| Summary: | 碩士 === 國立彰化師範大學 === 會計學系 === 104 === The biotech industry has become famous recently and is viewed as the second electronic industry. It’s much similar to electronic industry but sort of different. As biotech industry has long life cycle, intensive input and risky research process. According to these features, biotech industry is not an ideal investment to investors.
Our purpose is to construct a useful direction for investors by using data mining technique, which includes decision tree C5.0 and Apriori approaches to find out the implied relationship between earnings quality and corporate governance. We collected 376 samples from 2011 to 2015 to find out the potential knowledge. Also, we use Kothari’s (2005) modified-Jones model to estimate discretionary accruals as our earnings quality proxy, and we classify earnings quality into bad and normal. The bad earnings quality includes managing earnings downwards and upwards. Similarly, the normal earnings quality is defined as slightly managed earnings above or close to zero. The selections of corporate governance variables include two aspects, the board and ownership structure.
This thesis focuses on the relation between different earnings quality situations and corporate governance variables. The empirical results from decision tree show that (1) the number of block holders, directors and supervisor holdings and the seats of the independent director increases, company tend to manage earnings downwards. (2) In terms of the overall earnings quality, the holding structure is an important node to identify bad or normal. For instance, the ratio holdings between block holder and institutional shareholders. The association rules show that (1) the ratio between managers serving as board and the growth opportunity has relation with managing earnings downwards, (2) the signs of variables that are related to managed earnings that are close to zero include the seats of the independent director and the board size.
|
|---|
