西北大学学报：哲学社会科学版
西北大學學報：哲學社會科學版
서북대학학보：철학사회과학판
Journal of Northwest University(Philosophy and Social Sciences Edition)
2009年
5期
79~83
,共null页
上市公司 财务危机 预警模型 Logistic回归分析
上市公司 財務危機 預警模型 Logistic迴歸分析
상시공사 재무위궤 예경모형 Logistic회귀분석
listed companies; financial distress; early warning model; Logistic regress analysis
通过采用样本分析等方法对中国证券市场上市公司的财务危机预警进行实证研究，首先对国内外财务预警的发展历程和研究成果进行了评述，对上市公司财务预警的相关理论及方法作了总结和分析。然后结合中国上市公司的现状和特点，设计了样本A，对样本分别从行业和规模方面进行了分析。选取了37个财务指标，并按照财务指标的特性分为6个组，根据t检验的P值大小在37个财务指标中筛选出若干指标，作为预测模型的初始输入变量。运用Logistic回归分析作为主要建模方法，数据收集的时间延至样本公司可能出现财务危机的前五年或财务恶化的前三年。在此基础上，通过实证研究检验出若干预警能力强的财务指标，建立了一套动态财务危机预警模型。最后，通过有效性检验，得出了2007年上市公司有54家出现财务危机，模型的总判别率为80．30％，并在此基础上指出了模型的实际运用情况。
通過採用樣本分析等方法對中國證券市場上市公司的財務危機預警進行實證研究，首先對國內外財務預警的髮展歷程和研究成果進行瞭評述，對上市公司財務預警的相關理論及方法作瞭總結和分析。然後結閤中國上市公司的現狀和特點，設計瞭樣本A，對樣本分彆從行業和規模方麵進行瞭分析。選取瞭37箇財務指標，併按照財務指標的特性分為6箇組，根據t檢驗的P值大小在37箇財務指標中篩選齣若榦指標，作為預測模型的初始輸入變量。運用Logistic迴歸分析作為主要建模方法，數據收集的時間延至樣本公司可能齣現財務危機的前五年或財務噁化的前三年。在此基礎上，通過實證研究檢驗齣若榦預警能力彊的財務指標，建立瞭一套動態財務危機預警模型。最後，通過有效性檢驗，得齣瞭2007年上市公司有54傢齣現財務危機，模型的總判彆率為80．30％，併在此基礎上指齣瞭模型的實際運用情況。
통과채용양본분석등방법대중국증권시장상시공사적재무위궤예경진행실증연구，수선대국내외재무예경적발전역정화연구성과진행료평술，대상시공사재무예경적상관이론급방법작료총결화분석。연후결합중국상시공사적현상화특점，설계료양본A，대양본분별종행업화규모방면진행료분석。선취료37개재무지표，병안조재무지표적특성분위6개조，근거t검험적P치대소재37개재무지표중사선출약간지표，작위예측모형적초시수입변량。운용Logistic회귀분석작위주요건모방법，수거수집적시간연지양본공사가능출현재무위궤적전오년혹재무악화적전삼년。재차기출상，통과실증연구검험출약간예경능력강적재무지표，건립료일투동태재무위궤예경모형。최후，통과유효성검험，득출료2007년상시공사유54가출현재무위궤，모형적총판별솔위80．30％，병재차기출상지출료모형적실제운용정황。
Through the use of a sample analysis of early Warn!ng Empirical Study of China＇s stock market listed companies to carry out the financial distress study, the appraisals of the financial forecast researches of domestic and foreign scholars based on the related theories methods of listed companiesfinancial forecast and their character- istics are given firstly. Then a sample A was anclyzed, which is designed in advance from two aspects of their indus- try and scale respectively. 37 financial indexes were selected and divided into 6 groups in terms of the characteristics of financial indexes and some indexes were screened out from 37 indexes according to the magnitude of p value from t - test as a set of initial input variables for forecast model. Logistic Recursive Analysis was used as the mainly model - built means. The time of data - collection extends to previous five years when the financial crisis occurred for the sample company, or even previous three years, when finance affairs begin to get worse. On the bases, some financial indexes were checked up with strongly forewarning capacity and a forewarning system was set up for dynam ic financial crisis of listed companies. Finally, through effective checkout, it was concluded that there are 54 listed companies in 2007 which lost money. The total different rate of the model is 80.30% and the practical exertion situation of the model is pointed out on the bases.
