Flexible quantile-based modeling of bivariate financial relationships: The case of ROA ratio

Ratios used in financial analysis suffer from several drawbacks, and the tools – ranging from linear least-squares regressions to neural networks – suggested as alternatives also have serious disadvantages. We propose an alternative approach, based on quantile regression techniques, which exploits financial information in a more efficient way, not achievable by conventional tools. Our proposal is applied to the ROA (return on assets) ratio, this being one of the most popular ratios among both economic analysts and researchers. An empirical analysis is carried out on real data. Results indicate that the quantile approach provides a more accurate assessment of the financial position of the firm.