Dissertation > Excellent graduate degree dissertation topics show

The Logistic Regression Model Applied to Credit Risk Measurement

Author: QiaoZhanWen
Tutor: ChenZuoFen
School: Zhejiang Technology and Business University
Course: Statistics
Keywords: Information entropy Logistic regression Credit Risk Factor Analysis
CLC: F832.2
Type: Master's thesis
Year: 2011
Downloads: 334
Quote: 0
Read: Download Dissertation

Abstract

Credit risk is the most important risk facing commercial banks, has a critical influence on the banks continued sound operation. Since the 1990s, financial liberalization and financial innovation to flourish, credit derivatives are widely adopted, which makes the credit risk of commercial banks face more severe, how effective management of credit risk accurately measure based on is a very challenging issues faced by the commercial banks. Can be found through analysis of the status of our credit risk measurement and management of credit risk measurement means is still relatively backward emphasis of qualitative and quantitative methods, the resulting credit risk management foundation is weak, commercial banks are generally facing than credit risk. Combing, the response to this situation in this article on the extensive international credit risk measurement models and analysis of the applicability of the model combined with the actual situation of our country at this stage. Through the analysis of the scope of application of the various model, data requirements, forecast accuracy and stability Logistic regression model was found in the input data and assumptions are more suitable for the actual situation of our country at this stage therefore the final selection logistic regression method to create a credit risk measurement model, and model the effect of an empirical test. Logistic regression method to establish credit risk metric, due to the lack of theoretical guidance of experts and scholars based on their subjective judgment to select the model input indicators, often resulting in the index selection indicators redundant or omission of some important indicators for this In this paper, the use of information entropy principle a problem and an objective choice of the Logistic model input parameters. In the empirical part of all the listed companies in China as research subjects abandoned the usual pairing method in the previous study, based on information entropy theory of financial indicators of credit risk prediction ability objective determination, and to select the most predictive ability The 11 indicators as input variables to build an empirical test based Logistic regression of credit risk measurement models and model results. The research results show that the net assets per share, operating margin of 11 absolute financial indicators and relative financial indicators of the company, whether the breach has strong predictive ability; credit risk prediction model based on Logistic regression established stable and higher forecast accuracy, overall accuracy rate of 90.7% and 86.4%, and therefore can use the model to credit risk assessment and prediction. In addition to test this article using the method established the relative credit risk measurement models created with traditional methods of credit risk measurement models is improved, this paper the stepwise discriminant Act Fisher discriminant model, and two ways to establish credit risk metric conducted a comprehensive contrast to the predictions of the model from the overall prediction correct rate is still guilty of two models can be found by comparing the probability of two types of errors, use this method to establish the credit risk measurement model are comprehensive is better than using the traditional the Fisher multiple discriminant established model.

Related Dissertations

CLC: > Economic > Fiscal, monetary > Finance, banking > China's financial,banking > Banking system and business
© 2012 www.DissertationTopic.Net Mobile