Operational Efficiency And Total Factor Productivity Of Commercial Banks

As the most important financial institution in China’s financial system,commercial banks play a leading role in its financial system,and play an important role in realizing the stable and healthy development of the entire financial system and promoting the sustainable growth of the national economy.In the new era,especially since the global pandemic of the new coronavirus,the banking industry has faced new regulations on liquidity,new regulations on asset management,and a series of documents.The series of challenges may be greater than the development opportunities.The reason is that the challenges are the certainties that exist in reality,and the opportunities are to win opportunities for increments,which are uncertainties that cannot be grasped.Under the current background of the times and the economic situation at home and abroad,commercial banks that aim to serve the development of the real economy and achieve sustainable and healthy growth under the constraints of tighter capital and stricter risks have become the core goal of banking business management.In order to achieve this goal,the essence is to gradually improve the operating efficiency and total factor productivity of commercial banks.Therefore,it is of great theoretical and practical significance for promoting the sustainable and healthy development of China’s banking industry to reasonably measure the operating efficiency and total factor productivity of domestic and foreign commercial banks in Chinese mainland,and to explore the internal causes of operational inefficiency and total factor productivity decline.In the existing literature,the Data Envelopment Analysis(DEA)method is the most widely used non-parametric efficiency analysis method by many experts and scholars in various fields.Therefore,this paper also adopts the extension method based on the DEA method to measure the operating efficiency of commercial banks and total factor productivity(TFP).In order to measure the operational efficiency of commercial banks,this paper constructs a new Se Uo-SBM-DEA model(Super-efficient Undesirable-outputs Slacks-based Measure DEA)and DNUo-SBM-DEA model(Dynamic Network Undesirable-outputs SBM-DEA).Due to the rapid development of the DEA model,the current model structure is becoming more and more complex.At the same time,in the era of big data,the number of samples is becoming larger and larger and the data dimension is getting higher and higher.Therefore,the computing resources required to solve the DEA model are also increasing.Larger,this paper intends to introduce machine learning methods and their ensemble models to solve this problem.In addition,in order to measure the total factor productivity of commercial banks,this paper constructs a new TFP index(Allocative Malmquist-Luenberger Productivity Index,AMLPI).The main innovations of the model constructed in this paper are as follows:(1)A new Se Uo-SBM-DEA model is extended by adding new constraints to reconstruct the reference set.This model is a universally applicable method for evaluating the comprehensive efficiency of decision-making units(DMUs),which can not only consider the common effects of expected and undesired outputs at the same time,but also achieve a complete ordering of all DMUs.In addition,by constructing the complex dynamic network structure of DMU production and operation activities and the corresponding DNUo-SBM-DEA model,its comprehensive efficiency is internally decomposed.(2)Due to the inherent defects of the traditional DEA model,that is,the production possibility set may be changed every time a new decision unit is added,so it needs to be re-modeled and solved,resulting in a lot of waste of computing resources.By building a bridge between various machine learning methods such as BP neural network,support vector machine and its improved support vector machine,genetic neural network,and their integrated models and DEA models,namely DEA-ML(A Combined Machine Learning and DEA Method)two-stage method is used to solve the problem of computational loss of the traditional DEA model due to the addition of new decision-making units.(3)By introducing a directional distance function,a new AMLPI index is expanded and decomposed into allocation efficiency change with undesirableoutputs(AECU)and allocation-technical change with undesirable-outputs(ATCU)two parts.The index can take into account the combined effects of both desired and undesired outputs when measuring productivity,and its decomposition part provides a more intuitive explanation for revealing the root causes of productivity changes.Based on the model constructed in this paper,the operating efficiency and internal decomposition of various commercial banks in Chinese mainland during the“13th Five-Year Plan”(2016-2020),as well as the AMLPI and its decomposition,are measured and calculated,and the two stage DEA-ML is applied.The method measures and forecasts the operating efficiency of commercial banks.The research results show that during the research period:(1)Although the overall operating efficiency of various commercial banks in my country is low(the average value is 0.473),the operating efficiency of commercial banks has maintained a continuous growth trend as a whole,with an average annual increase of about 5.27 percentage points throughout the period.Secondly,the operating efficiency of various commercial banks in China is ranked as: private banks,joint-stock banks,foreign-funded banks,city commercial banks,state-owned banks,and rural commercial banks.In addition,the operating efficiency of various types of commercial banks in my country in different years presents obvious non-equilibrium characteristics,but this non-equilibrium has a tendency to gradually shrink.(2)The AMLPI of Chinese commercial banks decreased slightly on the whole(0.998),with a decrease rate of 0.2%;From the measurement results of AECU and ATCU,the main reason for the slight decrease in the TFP during the study period is the decline of ATCU;The AMLPI of Chinese commercial banks has increased in order of joint-stock commercial banks(1.094),rural commercial banks(1.055),large state-owned commercial banks(1.017),while private banks,city commercial banks and foreign-funded commercial banks have experienced different degrees of decline.(3)During the study period,the average capital flow efficiency of various commercial banks in my country was 0.761,and the average asset profitability efficiency was 0.256.It can be seen that the main internal reason for the inefficiency of various commercial banks in China during the “13th Five-Year Plan”period is the insufficient efficiency of asset profitability.(4)Based on the sample data and characteristic variables selected in this paper,the prediction accuracy,correlation coefficient R,mean square error MSE and Spearman’s Rho of each model are comprehensively analyzed.It is found that the comprehensive performance of the four ML-DEA algorithms can be ranked as follows : GANN-DEA > BPNN-DEA ≈ ISVM-DEA > SVM-DEA.