Evaluation Of Earnings Qualiy For Commercial Banks Based On DEA And SVM

Earnings quality is one of the most important indexes of evaluating company achievement, and also the key index of enhancing enterprises'competitive power and keeping their sustainable development. As"highe debts,high risks"enterprises for operating currency, higher earnings quality is the basic of survival and development for commercial banks. Only the good earnings quality, can the commercial banks have strong risk bearing capacity. Therefore, it has important realistic significance to evaluate the earnings quality of commercial banks. However, now the studies to earnings quality from inside and outside the country focus on qualitative analysis, quantitative analysis is still in an exploratory stage, and the mian quantitative analysis are traditional statistical methods and structural equation. Due to the restrictions assumption conditions, earnings quality from these two methods is not true. Overall, it is lack of a perfect method, and can not evaluate fully, accurately, deeply, so the evaluating result has a little reference value.Based on the above-mentioned situations, this paper takes the consulting a large number of domestic and foreign documents concerning this subject as the basis, from the starting point of commercial banks'concrete conditions, considers the DEA does not require a priori knowledge onfunctional form and parameter and good prediction ability of SVM, first designs the train of thought of DEA-SVM evaluating model, and elaborates the operation procedure of this model. Second, based on data of 13 commercial banks from 2002～2008, uses DEA-SVM to evaluate earning quality of 13 commercial banks. With technical efficiency of each commercial bank from DEA calculating, finds that 13 commercial banks are in the state of technical inefficiency, and technical efficiency of State-owned Commercial Banks are higher than Joint-stock Commercial Banks'. Apply the principal component analysis to get Weights, and according to these weights structural the earnings quality index.Uses SVM to regress the earning qualities of 13 commercial banks, and empirical test shows that the DEA-SVM model is efficient. On this base, calculates the prediction value of each commercial bank in 2009. At last, according to the evaluation result, proposes several suggestions to enhance the earning quality of commercial banks.