Financial reporting is a mix of mandated and voluntary disclosure and a mix of quantitative and textual disclosure. Since financial reporting is designed for outside users, the quality of financial reporting is very important. Based on the concept of agency problem and information asymmetry and using the computer-intensive techniques and Taiwan sample in this research, we build up a comprehensive framework to investigating two research issues regarding financial reporting texts in two years. Issue in year 1: to investigate the attributes of the report to the shareholders and investigate the relationship between these textual attributes and earnings quality in annual report.Issue in year 2: to investigate whether texts in financial reporting have incremental information for firm risk prediction. Contributions in issue for year 1:(1)Different from the literature in English language investigation, we are the first to consider the characteristics of Chinese language and the regulation for the report to the shareholders in order to comprehensively explore the attributes of the report to the shareholders and the difference between the textual characteristics in Chinese and English.(2)The characteristics and grammar of Chinese language is totally different from English. We are the first to use Chinese text mining technology to investigate our research issues and hope to provide a new methodology and share research results in understanding the implication of Chinese textual narratives.(3)We consider the concept of information asymmetry and the motivation of firms when they disclose their information. In most circumstances, the information that firms would like to hide or distort is bad news. Therefore, we would like to investigate what the attributes of the textual narrative will be among different earnings quality levels. And how does the firm hide their information in textual narrative if it has incentives to hide the information, vice versa.Contributions in issue for year 2:(1)Since we expect the information that firms would like to hide or distort is bad news, we then consider the possible consequence of these firms may result in firm failure in the near future. This project is the first study to use financial reporting text for firm risk prediction. (2)We consider that reasons for a firm failure are various, so we use the machine learnings named Random forest. Random forests can use for classification and regression problems, can handle many features and can help estimate the importance of modeling data variables. Our research uses different models in various aspects to build a random forest model for the firm risk prediction in order to increase the prediction power of firm risk.
|Effective start/end date||1/08/20 → 31/12/21|
UN Sustainable Development Goals
In 2015, UN member states agreed to 17 global Sustainable Development Goals (SDGs) to end poverty, protect the planet and ensure prosperity for all. This project contributes towards the following SDG(s):
- Information asymmetry
- the report to the shareholders
- textual attributes
- text mining
- earnings quality
- machine learning
- firm failure prediction models
- random forests
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.