with those obtained from standard ratios.
Financial ratios, i.e., ratios comparing the magnitudes of accounts in financial statements, constitute a case of researchers' and professionals' interest in relative rather than absolute account magnitudes. From the classical work on bankruptcy prediction by Altman (1968), the use of financial ratios has spread along and across many research lines (Willer do Prado et al., 2016), such as stock market returns (e.g., Dimitropoulos, Asteriou, & Koumanakos, 2010), firm survival analysis (e.g., Kalak & Hudson, 2016), credit scoring (e.g., Amat, Manini, & Antón Renart, 2017), assessing the impact of International financial reporting standards (e.g., Lueg, Punda, & Burkert, 2014), predicting donations to charitable organizations (e.g., Trussel & Parsons, 2007), accounting restatements (e.g., Jiang, Habib, & Zhou, 2015), and earnings manipulation (e.g., Campa, 2015). This article focuses on another frequent use of financial ratios: to classify firms according to similarity of the structure of their financial statements, searching for different profiles of financial structure, performance or distress. Since the seminal works of Cowen and Hoffer (1982), and Gupta and Huefner (1972), through the relevant contributions by Dahlstedt, Salmi, Luoma, and Laakkonen (1994); Ganesalingam and Kumar (2001); Mar Molinero, Apellaniz Gomez, and Serrano Cinca (1996); Serrano Cinca (1998); and Voulgaris, Doumpos, and Zopounidis (2000), the interest in clustering firms according to their financial ratios remains current (Feranecová & Krigovská, 2016; Lukason & Laitinen, 2016; Luptak, Boda, & Szucs, 2016; Martín-Oliver, Ruano, & Salas-Fumás, 2017; Momeni, Mohseni, & Soofi, 2015; Santis, Albuquerque, & Lizarelli, 2016; Sharma, Shebalkov, & Yukhanaev, 2016; Yoshino & Taghizadeh-Hesary, 2015; Yoshino, TaghizadehHesary, Charoensivakorn, & Niraula, 2016).