| Summary: | 碩士 === 國立高雄應用科技大學 === 金融資訊研究所 === 98 === In the second half year of 2008, comes under the global financial crisis which has lead to contraction and sluggish consumption investment. Many of the domestic enterprises have face shortage of funds and financing difficulties among which primarily in small and medium size enterprises (SMEs). The SMEs are limited by unsound financial and sales data that procuring banks to increase credit risk assessment which tend to be more conservative in extending loans that has lead to financing difficulties and increasing business risk. This study attempts to find SMEs financial variables, identify the causes of financial crisis in SMEs and to create a more practically applicable prediction model as an early warning system. It is aim at improving SMEs creditability that conduce willingness of bank to meet the financing demands of SMEs.
The study shows that financial variables that affect financial crisis in SMEs are quick ratio, debt ratio and total asset turnover; in the first three years of the financial crisis, account receivable turnover, debt ratio, equity ratio, total asset growth rate, revenue growth rate extract from the factor analysis are also the significant variables that affect SMEs financial crisis.
The accuracy of the two models of this study is between 68%-75%, feasible for predicting SMEs financial crisis. Type-I error in model one are 60%,70%,80% and in model two 45%,47.50%,47.50%. It has highlighted the financial characteristics of SMEs in Taiwan, which is unsound, easy modification of financial statements, beautification of financial date, result in errors of financial statements, increasing the difficulties for bank to audit.
The two models established by logistic regression shown that models built up by primary components to the occurrence of financial crisis in overall prediction accuracy and default enterprise accuracy have outperformed models built up by correlation matrix analysis. The result shows that financial variables of the models composed by primary components analysis have effectively improved the prediction.
|
|---|
