A novel method for financial distress prediction based on sparse neural networks with $$L_{1/2}$$ regularization

Abstract

Corporate financial distress is related to the interests of the enterprise and stakeholders. Therefore, its accurate prediction is of great significance to avoid huge losses from them. Despite significant effort and progress in this field, the existing prediction methods are either limited by the number of input variables or restricted to those financial predictors. To alleviate those issues, both financial variables and non-financial variables are screened out from the existing accounting and finance theory to use as financial distress predictors. In addition, a novel method for financial distress prediction (FDP) based on sparse neural networks is proposed, namely FDP-SNN, in which the weight of the hidden layer is constrained with regularization to achieve the sparsity, so as to select relevant and important predictors, improving the predicted accuracy. It also provides support for the interpretability of the model. The results show that non-financial variables, such as investor protection and governance structure, play a key role in financial distress prediction than those financial ones, especially when the forecast period grows longer. By comparing those classic models proposed by predominant researchers in accounting and finance, the proposed model outperforms in terms of accuracy, precision, and AUC performance.

Explainable Gradient Boosting for Corporate Crisis Forecasting in Italian Businesses

“…Support Vector Machines (SVM) and Random Forests (RF) are widely used due to their robustness in handling non-linear data relationships, with studies such as [19] demonstrating SVM's superiority over traditional logistic regression in predicting bankruptcy a year prior using financial ratios. Random Forest, an ensemble learning method, is particularly noted for its robustness in handling large, high-dimensional datasets, often outperforming other models in predicting financial distress due to its ability to manage missing data and assess the importance of different financial indicators [20,21]. Boosting algorithms like AdaBoost and Gradient Boosting have also been effective, especially in scenarios with imbalanced datasets, by focusing on misclassified instances to improve model accuracy [20].…”

Why do employees commit fraud? Theory, measurement, and validation

“…As shown in column (3), the coefficient of Frauds is significantly positive at 1% level, which shows that the employee fraud motive significantly increases the number of frauds. As shown in column (4), the coefficient of 3 Following Chen et al (2022), the developed provinces we define include Beijing,Shanghai,Guangdong,Jiangsu and Zhejiang. 4 The data of corporate basic information is manually collected from Qichacha (https://www.qcc.com/), one of the largest enterprise credit information inquiry platforms in China.…”

“… 3 Following Chen et al (2022) , the developed provinces we define include Beijing, Shanghai, Guangdong, Jiangsu and Zhejiang. …”

NeVer2: Learning and Verification of Neural Networks

“…Deep Neural Networks (DNNs) and Machine Learning (ML) models in general are gaining momentum in several applications, thanks to their astounding performances and ability to deal with different domains. There are many success stories involving, e.g., financial analysis (Chen et al, 2022, Sako et al, 2022, stock price prediction (Yang et al, 2023), analysis and control of industrial processes (Guidotti et al, 2023a,b, Jahanbakhti et al, 2023, Pavithra et al, 2023, Zhang et al, 2023 or medical analysis (Dash et al, 2023, Gul et al, 2023 and, in recent times, interpretation of MRI images in the context of COVID-19 pandemics (Al-Waisy et al, 2023, Dansana et al, 2023. The application of DNNs may involve safety-or security-sensitive applications (Demarchi et al, 2022, Katz et al, 2017 where the need of formal guarantees on their behavior has great significance.…”