We propose a financial statement (FS) fraud detection framework, called PeerMeta, that makes improvements in all three components of the detection procedure: label measurement, feature set, and detection model. For the label measurement, prior studies mainly adopt FS fraud events that have already been disclosed and confirmed.
Jianqing Fan, Qingfu Liu, Kaixin Zheng
The results
We construct a new measure based on news coverage that can reflect unrevealed FS fraud behaviors as well. For the feature set, we innovatively add peer factors learned through the business description texts in financial reports. For the detection model, two meta-learning algorithms are applied to aggregate the 18 popular classifiers. The results indicate that the proposed method has amazingly high recall, reaching a staggering value of 0.98. We document that all components in PeerMeta contribute to the improvements of FS fraud detections and also showcase the significant economic value of the detection framework and find that recall is more crucial for the economic value than precision.
FinEML seminar series
The FinEML seminar series is designed to create a collaborative platform for the exchange of insights and findings within the field. We aim to foster a friendly atmosphere that encourages constructive feedback, providing an opportunity for both junior and senior researchers to share their work.
Submit research paper
Submit original research papers in the following topics, but not limited to:
- Asset Pricing
- Big Data
- Forecasting with Machine Learning
- Macro Finance
- Option Pricing
