An Empirical Study Of Financial Crisis Early-warning On Manufacturing Listed Companies

On the basis of the review of previous empirical results, this paper holds that financialdistress research mainly includes definition of financial distress concept,selection of financialdistress prediction variables, and establishment of financial distress prediction model. Andfinancial distress concept is the foundation,while the prediction variables selection is thekey,and prediction model establishment is the focal point. When defined the financial distressconcept, this paper used the company which was special treatment (ST) because of the specialfinancial situation as the financial distressed company.In the end,this paper chose 145 chineselisted manufacturing companies as the research objects. This paper put forward the principleof index selection, and chose the companies financial index covered by solvency index,profitability index, etc , five aspects of financial indicators, and covered by ownershipstructure index,equity balance index, etc, six aspects of non-financial indicators,37 indictorswere constructed the financial early-warning indicators,through the significant test and factoranalysis methods, 6 early-warning indicators were screened out finally.In the selection of theearly-warning model, first, the paper used the Logistic regression to build the financialdistress early-warning model, and used the 6 early-warning indicators information which werefrom the two years before training samples taking place financial crises,and constructed theearly-warning model.Then through the analysis of the general regression neural networkcharacteristics ,the paper thought it had feasible as the financial distress early-warningmethods, the paper selected the general regression neural network as the early-warning modeland used the same training samples information to construct the early-warning model. Finally,we used the training samples and the test samples to test the model's forecast accuracy . Thetest results showed that:The forecast accuracy rates tested by Logistic regression model are89.5 percent and 84.25 percent, the forecast accuracy rates tested by GRNN model are 95.35percent and 93.1 percent.The results showed that: firstly,the 6 early-warning indicators were good at reflect thedifferences between the financial crisis companies and financial normal companies,and the 6early-warning indicators included the effective information whether Manufacturing ListedCompanies happened financial distress.Secondly, the two financial early-warning models hadgood prediction ability,they can provided information for the internal managers and externalinvestors to make decisions.Thirdly,general regression neural network had feasible as thefinancial distress early-warning method, it had a higher discriminant accuracy and applicationvalue. Finally,The enterprise managers, Shareholders ,creditors and other stakeholders shouldpay attention to the audit reports issued by the certified public accountant to the companyfinancial statements,in order to safegard their own interest.