The Determinants Of PD Of Agriculture-Related Business Loans In Commercial Banks

With the increasing of mercerization and monetization, the development of agricultural industry, growth of rural economic and income growth of farmers can not do without financial support currently. The analysis of agriculture-related asset quality, operational efficiency and its influencing factors of rural financial institutions has become the focus of great concern of officials and scholars.(1) The thesis consists of four parts:1) Make econometric models based on the economic theory combined with relevant characteristics of commercial banks' agriculture-related loans and analyze the effects of explanatory variables using cross-sectional data and panel data.2) According to the preliminary econometric model, select factors can be observed from the database as independent variable. They are loan guarantee ways, loan amount, loan term, interest rate, repayment methods, enterprise scale, the economic nature and characteristics of enterprise management, respectively. Secondly, describe relevant statistics data and make calculation of simple correlation coefficient matrix.3) Do empirical analysis. First, make regression analysis to section data and panel data and compare the unity and difference in regression results of method of linear probability model (LPM) and Logit model, taking robust estimation, cluster estimates and fixed effects estimation with Stata. Secondly, make autocorrelation, heteroscedasticity test and stability inspection. Meanwhile, do robustness test of LPM and Logit model using defaultl and default2 which are defined according to different criteria of loan defaults.4) Conclusions:①Relative to short-term loans, medium-term loans have a significant positive effect to probability of default (PD) and long-term loans should be also positively correlated with PD according to economic theory.②The change of total interest rate will make a positive effect on PD of loans.③Relative to credit loans, guaranteed loans has significant positive effects on PD, or has no significant difference in PD with the former; Mortgage loans were highly correlated with PD no matter in Logit model or the linear model; Secured loans have no significant difference with credit loans about PD. Additional, discounted bills are related to PD of loans in inversely proportional manner.④Relative to small businesses, special large enterprises have a significant negative effect on PD of loans and large enterprise has a significant positive effect on it. There is no significant statistical difference about PD of loans between small and medium enterprises.⑤Relative to the phase-out of customers, good customers, general customers and restricted clients have a low PD of loans.⑥Relative to other enterprises, state-owned enterprises, cooperative enterprises, joint ventures, limited liability companies, joint stock, private enterprises, foreign-funded enterprises have significant negative effects on the loan delinquencies, but there is a significant positive effect for collective enterprises.⑦Compared with the baseline, annual dummy variables and regional dummy variables are statistically significant.(2) The main contributions of thesis:1) It is innovative to make empirical research in selecting agriculture-related loan data of commercial bank in China.2) The thesis combines correlation analysis with econometric analysis using the respective section data and panel data.(3) Difficulties of Study:1) This study involves mixing section and panel data in data analysis, using a mixed OLS, cluster estimation, Logit models and methods such as fixed effects regression, which results in more complex correction and test due to the large amount of data;2) The procedure which changes unbalanced panel data into the required panel data drops some sample data, leaving certain default and loss of data which may affect forecast accuracy.