2018/11/13 by Rastin Matin, Casper Worm Hansen, Matin, Rastin +5
Business, Management and Accounting · Decision Sciences · Economics, Econometrics and Finance · #Auditing, Earnings Management, Governance #Computation and Language (cs.CL) #Computational Finance (q-fin.CP) #Credit Risk and Financial Regulations #FOS: Computer and information sciences #FOS: Economics and business #Financial Distress and Bankruptcy Prediction #Risk Management (q-fin.RM) #Stock Market Forecasting Methods
paper · pdf · doi:10.48550/arxiv.1811.05270
openalex publication_date 2018/11/13 · openalex created_date 2022/08/02 · openalex updated_date 2026/07/28
Corporate distress models typically only employ the numerical financial\nvariables in the firms' annual reports. We develop a model that employs the\nunstructured textual data in the reports as well, namely the auditors' reports\nand managements' statements. Our model consists of a convolutional recurrent\nneural network which, when concatenated with the numerical financial variables,\nlearns a descriptive representation of the text that is suited for corporate\ndistress prediction. We find that the unstructured data provides a\nstatistically significant enhancement of the distress prediction performance,\nin particular for large firms where accurate predictions are of the utmost\nimportance. Furthermore, we find that auditors' reports are more informative\nthan managements' statements and that a joint model including both managements'\nstatements and auditors' reports displays no enhancement relative to a model\nincluding only auditors' reports. Our model demonstrates a direct improvement\nover existing state-of-the-art models.\n