The Characteristics And Prediction Model On Recovery Rate For Non-performing Assets By Business Acquisition

Non-performing assets are very common in domestic banking sector, as non-performing assets play important roles in our financial security system; it has been attracting wide concern from both home and abroad. In order to increase the recovery rate of non-performing assets, speed up the disposition process and reduce loss by non-performing assets, four financial Assets Management Companies- established in 1999 by our government, including Cinda, Orient, Great Wall and Huarong-were assigned the special task of recovering non-performing assets from the four major state-owned commercial banks.According to the difference of transfer mechanism, the non-performing assets can be classified into two groups: policy-off and business acquisition. The disposition of non-performing assets by policy-off is closely related to the assets management under the planned economy system for China's banking, which means the end of one disposition. Business acquisition is the main disposition form at this stage in commercial bank for non-performing assets, so the disposition for this part is closer to the risk management objectives for our commercial bank. Based on the background, this study is focused on the analysis of characteristics on the recovery rate for non-performing assets by business acquisition and tries to establish the forecast model for recovery rate.This article begins with the definition of recovery rate; accurate definition is the foundation and premise for the study of recovery rate characteristics. And then, based on the previous studies, the international research and industry practice on recovery rate for non-performing assets were summarized and concluded. Based on the largest default recovery rate database of China-LossMetricsTM, considering the actual situation in China, the author gives a characterization and description on recovery rate for non-performing rates at different levels. This includes not only the non-performing assets qualification characteristics, such as collaterals, five-category classification, but also the debtor's own information, such as industry, region and business conditions and so on. Meanwhile, the differences of the distribution and influencing factor for recovery rate for business acquisition and police-off were analyzed by comparison. In particular, this paper analysis the factor of industry in separately, Using Generalized Beta Regression approach, we estimate the distribution of recovery rate conditional on different business conditions, areas, collaterals and disposal manners across various industries. Thus, the whole picture of recovery rates in different industries is presented. Based on the results, the influence pattern of the risk drivers on recovery rate in different industries is captured, and the relevant advices for disposal process could be provided. The results show that, the performance of generalized beta regression model is better than traditional beta model in different industry sector.Finally, by choosing the traditional linear model and the Beta-normal transform model, this paper attempts to develop a prediction model for recovery rate (1-LGD) of the non-performing assets. In order to test the reliability of the models, we show multi-angle validation approaches such as multicollinearity, accuracy, distinction capacity test and stability of the model. The empirical results shows that the above two types of models have better predictive and high robustness, and the prediction results for acquisition of commercial samples is better than the business acquisition of full sample.