MuST: An Interpretable Multidimensional Strain Theory Model for Corporate Misreporting Prediction
Document Type
Article
Publication Date
2023
Journal / Book Title
Electronic Commerce Research and Applications
Abstract
Corporate misreporting has become a serious problem in both e-commerce and offline economies. Developing models to predict corporate misreporting involves challenges related to (1) the lack of interpretability in the past literature, (2) corporate fraud prevention, (3) the heterogeneity in both the corporate ability and complexity of regulatory requirements, and (4) the sparsity of misreporting observations. This study proposes a novel multidimensional strain theory (MuST) model to address these challenges. Inspired by strain theory, we quantify the incentive to misreport as the gap between corporate abilities and the complexity of regulatory requirements. Experiments show that our model achieves high accuracy in identifying various types of corporate misreporting indicated by comment letters from regulators. Furthermore, the interpretability and application analyses show that latent corporate abilities learned from the MuST model effectively reflect inherent corporate risks and can further be applied to the classified supervision of e-commerce companies. Our paper contributes to the emerging literature on misreporting prediction and provides a precise and interpretable prediction model for e-commerce supervision.
DOI
https://doi.org/10.1016/j.elerap.2022.101225
Book Publisher
Elsevier
MSU Digital Commons Citation
Liu, Chunli; Yang, Liu; Gao, Weibo; Li, Yang; and Liu, Yezheng, "MuST: An Interpretable Multidimensional Strain Theory Model for Corporate Misreporting Prediction" (2023). Department of Information Management and Business Analytics Faculty Scholarship and Creative Works. 157.
https://digitalcommons.montclair.edu/infomgmt-busanalytics-facpubs/157