An Analysis Of Credit Risk On China's Manufacturing Industry

Manufacturing industry has always been the pillar industry of the national economy. If manufacturing enterprises develop healthily and orderly, the economic would develop quickly. In order to promote further development of manufacturing industry, the state council announced the "made in China 2025" strategic planning in May 2015. Striving to through "three steps" strategy, the nation would like to achieve the goal of China's manufacturing powerhouse. However, the current manufacturing industry still faces many problems:the world economic growth is slow, and external demand has been reducing. China's current economic increases slowly, and manufacturing PMI in the past two years has been hovering near the line of vicissitudes. On the other hand, internal demand is insufficient, and most of China's manufacturing enterprises are lack of core competitiveness. These all led to a sharp economic contraction, and manufacturing enterprises has become worse, credit risk has also been gradually exposed and accumulation. So the credit risk of China's manufacturing industry has become a hot topic widely concerned by the scholars.Combing the literature, we can find that domestic and foreign scholars mainly focus on the measures of credit risk. The most popular method of traditional method for measuring credit risk is Z score method, which is based on the financial index (Kuma and Rao.2015). Then a variety of measures are developed, such as the KMV model based on option pricing theory (Tomas, etc. al,2015), the CreditMetrics model based on value at risk (Kollar, etc. al,2015), etc. However, Z score model has some subjective qualitative analysis; KMV model is based on the capital market data, if the price is too much under the influence of speculative factors, the accuracy of the empirical results will be affected; CreditMetrics model assumes that the credit rating transition probability is not affected by industry, time, or the economic cycle, so it is stable. But this assumption has certain deviation with the fact. Therefore, it is necessary to find out an approach for the analysis of credit risk of manufacturing.This paper mainly uses DEA model and Logit model to analyze credit risk of manufacturing. The indicator data are based on manufacturing enterprise's financial data. Financial data are widely used in practice because they are simple, flexible, easy to operate and widely applicable. DEA method can be used to measure the enterprise relative credit scores and rankings, and Logit model can be used to measure the probability of default of the enterprise. Two models are beneficial to form a complete theoretical framework and reduce the risk of a single model prediction error. This method would also provide a reference for analysis of credit risk of china's manufacturing.This paper mainly uses normative analysis and empirical research method. The paper selects twenty-eight financial index data of ninety manufacturing enterprises form 2012 to 2014, then process these dataaccording to the weighted average of the time method. The results as the initial index data evaluate credit risk of manufacturing enterprises. This paper uses Excel software, SPSS 19.0 and DEAFrontier software to analyze data, and uses a variety of methods, such as principal component analysis, the data envelopment analysis, KS test, Logistic regression method and ROC analysis, to evaluate credit risk. At the same time, this paper will use a large number of charts to show statistical results, which would reflect the research object more profoundly.Based on the above research methods, this paper is organized in five chapters: The first chapter is an introduction, which describes not only the background and significance of the topic, but also includes the framework, research methods and innovation. The second chapter is literature review. This part mainly review related research literature from three aspects of credit risk, including the reasons, characteristics and the measurement methods. This chapter summarizes the traditional measurement model, modern measurement model and other credit risk measurement method. Through analysis of the existing literature at home and abroad, this chapter summarizes the advantages and disadvantages of each measurement method, in order to select more suitable method for evaluating manufacturing model of credit risk in our country.The third chapter is the research design. This chapter aims to put forward the analysis of manufacturing enterprise's credit risk measurement scheme, and introduce the empirical model and main method. This part first introduces sample index selection principle, data sources and processing methods, secondly introduces the DEA method and the reason why the method can be used to assess relative credit scores. This part also introduces the Logit model and why the model is used to assess expected default rates. Finally, do a summary of this chapter.The fourth chapter is the empirical analysis. First, according to the trend of the initial indicators, the input and output indicators are determined. Then, the output indicator data are enter into principal component analysis to simplify the index number, on the basis of this result, DEA model is constructed to calculate the relative credit rating score and relative credit ranking. Secondly, by using KS test, nonparametric test and independent samples T test find the data which can significant distinguish ST enterprise and Non-ST enterprise. Then, put this data into principal component analysis and establish a logit model; at the same time, introduce ROC method to determine the best default probability, then use this value as the critical point of the model, and test model validity. Thirdly, put the DEA model results and Logit model results together to show the research of credit risks of the manufacturing industry. Finally, do a summary of this chapter. Chapter five is about conclusion and outlook. According to the empirical findings, conclusions and relevant policy recommendations are made, as well as shortcomings and things which are needed for further research.With research method mentioned, the findings and conclusions are as follows: Firstly, the DEA method and Logit model can make full use of all indicators. Capital structure index, profitability index, growth ability, operating ability and share index are used in the two models, the cash flow indicator is also used to structure DEA model, and Logit model includes solvency indicators. Secondly, according to the results of 90 listed companies, we can find that the number of enterprises which have higher credit rating score (top ten) and higher default rate (greater than 0.23) is one. However, the number of enterprises which have lower credit rating score (last ten) and lower default rate (less than 0.23) is eight. The results show that combining with the Logit model evaluating default probability can make up the disadvantages of DEA method, and reduce credit risk caused by relying on a single model. Thirdly, the ROC method can calculate the best expected default probability of enterprises. Then, use this value as the critical value of the Logit model, after analyzing the result, the chapter finds that the accuracy of using Logit model to predict the result is higher (88.9%), and the percentage of the first category mistake declines significantly (down 20%) than selecting 0.5 as the critical value. This founding is very important to the accuracy of forecasting Logit model.Compared with previous studies, this paper implements the following innovations:First, to evaluate credit risk of manufacturing industry, most of literature estimate the default rate of enterprise, or evaluate credit rating of enterprise alone. However, no one considered to put the two results together. So this paper uses DEA model to evaluate the relative credit rating of listed companies, and uses Logit model to measure the company's expected default rates, which helps creditors to evaluate credit risks of the manufacturing companies. Secondly, most of literatures use one year data to assess credit risk, but those data change over time or by accident, which would cause the error. So this paper proposes using time weighted method to deal with the indicators. Thirdly, the critical value of Logit model is usually set to 0.5. However, this paper introduce the ROC analysis to finding out the best default rate of the model, and find that prediction accuracy of the model is higher when using this value as the critical value of model, and the percentage of the first category mistake has dropped significantly.However, due to my limited abilities and practical reasons, this paper also has two shortcomings, to be further expanded. Firstly, this paper only selects financial data of the manufacturing enterprise to carry on the empirical analysis. However, non-financial data, macroeconomic data and capital market data are not considered. It will have an influence on the empirical results of the model. Secondly, this article selects the manufacturing data of listed companies. But there are many unlisted companies in manufacturing industry, which are not selected. At the same time, due to financial data of listed companies are not necessarily true, which will influence the evaluation of credit risk of manufacturing enterprises.In this paper, the future direction of my research is the following two aspects: Firstly, non-financial data of the unlisted companies and the listed companies, macro economic variables, capital market data are all into the index system by using reasonable method. Secondly, the relative credit rating and default rates can be combined with RAROC method to the loan pricing.