Abstract: Accurate financial distress prediction models are of critical importance to various stakeholders, i.e. management, investors, employees, shareholders and other interested parties, as the models provide them with timely warnings. The selection of appropriate indicators, which has significant impact on the accuracy of the prediction model, has been widely studied by researchers. Generally, most of the prediction models presented in the literature select indicators directly from various financial ratios based on information that appears in the corporations' financial statements. However, it is widely recognized that a main cause of financial distress is poor management. Therefore, it is believed that the efficiency of business operations should be included in the financial distress prediction model.
Different from previous researches that construct a prediction model based on the integration of financial ratios and a single-dimensional efficiency indicator, it is argued that multi-dimensional efficiency indicators can comprehensively and effectively reflect the overall situation of corporations in various industries with different scales. Therefore, they should be considered in financial distress prediction in order to achieve better performance. In this study, four input-output systems are put forward for efficiency evaluation with respect to four dimensions, i.e., operational efficiency, financial efficiency, financing efficiency and human capital efficiency. The data envelopment analysis is employed to evaluate the relative efficiencies of corporations in each dimension. And then, the financial distress prediction model, which integrates the multi-dimensional efficiency indicators and financial ratios, is established.
In order to verify the validity of the proposed model, 148 listed corporations in Shanghai Stock Exchange are selected and Shenzhen Stock Exchange for empirical study. Among them, 74 corporations are financial healthy, while the other 74 corporations fall into financial distress. Three commonly used financial distress prediction techniques, i.e., support vector machine, artificial neural network and decision tree, are embedded in the model. To further investigate the effectiveness of the proposed model, we adopt three different financial ratios systems are adopted as benchmarks. The results show that the proposed model that combines multi-dimensional efficiency indicators and financial ratios can effectively improve the accuracy of financial distress prediction. The findings of the empirical study also reveal that when different financial ratios are used as the input variables, adding multi-dimensional efficiency indicators could constantly improve the prediction performance, which exhibit the robustness of the proposed approach.

Key words: financial distress prediction, data envelopment analysis, support vector machine, artificial neural network, decision tree