基于文本挖掘的上市公司财务风险预警研究

doi:10.3778/j.issn.1002-8331.2107-0423

摘要/Abstract

摘要： 上市公司年报中的描述性文本信息是上市公司信息披露的重要组成部分，通过对上市公司信息披露文本的挖掘与分析可以提高对其财务风险的预测能力。基于BERT（bidirectional encoder representations from transformer）模型与自编码器（autoencoder，AE），提出了BERT-AE融合文本特征提取模型，提取A股市场531家上市公司年报中“经营情况讨论与分析”和“审计报告”的文本特征，构建能够反映财务困境公司与正常公司的文本特征指标，随后将文本特征指标与财务指标数据结合，分别使用Logistic回归、极端梯度提升（extreme gradient boosting，XGBoost）、人工神经网络（artificial neural networks，ANN）、卷积神经网络（convolutional neural networks，CNN）四种模型，检验加入文本特征指标后财务风险预测的准确性是否得到提高，并使用Word2Vec-CNN-AE、Word2Vec-LSTM-AE模型提取财务文本特征进行对比实验。结果表明，三种模型提取的财务文本特征均能使财务预警模型预测的AUC得到提升，且BERT-AE模型提取的财务文本特征使得四种财务预警模型预测的AUC值提升效果更为显著，表明BERT-AE模型有效地提取了财务文本特征，提高了上市公司财务风险预警模型的预测能力。

关键词: 财务风险预警, 文本挖掘, BERT模型, 自编码器, 文本特征

Abstract: The descriptive text information in the annual report of a listed company is an important part of the information disclosure of a listed company. Through the mining and analysis of the information disclosure text of a listed company, the ability to predict its financial risks can be improved. Based on the BERT（bidirectional encoder representations from transformer） model and autoencoder（AE）, this paper proposes a BERT-AE fusion text feature extraction model, which extracts the text features of “business discussion and analysis” and “audit report” in the annual reports of 531 listed companies in the A-share market, and constructs the text feature indicators which can reflect financially distressed companies and normal companies, and then combines the text feature indicators with financial indicator data, uses Logistic regression, extreme gradient boosting（XGBoost）, artificial neural networks（ANN） and convolutional neural networks（CNN） four models to test whether the accuracy of financial risk prediction is improved after adding text feature indicators, and uses Word2Vec-CNN-AE and Word2Vec-LSTM-AE model to extract financial text features for comparative experiments. The results show that the financial text features extracted by the three models can all improve the AUC predicted by the financial early warning model, and the financial text features extracted by the BERT-AE model make the AUC value predicted by the four financial early warning models more significant, indicating that the BERT-AE model effectively extracts the financial text features and improves the predictive ability of the financial risk early warning model of listed companies.

Key words: financial risk early warning, text mining, BERT model, AutoEncoder, text features

梁龙跃, 刘波. 基于文本挖掘的上市公司财务风险预警研究[J]. 计算机工程与应用, 2022, 58(4): 255-266.

LIANG Longyue, LIU Bo. Research on Financial Risk Early Warning of Listed Companies Based on Text Mining[J]. Computer Engineering and Applications, 2022, 58(4): 255-266.

参考文献

[1] TENNYSON B M，INGRAM R W，DUGAN M T.Assessing the information content of narrative disclosures in explaining bankruptcy[J].Journal of Business Finance & Accounting，1990，17（3）：391-410.
[2] 陈艺云.基于信息披露文本的上市公司财务困境预测：以中文年报管理层讨论与分析为样本的研究[J].中国管理科学，2019，27（7）：23-34.
CHEN Y Y.Forecasting financial distress of listed companies with textual content of the information disclosure：a study based MD&A in Chinese annual reports[J].Chinese Journal of Management Science，2019，27（7）：23-34.
[3] MATIN R，HANSEN C，HANSEN C，et al.Predicting distresses using deep learning of text segments in annual reports[J].Expert Systems with Application，2019，132：199-208.
[4] ALTMAN E I.Financial ratios，discriminant analysis and the prediction of corporate bankruptcy[J].The Journal of Finance，1968，23（4）：589-609.
[5] OHLSON J.Financial ratios and the probabilistic prediction of bankruptcy[J].Journal of Accounting Research，1980，18（1）：109-131.
[6] ODOM M D，SHARDA R.A neural network model for bankruptcy prediction[C]//1990 IJCNN International Joint Conference on Neural Networks，1990.
[7] 杨淑娥，王乐平.基于BP神经网络和面板数据的上市公司财务危机预警[J].系统工程理论与实践，2007（2）：61-67.
YANG S E，WANG L P.Research on financial warning for listed companies by using BP neural network and panel data[J].Systems Engineering-Theory & Practice，2007（2）：61-67.
[8] 方匡南，杨阳.SGL-SVM方法研究及其在财务困境预测中的应用[J].统计研究，2018，35（8）：104-115.
FANG K N，YANG Y.SGL-SVM with its application in forecasting corporate financial distress[J].Statistical Research，2018，35（8）：104-115.
[9] 宋歌，马涛.基于深度学习的上市公司财务风险预警模型研究[J].价值工程，2019，38（1）：53-55.
SONG G，MA T.Research on financial risk early warning model of listed companies based on deep learning[J].Value Engineering，2019，38（1）：53-5.
[10] WANG G，WANG K M，ZHOU Y Y，et al.Establishment of a financial crisis early warning system for domestic listed companies based on three decision tree models[J].Mathematical Problems in Engineering，2020：8036154.
[11] SHIRATA C Y，TAKEUCHI H，OGINO S，et al.Extracting key phrases as predictors of corporate bankruptcy：empirical analysis of annual reports by text mining[J].Journal of Emerging Technologies in Accounting，2011，8（1）：31-44.
[12] BACH M P，KRSTI? ?，TURULJA L，et al.Text mining for big data analysis in financial sector：a literature review[J].Sustainability，2019，11（5）：1277.
[13] HáJEK P，OLEJ V.Evaluating sentiment in annual reports for financial distress prediction using neural networks and support vector machines[C]//14th International Conference on Engineering Applications of Neural Networks，Halkidiki，Sep 13-16，2013.Berlin，Heidelberg：Springer，2013.
[14] HáJEK P，OLEJ V，MY?KOVá R.Forecasting stock prices using sentiment information in annual reports—a neural network and support vector regression approach[J].WSEAS Transactions on Business & Economics，2013，10（4）：293-305.
[15] 谢德仁，林乐.管理层语调能预示公司未来业绩吗?——基于我国上市公司年度业绩说明会的文本分析[J].会计研究，2015（2）：20-27.
XIE D R，LIN L.Do management tone help to forecast firms’ future performance of the company?—A text analysis based on annual earnings communication conferences of listed companies in China[J].Accounting Research，2015（2）：20-27.
[16] HENRY E.Are investors influenced by how earnings press releases are written?[J].Social Science Electronic Publishing，2008，45（4）：363-407.
[17] LOUGHRAN T，MCDONALD B.When is a liability not a liability? Textual analysis，dictionaries，and 10-Ks[J].Journal of Finance，2011，66（1）：35-65.
[18] 陈艺云，贺建风，覃福东.基于中文年报管理层讨论与分析文本特征的上市公司财务困境预测研究[J].预测，2018，37（4）：53-59.
CHEN Y Y，HE J F，QIN F D.Financial distress prediction based on the textual characteristics of management discussion & analysis in the Chinese annual reports[J].Forecasting，2018，37（4）：53-59.
[19] DEVLIN J，CHANG M W，LEE K，et al.BERT：pre-training of deep bidirectional transformers for language understanding[J].arXiv：1810.04805，2018.
[20] VASWANI A，SHAZEER N，PARMAR N，et al.Attention is all you need[C]//Advances in Neural Information Processing Systems，2017：5998-6008.
[21] HINTON G E，SALAKHUTDINOV R R.Reducing the dimensionality of data with neural networks[J].Science，2006，313（5786）：504-507.
[22] 李恒辉，郭交，韩文霆，等.栈式稀疏自编码网络的多时相全极化SAR散射特征降维[J].遥感学报，2020，24（11）：1379-1391.
LI H H，GUO J，HAN W T，et al.Scattering feature dimension reduction of multitemporal fully PolSAR image based on stacked sparse AutoEncoder[J].Journal of Remote Sensing，2020，24（11）：1379-1391.
[23] 李贺，冯天瑾，丁香乾，等.企业财务预警PCA-SVM模型研究[J].计算机工程，2006，32（9）：233-235.
LI H，FENG T J，DING X Q，et al.PCA-SVM models for early warning of enterprises through finance data[J].Computer Engineering，2006，32（9）：233-235.
[24] 肖毅，熊凯伦，张希.基于TEI@I方法论的企业财务风险预警模型研究[J].管理评论，2020，32（7）：226-235.
XIAO Y，XIONG K L，ZHANG X.Enterprise financial risk early warning model based on TEI@I methodology[J].Management Review，2020，32（7）：226-235.
[25] FAN J Q，XUE L R，ZHOU Y.How much can machines learn finance from chinese text data?[J].Social Science Electronic Publishing，2021.DOI：10.2139/ssrn.3765862.