The Study of Building a Financial Crisis Warning Model

• Main Authors: Shu-Yu Lin, 林書羽
• Other Authors: Ding-Tsair Chang
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/82551409057132542995

碩士 === 中華大學 === 工業管理學系碩士班 === 102 === Financial distress warning model is extensively applying on research such as in the area of electronic industry, food industry or construction industry, either in Taiwan or overseas. In the early period, most researchers conducted univariate analysis (Beaver, 1966), multivariate discriminant analysis (Altman, 1968), linear probability model (Ohlson, 1980) on their researches which performed effective prediction. This study is based on listed, OTC (over the counter) and emerging companies in the above mentioned industries which are encountering financial crisis. Furthermore, we will review both the financial normal and financial crisis companies over the period 2008 - 2013. We will build up a warning model which will be tested its accuracy by using Mahalanobis-Taguchi System (MTS) and Logistic Regression Model. Mahalanobis-Taguchi is developed by Genichi Taguchi. This method is in combination of quality engineering and Mahalanobis Distance (MD). The concept is to use multivariate scale to diagnosis and forecast the object on the purpose to support decision making process. Moreover, Logistic Regression Model is brought up by Joseph Berkson. This model can be used on regression model which has qualitative variable as dependent variable. It is not limited by the assumption of normal distribution in plus can analyze discrete, continuous, qualitative or quantitative data. As a consequence, it can calculate the probability of occurrence. In this study, MTS and Logit regression model will be used in financial ratios cut so that we can construct a variable model to select important characters in order to forecast and test. Thereby, less important financial indicators will be reduced as well as time and cost can be saved. We would like to develop a measuring scale showing accurate prediction in multidimensional system.