Corporate financial distress prediction with multiperiod annual report data: A fusion deep neural network model

Chongren Wang, Pimei Gong, Jiawang Li and Zhiyi Wang

PLOS ONE, 2025, vol. 20, issue 9, 1-25

Abstract: The occurrence of financial distress in enterprises not only leads to operational difficulties but also may trigger chain reactions such as bankruptcy, debt arrears, layoffs, etc., which in turn have a negative effect on investors, creditors, and the entire economic system. Therefore, accurately and timely predicting the financial distress of enterprises is highly important. To address this, a fusion deep neural network based on multiple annual report text data and financial data (MTF-FDNN) model is proposed for financial distress prediction. This model can simultaneously extract long text features of multiple annual reports and financial indicator features of multiple periods of enterprises. Specifically, the model first constructs a multiperiod financial feature extraction model on the basis of a fully connected neural network. Next, it uses a fine-tuned longformer pretrained model to convert long texts into vector representations. Subsequently, Bi-LSTM and TextCNN are employed to extract semantic features from long texts both globally and locally. Finally, the fused financial features and semantic features of long texts are used to identify the financial distress of listed companies. Additionally, on the basis of the experimental results from the test set, the proposed model demonstrates significant improvements over traditional multiperiod financial indicator-based prediction models, with increases of 4.98% in AUC, 6.54% in accuracy, 10.58% in recall, and 6.48% in the F1 score. It is evident that introducing multiperiod textual features significantly enhances model predictive performance. This model effectively predicts corporate financial distress, thereby assisting business managers, external investors, and other stakeholders in mitigating risk.

Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0333064 (text/html)
https://journals.plos.org/plosone/article/file?id= ... 33064&type=printable (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:plo:pone00:0333064

DOI: 10.1371/journal.pone.0333064

Access Statistics for this article

More articles in PLOS ONE from Public Library of Science
Bibliographic data for series maintained by plosone ().