What drives financial competitiveness of industrial sectors in Visegrad Four countries? Evidence by use of machine learning techniques

Abstract

This article presents machine learning (ML)-based empirical research with a specific focus on the financial competitiveness of different industrial sectors in Visegrad Four (V4) countries. Financial competitiveness is measured by the two most widely applied profitability ratios: return on assets (ROA) and return on equity (ROE). Several sectoral average financial ratios are considered as input variables from the 4 countries and 27 sectors, with data collected between 2016-2020 in a cross-sectional approach. Explorative data analysis reveals that the three strongest clustering features of V4 sector-level financial data are found in country classification, total assets per employee, and gross margin ratios. Hypothesis examination has justified a view that drivers of financial competitiveness are not necessarily identical to factors explaining variance between sectoral average financial ratios. Six methods have been applied to develop predictive models for ROA and ROE. Results demonstrate that the traditional generalized linear model (GENLIN) delivers insufficient predictive power despite fulfilment of each statistical assumption. The k-nearest neighbor (KNN) and random forest (RF) methods are demonstrated to be the best ML techniques to predict the sectoral financial competitiveness of V4 companies. Beyond country classification, the best predictors of ROA and ROE at the V4 sectoral level are found in income margin, turnover, and leverage ratios as compressed components by use of principal component analysis (PCA). The article also provides added value to literature on sectoral and financial competitiveness research, analysis of financial features of V4 companies, and the efficient application of ML methods.