| dc.description.abstract | In the past, most of the research was to detect risks with financial information, and less to explore with non-financial information. At present, there is a few relevant research in the United States, and there is no relevant research in Taiwan, so this paper intends to explore the influence of different financial status companies on the characteristics of shareholder reports. After collecting 9 companies with fraud and 7 companies with crisis, and matching them with the same size and leading companies according to their industry search, the final sample number is a total of 53 companies. Take t-2 to t+2 for a total of five years for a comparison. The M-Score proposed by the Beneish (1999) and the Z-Score model proposed by Altman (1968) are used to assist in the detection of fraud and crisis companies, and observe whether the content of the shareholders’ report has changed when the company is facing different financial conditions.
After training the shareholder report of the leading company into a module, the five consecutive years of the shareholder report of other companies are put into the module, the cosine and European distance are quantified, and is drawn as a scatter map for observation. The results show two extreme cases that one is the change of cosine of fraudulent companies and crisis companies is larger than that of leading companies and the other is companies of the same size, the other is the cosine and European are almost unchanged, that is, the content is almost copied and pasted. Unexpectedly , it is also found that leading company’s European distance changes more , compared with other companies with different financial status. | en_US |