Sunil K. Mohanty. Brooklyn College of the City University of New York. Phone:

Transcription

1 Efficiency in China s Banking Sector: Application of SFA to Pre- and Post- Basel II Eras Sunil K. Mohanty Brooklyn College of the City University of New York Koppelman School of Business Phone: skmohanty@brooklyn.cuny.edu Hong-Jen Lin Brooklyn College of the City University of New York Koppelman School of Business Phone: ext HJLin@brooklyn.cuny.edu Winston T. Lin University at Buffalo, SUNY Phone: mgtfewtl@buffalo.edu Revised Draft: March

2 Efficiency in China s Banking Sector: Application of SFA to Pre- and Post- Basel II Eras Abstract We compare cost and profit efficiencies of Chinese banks during the pre- Basel II era ( ) with that of post- Basel II era ( ). We use a stochastic frontier approach (SFA) with onestep estimation procedure following Wang and Schmidt s (2002) methodology. Our results show that the cost efficiency of the banking industry improved over the period and the same fluctuated a little over the second sub-period ( ). On the other hand, the profit efficiency increased over the , but it continued to decline over the period. During the post- Basel II era ( ), the cost efficiency of state- owned banks outperformed that of regional commercial banks as well as joint- stock banks. Our study also finds that the Tier-1 riskbased capital ratio is positively associated with the cost efficiency. We also document that the profit efficiency of regional commercial banks outperformed those of other two categories of banks (stateowned and joint-stock banks) for the period. JEL Classification: G21 Key Word: Cost efficiency, profit efficiency, banking efficiency, and stochastic frontier 2

3 1. Introduction China, the second largest economy in the world, has seen continuously rapid economic growth in past three decades. The Chinese financial system is dominated by large but inefficient banking sector in absence of well-developed bond and equity markets. Since China initiated comprehensive economic reform in early 1980s, the banking sector has experienced fundamental and structural reforms which were aimed to liberalize, deregulate, and promote efficiency in the Chinese financial system. For example, in 1995, the Commercial Bank Law was first promulgated followed by the formation of the commercial bank system and organizational structure. The Commercial Banking Law provided a legal means for changing specialized state banks to state-owned commercial banks (SOCBs). These reforms helped transform the banking sector from a state-owned monopolistic and government policy-driven one to a market-oriented, multi-ownership, competitive, and profit-oriented banking system. Since the enactment of the Commercial Bank Law in 1995, the organizational structure of the financial system gradually improved. In 1995, capital adequacy requirements (Basel I) were introduced in all commercial banks, as wells as ratios such as the loan to deposit or assets, and assets to liquid liabilities. Prior to joining WTO in 2001, Chinese banking sector, particularly the state-owned banking, had a huge amount of non-performing loans (NPLs), weak capital base, lack of adequate corporate governance practices, and low profitability (e.g., He, 2001; Li and Ma, 2004). Deregulations and further reforms in banking sector in China began to accelerate after China joined the World Trade Organization (WTO) in 2001 (e.g., Yao et al, 2007; and Barth, Li, and Li 2013). In addition, under the rules and agreements set by the WTO, China opened up its financial markets to foreign competition by establishing a level-playing fields for foreign banks to compete with domestic banks by the end of

4 Bank reforms and liberalization measures in China since joining WTO were accompanied by improvements in corporate governance structures, bank regulations and supervision. For example, the China Banking Regulatory Commission (CBRC) was formally established in April Since then, a financial regulatory system was created in which the CBRC, the China Securities Regulatory Commission (CSRC) and the China Insurance Regulatory Commission (CIRC) could work together in coordination with one another. In response to the globalization of financial markets and institutions, and competition from foreign banks, the bank managers operating Chinese banks emulated strategies of their Western counterparts by adopting new technology, implementing advanced risk management systems, diversifying their product mix, and adhering to Basel II and Basel III capital adequacy rules which went into effect in 2007 and 2010 respectively. On March 13, 2007, CBRC formally issued The Guidelines on the Implementation of the New Basel Accord (Basel II) by China s Banking Sector. These Guidelines are formulated with a view to steadily pushing forward the adoption of Basel II in China. Adoption of Basel II was aimed at enhancing the effectiveness of the risk-based capital regulation, thus underpinning the stability of the Chinese banking system (e.g., CBRC, 2007). Indeed, banking regulators of China adopted more stringent regulatory standards than that of Western counterparts to leapfrog over the Western banking industry performance (e.g., Knaak, 2014). The Big Four banks in China were ranked as top 10 banks in the world in 2013 in terms of asset size after meeting risk-based capital ratios and the countercyclical capital buffer requirements under Basel III (e.g., Molyneux, Liu and Jiang, 2014; Chen, Jiang, and Lin, 2014; and Huang and Xiong, 2015). Lessons from the global financial crisis taught us that a well-performing banking system which provides certain essential services to the economy such as credit- 4

5 supply, and liquidity is key to the economic growth and stability. Thus, Basel II and III capital adequacy rules and their implementation for Chinese banks are likely to contribute to financial stability via their effects on higher quality of assets, lower future nonperforming loans, reduced risk-taking behavior of banks, and lower probability of bank failures. Inextricably linked to the policy of banking reforms related to Basel II and III capital rules, market disclosures, and bank supervision is the improvement of the efficiency of the Chinese banking sector over the pre- Basel II era to post- Basel II era. Enhanced banking reforms in light of implementation of Basel II and Basel III aimed at preparing Chinese banking sector to remain internationally competitive and to put them at pace with the transformations observed in the overall global economy. On one hand, one would expect continued improvement in banking efficiency based on the competitive market conditions in light of ongoing banking reforms, but on the other hand, the social-driven agenda of the central government concerning unemployment and regional inequalities might pose challenges for bank managers to achieve improved banking efficiency despite ongoing banking reforms. In light of these reforms and challenges, a number of research questions are of particular interest and importance. How has the bank performance in terms of the cost efficiency as well as of the profit efficiency changed over time? Has the bank performance improved during the post- Basel II era in comparison to that of pre-basel II era? Does risk-based capital ratio affect bank efficiency? This paper addresses these questions and contribute to existing literature related to banking efficiency in China. The main purpose of the study is to investigate the impact of regulatory reforms on banking efficiency in China particularly during the post- Basel II era. There are several reasons why understanding efficiency of Chinese banking sector is important. First, China is the second largest economy and the safety, soundness, and efficiency of banking system 5

6 are critical to global economy. Second, since China s equity and debt markets are not welldeveloped, the banking sector in China plays a critical role in its economy. Chinese banks are increasingly facing competition from their international rivals especially after the deregulation of entry of foreign banks under WTO agreement. Third, banks in China with major state-ownership and minor private or foreign ownership, have remained strong and kept growing during post- Basel II era and after the recent global financial crisis. Thus, after the financial meltdown, the belief in conventional banking is challenged. People started to re-evaluate the cost and benefit of private ownership and the free market. Finally, the findings of the study will be useful for policymakers and managers of Chinese banks to evaluate banks competitive viability. In this paper, we analyze time varying patterns of banking efficiency in China under different types of ownerships (Jiang, Yao and Feng 2013; Molyneux, Liu, and Jiang, 2014) over period. We use the stochastic frontier approach (SFA) to analyze cost and profit inefficiencies of banks in China. First, we estimate whether cost efficiency and profit efficiency of the Chinese banking sector have improved over the period. we estimate the volatility of cost and profit inefficiencies at the industry level. By doing so, we can observe the patterns of the cost and profit inefficiencies of banks in China over period. We examine whether the banking efficiency during post-basel II era ( ) has improved over the pre- Basel II era ( ). We also investigate whether size, type of bank ownership, and Tier1 capital ratio play important roles in determining cost or profit efficiency of a bank and whether Base II and III capital regulations have had an intended impact on Chinese banking efficiency. Our paper contributes to economics and finance literature in several ways. First, Chen, Skully, and Brown (2005) estimate technical, allocative, and cost efficiencies while Berger, Hasan, and Zhou (2009) investigate profit efficiency only. Second, neither of them 6

7 combines production side and profit side in a single paper. Howver, this study does incorporate both of them. Third, the volatility of efficiency is considered. Fourth, our paper estimates the mean and the variance (volatility) of cost and profit inefficiencies. 1 Fifth, this study employs an improved methodology following Battese and Coelli (1995) and Wang and Schmidt (2002). Finally, our empirical models account for exogenous factors and heteroscedasticity in the inefficiency terms. The remainder of this paper is structured as follows. Section 2 presents the related literature. Section 3 describes the empirical methodology used and specifies the models and variables in this study. Section 4 presents data sources, variables and empirical results of the stochastic frontier analyses of cost and profit inefficiencies. Section 5 concludes and provides policy implications for the banking industry in China. 2. Related Literature In early years, most studies on Chinese banking industry are simply descriptive (e.g., He, 2001; and Barth, Koepp, and Zhou, 2004 among others). Recently, more and more studies numerically analyze the banking efficiency in China due to availability of data ( e.g., Chen, Skully, and Brown; 2005; Lin and Lin, 2006; Berger, Hasan, and Zhou, 2009 & 2010; Fu and Hefferman, 2007 & 2008; Yao et al., 2007; and Shen and Lu. 2008). Chen, Skully, and Brown (2005) and Wang et al. (2014) have adopted DEA approach while the other two use the stochastic frontier approaches. Berger, Hasan, and Zhou (2008), Jiang, Yao and Feng (2009), and Lin and Zhang (2009) have analyzed the Chinese banking sector for different types of ownership structure from 1994 to Fu and Hefferman (2008) have shown that the stochastic frontier is an appropriate approach to 1 The results of the measures of the volatility of the efficiencies are omitted from the tables to save spaces. They are available upon requests. 7

8 estimate efficiencies as well as the economy of scale. Foo and Witkowska (2014) calculated the efficiency of Chinese banks and conclude that the government intervention and state ownership lead to low profitability. They have also found that the Chinese banks operated in the range of decreasing or constant economy of scale. Previous literature suggests that government ownership in Chinese banks is negatively related to the profit efficiency of a bank (e.g., Leung and Young, 2003; Lin and Zhang, 2009). Lin and Zhang (2009) find that the Big Four state-owned banks are less profitable and less profit efficient than citylevel commercial banks and domestic joint-equity banks. Kumbhakar and Wang (2007) find that joint-stock banks are more cost efficient than wholly state-owned commercial banks (SOCBs). Similarly, Jiang, Yao and Feng (2013) report that joint stock commercial banks and city commercial banks are more cost efficient than state-owned commercial banks. Jiang, Yao and Zhang (2009) show that joint-stock commercial banks are likely to more profit efficient than state-owned commercial banks. Capital adequacy requirements are one of the main regulatory tools for the banking system. Bank capital serves two useful purposes. First, it acts as a buffer against losses which protects depositors and limit the loss to deposit insurance system. Second, bank capital limits the moral hazard issue of shareholders incentive to take excessive risk. Previous studies related to bank efficiency and capital ratios (Molyneux, Liu and Jiang, 2014 and Fiordelisi Marques-Ibanez and Molyneux, 2011) find that less efficient banks tend to take more risk, and better capitalized banks are likely to be more efficient than lesscapitalized banks. Shen and Lu (2008) also find tight relationship between capital ratios and profitability. Knaack (2014) and IIF (2014) indicate that the large asset size may entail higher efficiency or better performance. Berger (2013) shows that the higher capital levels help improve cost efficiency. Molyneux et al (2014) demonstrate that higher capital ratio is positively related to the profitability and diversification. 8

9 3. Methodologies and Model Specifications 3.1 Methodologies Following Battese and Coelli (1995) and Wang and Schmidt (2002), we use one-step estimation approach to estimate the cost and profit efficiencies of 124 banks in China from 1996 to This model allows the inefficiency term to follow the shift-in-mean processes where the exogenous variables explain the means or the volatility of the cost or profit inefficiency. The model follows the intermediation method where the deposits, investments, and loans are regarded as output variables and labors and capitals are inputs. When banks are treated as financial intermediaries and they follow market mechanism more closely, the intermediation method is more applicable. Equation (3.1) describes the cost frontier of the banks under study by considering three different distributions of cost inefficiencies as follows: i ( y i,p i; ) + vi ui ln TC = f β + (3.1) where ln TC i is the natural logarithm of the ratio of the total cost to the assets; y t is the output vector; 2 p is the input-price vector. β means the parameters to be estimated in the t cost frontier f-function. vi and u i are both random variables: v i is the normally distributed random error and u i is the cost inefficiency term with Gamma distribution (See Greene, 2007). The subscript it stands for the ith bank for the time period t. In equation (1), θ and Ρ are the parameters of the Gamma-distributed inefficiency, u i ; and v is the standard deviation of the normal error. Battese and Coelli (1995) and Wang and Schmidt (2002) extended the truncatednormally-distributed equation (1) to the panel data (i.e., time-series and cross-sectional pooled) and the equation becomes it ( y it,p it ; ) + vit uit ln TC = f β + (3.2) 2 9

10 u it 2 (, ) + ~ N uit (3.3) 2 ( 0, ) v it ~ N (3.4) it zit = (3.5.) 2 uit = exp( z ' ) (3.6) it And the cost inefficiency is explained by some other factors, that is, Z variables. We use Z variables to control the asset management risk, size effect, ownership effect, and the impact from non-performing loans. The other specific parameters of equation (3.2) are λ= u / v ; and σ= u v Equations (3.7) to (3.10) depict the profit frontier. For other equations, we try to use similar notations of variables as we did in equation (3.2) if the natures of the variables are the same. Equation (3.7) is stated as follows. it ( y it,p it; ) + eit mit ln ROA = g β.(3.7) m 2 (, ) + it ~ N it mit 2 ( 0, ). (3.8) e it ~ N..(3.9) 2 mit = exp( z ' ) = '.., (3.10) it it z it Again, y it is the output vector; p it is the input-price vector; and β represents the parameters to be estimated in the profit frontier. The g-function means the optimal profit frontier. eit is the normally distributed random error and mit is the profit inefficiency term. ln ROA it is the natural logarithm the return on assets for bank i at year t. z is the vector of 10

11 factors that explain the profit inefficiency. 3 Equation (3.7) also follows Battese and Coelli 2 (1995) model and Wang (2002). The results of the estimated σ mit are not displayed to save spaces but available upon requests. Using the same specification as in the cost frontier helps us compare the empirical results between two different types of frontiers. 3.2 Model Specifications For both cost and profit frontiers, we take the translog functional form to specify the cost and profit functions. the f-function in Equation (3.11) is denoted as: f ( y it,p it;β) = ln y1 + 2 ln y2 + 3 ln y3 + 4 ln p1 + 5 ln p ln y1 + 7 ln y2 + 8 ln y3 + 9 ln p1 10 ln ln p1 ln p ln y1 ln y ln y1 y ln y2 ln y3 15 ln y1 ln 16 ln y1 ln p ln y2 ln p ln y2 ln p ln y3 ln p1 20 ln y3 ln + +,,,,,, 3.11 And the specification of g-function in Equation (3.0) is the same, though the estimation of parameters and coefficients differs. The dependent variables are TC and ROA for cost and profit frontiers, respectively. There are three output variables: y1=ln (the total loans), y2=td (total deposits), and y3=fi (Financial Investments). The two input price variables are p1 for cost of funds (total interest expenses divided by total deposits) and p2 for the cost of labor (the wage level of financial sectors at the country level). The other items in the cost and profit frontiers are the product of input prices and output following the translog functional form. Regarding the choice between the production method and intermediary method in terms of specifications of the cost and profit frontiers, we chose the intermediary method following Berger and Humphrey (1997) and Fu (2009) that the Chinese banking market is p 2 2 p p The choice of LnROA instead of Ln(profit) is caused by the problem of heterscedisty of the random error e. It is difficult to handle the heteroscedasty for both random error e and ransom inefficiency term m. LnROA deals with the heteroscedasty of the dependent variables of profits caused by the huge variety of the sizes 2 across firms and the σ mit deals with the heteroscedasty of the residual terms. 11

12 monopolistic competitive. It makes sense to adopt the intermediation method when the market mechanism works across financial institutions. That is, higher quality of services entails higher levels of outputs. In practice, there is no clear causality between the amount of deposits and that of loans in the Chinese banking system. Regarding the choices of outputs and inputs, we have checked the data sources and literature thoroughly. Although a recent study by Guarda, Rouabah, and Vardanyan (2013) suggests a data-oriented approach to choose these variables, our choice of variables is very limited. For instance, the data in recent years classify the deposit in detail and provide more fee- based items, but the length of the variables is short or not available for many non-stateowned banks. The related research in Chinese banking is confined to the availability of data. If the classification of variables is more detailed, then the available longitude of the variables is limited. When we target to investigate the time-varying patterns of efficiencies, the choices of data are limited. Therefore, we chose three output variables and two input price proxies of this paper. Key determinants of bank efficiency (Z) include SIZE, Tier1, Equities/Total assets, NPL reserve ratio. The SIZE variable is measured by the natural logarithm of the total assets of a bank. It is used to control the effect of the size of a bank. Tier 1, E/Total Assets, NPL reserve ratio are the factors depicting the risk level of a bank. Tier 1 is the Basel II required capital ratio; Capital ratio is the traditional capital adequacy ratio 4 ; Equities/total assets simply the total equities to total assets in percentage; and NPL reserves/ln is the proportion of NPL reserves to the total loans in percentage. NPL is a particularly important factor in the state-owned banks as documented by Fan (2003), Lu et al. (2005), and Lin and Zhang (2009). 4 12

13 4. Data Sources, Summary of Variables and Empirical Results 4.1 Data Sources The data of banks in China are collected from Bank Scope data bank. The banks and the time periods for the empirical work are listed in Table 1. In total, we have 884 observations for 124 banks and the time periods span from 1996 to Among these 124 banks, 10 are state-owned banks, 82 are joint stock banks, and 32 are regional commercial banks, while the Bank of Communication was classified as a joint stock bank before 2006 and has become a state owned one after 2006 up to now. The limited number of the regional commercial banks is caused by the time period of this study. Most city commercial banks are established or with available data after the year City commercial banks and banks established by towns or villages altogether are called regional commercial banks here. ================= Insert Table 1 here ================= The one hundred and twenty-four banks selected in this study are based on the availability of the related variables. The numbers of banks are very limited before year 2007 (21 banks). The variables of some banks are culled directly from the annual reports of the banks. The data from 2007 to 2014 are collected from the Bank Scope though variables of many banks are not available in year Only 14 banks are in the sample with complete sets of variables. Since the BC stochastic frontier model can deal with the unbalanced panel data and we would like to include as many observations as possible, we are using this data set with the various time periods for different banks, that is, an asymmetric panel data set. The banks are classified as three groups: State-owned banks 13

14 regional commercial banks, and joint stock banks. 5 Some previous studies have ignored existence of regional commercial banks (Lin, Mohanty and Lin, 2009). Shen and Lu (2008) have detailed the classification with one more category: Policy bank. Since we have only one policy bank (China Development Bank), it is combines with all other state-owned banks. The state-owned banks were called Big Four before, which are Bank of China, China Agriculture Bank, China Construction Bank and China Industrial and Commercial Bank. After the Bank of Communication has been re-classified as a public, state-owned bank after 2006, it becomes Big Five. The classification of banks under study basically stays the same in this data set. The regional commercial banks are in general regionally publicly-owned. They used to be city, town, or village credit unions that supported the development and funding of small and medium firms in the urban areas and have been reorganized as commercial banks from As local banks, the regional commercial banks risk is highly related to regional economic growth and policies. Since they were reorganized following the modern business law and banking regulations, they are less hierarchical in decision making and probably, more efficient in terms of management with better corporate governance structure. Nevertheless, their sizes vary a lot from bank to bank though most regional commercial banks are of medium and small size. Some regional commercial banks have all local branches while some banks such as Bank of Shanghai, have expanded business to several cities and listed as one of the five hundred largest banks all over the world (see In our sample, since we consider only the banks with longer history, these regional commercial banks under study are comparatively more solid and robust. In addition, all of them are located in the eastern prosperous regions in China. Thus, their 5 Foreign banks are not considered because of their different features. It reduces statistical efficiency when foreign banks in China are considered in the stochastic frontiers. 14

15 performance is better than other commercial banks located in the western areas with short history. 6 The joint stock banks are referred to as private banks with publicly traded shares outstanding in the market. For instance, Minsheng Bank is the first so-called privately owned, namely, joint stock. Several joint stock banks are featured with partially foreign ownership. In general, most joint stock banks are national banks and connected with some foreign banks so they usually enjoy the strengths in management, technologies, and sufficient capital flows. Because the history of the joint stock banks is still shorter than the state-owned banks, in terms of sizes and the number of branches, the state-owned banks still dominate the whole national market, while the regional commercial banks may have some domestic know-hows and strengths. It is of interest that all state-owned banks now are mixed with small proportions of foreign shares. After a series of reforms, the participation of foreign banks share entails the traditionally state-owned banks to be monitored by outsiders from the stock markets. Hence, the state-owned banks have become much more transparent especially after Summary Statistics ================ Insert Table 2 here ================ Table 2 summarizes the variables used. In Table 2, TC, y1, y2, y3, and p2 are denominated in 1996 million Renminbi (Chinese currency) while ROA, p1, Tier1, equity to 6 Some regional commercial banks have transformed into joint stock banks. This has been taken care of in the sample under study. For instance, the Bank of Beijing changed from reginal banks to joint-stock in 2000; Bank of Tianjin, 2007; Bank of Dalian, 2007; and Bank of Nanjing,

16 assets ratio and NPL reserve ratio are ratios in percentage. 7 From Table 2, we observe that the state-owned banks have highest TC ($97,821.7vs. $ and $ ), most outputs (see the values of y1, y2, and y3), and largest total assets ($3,522,380 vs. $205,110 and $68,997). In contrast, the regional commercial banks are featured with lowest TC, smallest Net Income, smallest outputs y1, y2, and y3, and the smallest size. We also find the state-owned banks bear highest cost of funds (p1) 7.720% against the regional commercial banks lowest 2.153%. In addition, the regional commercial banks outperform the state-owned banks with higher tier 1 ratio (11.766%), while the state owned banks Tier 1 ratio are 8.996%.The state-owned banks enjoy the lowest equities to total assets ratio (ET % vs and 6.425%) because they are backed by the government s share. The joint stock banks own the highest E/L ratio, which indicates they enjoy proportionally higher capital infusions than the regional commercial and state-owned banks. However, among these three types of banks, the joint stock banks have highest non-performing loans reserve ratio. It is possible that the joint stock banks are more aggressive in making loans to customers and thus prepare more reserves for more non-performing loans. The high NPL reserve harms the profitability and cost efficiency of the joint stock banks. It is noted that during these bank reforms, the state-owned banks had the privilege that the infusion of national capitals helped wipe out the non-performing loans. And the cost of these national capitals was very low or even costless. On the contrary, the joint stock banks needed to disclose all the cost of writing off the nonperforming loans in the financial statements. 4.3 Empirical Results 7 There are too may missing observations in the variables of the off-balance sheet items and off-shore banking. Therefore, these variables are not considered. 16

17 Table 3 displays the results of the one-step stochastic frontier following Battesse and Coelli (1995). Both cost and profit frontiers are estimated. We have also run the simple stochastic frontier with half-normal distributed inefficiency and the results are similar to the one step model. The frontiers with Gamma-distributed inefficiency are also computed, but the results of different distributions do not differ much so they are not displayed to save spaces. Panel I of Table 3 describes the results from 1996 to 2014; Panel II shows those from 1996 to 2006 for the pre-crisis years; and Panel III displays those from 2007 to 2014 for the post crisis period. The cost and profit efficiencies for each bank are measured and summarized for each year in Table 4. Figure 1 characterizes the changes in cost and profit efficiencies during this transitional period from 1996 to These empirical results are explained and interpreted from subsections to Cost Frontier The dependent variable of this cost frontier is the natural logarithm of the total costs that sum the operation costs (overhead) and the interest expense. Here, we focus on the cost frontier of the whole sample from 1996 to For the cost frontier, the coefficients of lny2, and lnp1 are significant at least at the 1% level and lnp2 at the 10 percent level. The parameters λ(lambda) and σ (Sigma) are statistically significant at the 1% level, which show that there exists the cost inefficiency and the use of the stochastic frontier approach is appropriate Profit Frontier Again, we focus on the profit frontier of the whole time period from 1996 to The dependent variable of the profit frontier is the natural logarithm of ROA. The independent variables are lny1, lny2, lny3, lnp1, and lnp2 that follow the translog functional form. The specification of this is named alternative profit frontier. The results of 17

18 profit frontier are also displayed in Table 3 where lny1, lny2, lnp2 are significant at the 5% significance level or less. Compared with the results of the cost frontier, the results of profit frontier are less significant since the exact specification of profit frontier is not available. This coincides with a majority of the so-called the empirical results of the alternative profit frontier from other papers The Factors of the Cost and Profit Efficiencies The means of cost inefficiency term u and the profit inefficiency m are explained by several factors. Wang and Schmidt (2002) and Wang (2002) estimated the inefficiency term along with the relevant control variables in the mean Both the frontiers with and without factors are estimated. The results of show that cost factors here (TT, SIZE, Tier1, NPL reserve to total loan) are negatively related to the profit inefficiency. All factors of the profit efficiency also show the same pattern. This shows the profit efficiency improves as time goes by and the Tier 1 ratio and NPL reserves help improving profit efficiency. Nevertheless, the CITY and STATE dummies are positively associated with the cost inefficiency so the regional commercial banks and state-owned banks may have lower cost efficiency than the other types of banks. And the commercial regional banks are less profit-efficient than the other types of banks. The factors of the cost and profit efficiencies are further analyzed in the two subperiods: and , where the year 2007 is the breakpoint when the Western financial crisis started. Before 2007, all factors of the cost inefficiency and profit inefficiency are insignificant. From 2007 to 2014, ET and NPL are negatively correlated to the cost inefficiency (i.e., positively related to cost efficiency) significantly at the 1 percent level. It is intuitive that the non-performing reserve to loan ratio is positively affiliated with the efficiency measure. Higher equity to total asset ratio helps a bank secure its wealth 18

19 and enhance its cost efficiency. Regional ownership dummy (CITY) is negatively and government ownership STATE dummy are positively and significantly (at the 5% level) correlated to the cost inefficiency. That is, the regional banks tend to be more cost efficient than state-owned banks. The state-owned banks are not as cost-efficient as the city (regional) commercial banks. Among the factors of the profit inefficiency, size and TIER 1 are negatively related to the profit inefficiency (positively correlated to profit efficiency) at the 1% level. Therefore, higher Tier 1 ratio of Basel accord helps a bank improve profit efficiency. The large size also contributes higher profit efficiency from 2007 to In the same sub-period, the CITY dummy is positively related to the profit inefficiency. In other words, the regional ownership may signal lower profit efficiency during It is more likely that the financial crisis affected the medium and smaller enterprises that rely more on the financing from the regional banks. When firms suffered from lower exports because of the financial crisis, the profit efficiency of the related banks may also decrease Time-Varying Patterns of Cost and Profit Efficiencies Table 4 and Figure 1 show the cost and profit efficiencies for each bank and the corresponding rankings. The cost efficiency CE=exp(u_min-u_i) and the profit efficiency PE=exp(-m ). It is particularly important to investigate the patterns of efficiencies along the series of bank reforms during periods. China joined the WTO in 2001 and has complied with the WTO requirements in In addition, risk-based capital standards (Pillar I), bank supervision (Pillar 2) and market disclosure (Pillar 3) requirements under Basel II are implemented in The cost and profit efficiencies are calculated and discussed for the three different types of banks. The changes in the profit efficiency and cost efficiency are displayed in Figure 1. It is obvious that the cost efficiency fluctuated up and down from 1996 to 2014 and the cost 19

20 efficiencies of three types of banks intertwined with one another. Profit efficiency (PE) fluctuates during this time period. From 1996 to 2001, it swings up and down. After 2001, PE increases smoothly all the way to 2007 and begins to drop after The decrease in PE after 2007 may be caused by the high competition in the banking market. High competition among banks will lead to low profitability for each bank and thus the profit efficiency will be low. The PE of the state-owned banks is the highest from 1996 to 2004 and it becomes the lowest among three types of banks from 2006 to It is likely that the profit efficiency of the state-owned banks was affected most deeply by the western financial meltdown given the fact that the Basel Tier 1 capitals have been reserved to reduce the risk. And the profit efficiencies for the regional commercial banks and joint venture banks remained higher than that of the state-owned banks. We may conclude that the state-owned banks may not perform very differently in terms of the cost efficiency but their profit efficiencies may underperform, compared with the other two types of banks. It is worth noting that PE in year 2009 is , falling to in The fast decline in the profit efficiency signals the declining profitability while the cost efficiency of banks stays high. More studies need to be done to address this issue and provide reasoning of the decaying profit efficiency. In terms of bank management, the management needs to find out some strategies to improve the profit efficiency. In terms of the policy makers perspective, they have to detect the roots of the decreasing profit in macroeconomy. 5. Concluding Remarks and Policy Implications This paper estimates the cost and profit efficiencies for124 banks in China during period. Three types of banks: state-owned, regional commercial, and joint stock banks are analyzed. China joined the WTO in Since then, Chinese regulatory authority has undertaken a series of reforms such as privatization via initial public offerings 20

21 and foreign ownership. As a result, the landscape of the Chinese banking industry has been changed dramatically. Among three different types of banks under study, the state-owned commercial banks (SOCBs) tend to be large in terms of asset size. However, our findings show that the joint stock banks benefit from higher Tier 1 capital ratio, and lower nonperforming loan reserve ratio. That is, the regional commercial banks gain from higher capital adequacy and lower asset risk. Thus, their return on assets is more than those of the other two types of banks. And the joint stock banks have higher leverage ratio and the higher non-performing loan reserve ratio. The results show that the state ownership and Tier I capital ratio help reduce the cost inefficiency of a bank while the equities to total assets ratio is negatively associated with the cost efficiency. The overall trend of cost efficiency has improved over time from 1996 to 2014, though the cost efficiency scores tend to converge after The trend of the profit efficiency is not clear because it fluctuates prior to 2001 and then it increases between 2002 and 2007; and it declines again after 2007 until These findings here coincide with the results from the cost and profit frontier estimations. Overall, the results show that the series of bank reforms in general have helped improve banking efficiency in China. These results are consistent with the findings reported by Lin, Sun, and Wu (2015). There are several implications based on the findings of our study. First, the bank reforms in China have helped improve the cost efficiency of Chinese banks. In particular, the state-owned commercial banks have become more cost efficient relative to other banks during the post- Basel III era ( ). Second, the cost efficiency in the regional commercial banks and the joint stock banks continued to improve over the period. Third, the regional commercial banks seemed to have benefitted from higher profit efficiency. The high profit efficiency attributed to regional commercial banks are likely to be linked to relationship banking (or the so-called domestic wisdom) and smart marketing 21

22 strategies. These results are consistent with findings reported by Hasan et al Since banking markets in China are geographically segmented, it is likely that a regional commercial bank might gain from its local connections with regional businesses (e.g., Dobson and Kashap, 2006; Hasan et al.; 2015). This evidence is also supported by the city or village government, while the joint stock banks are usually affiliated with foreign banks. In addition, the regional commercial banks are comparatively more flexible in their deposittaking and loan approval policies. Efficient management with fewer hierarchies could be the other reason for regional banks to remain more efficient. Fourth, China s participation in WTO and its adoption of Basel II and III capital ratios and risk management policies are likely to have helped improve cost efficiency of the banking sector as whole. Our results show that Tier 1 capital ratio is positively significantly related to the cost efficiency during both pre- and post- Basel II eras. Thus, the increased competition with higher capital ratio requirements triggered by bank reforms might have led to increased cost efficiency, but the bank reforms do not guarantee an increased profit efficiency for banks. It is worth noting that the continued decline in profit efficiency may signal the future instability of the banking sector in China. Moreover, regulators have to tightly monitor the changes in economic environment as well as the risk-taking behavior of profit-seeking managers. 22

23 References Barth, J. R.,Koepp, R., and Zhou, Z. (2004), Banking Reform in China: Catalyzing the Nation s Financial Future, SSRN Working Paper. Battese, G. and T. J. Coelli (1995), "A Model for Technical Inefficiency Effects in a Stochastic Frontier Production for Panel Data." Empirical Economics 20, Berger A.N. and D.B. Humphrey (1997). Efficiency of Financial Institutions: International Survey and Directions for Future Research. European Journal of Operational Research 98(2): Berger, A.N.; Klapper, L. F. and Peria M.S.M. and Zaidi, R. (2008), Bank ownership type and banking relationships, Journal of Financial Intermediation 17, Berger,A. N., Hasan, I. and Zhou, M.,(2009), Bank Ownership and Efficiency in China: What Will Happen in The World s Largest Nation? Journal of Banking and Finance, 33(1), pages Berger, A. N., I. Hasan, M. Zhou (2010), The Effects of Focus versus Diversification on Bank Performance: Evidence from Chinese Banks, Journal of Banking and Finance 34(7), Berger, A. N. and C. H.S. Bouwman (2011) How Does Capital Affect Bank Performance during Financial Crises? Wharton Financial Institution Center Working Paper Series No Berger, A.N. (2013), How does capital affect bank performance during financial crises? Journal of Financial Economics 109(1), Boyd, J.H. and Gertler, M. (1994), Are banks dead? Or are the reports greatly exaggerated? Federal Reserve Bank of Minneapolis Quarterly Review, Summer, Cai, Zhuang and Peter Wheale (2007), The new capital accord and the Chinese banking industry, Journal of Banking Regulation 8,

24 CBRC. (2007). The Guidelines on the Implementation of the New Basel Accord by China s Banking Sector. Chen, J., Jiang, C., and Lin, Y. (2014) " What determine firms' capital structure in China?", Managerial Finance 40(10), Chen, X., Skully, M. and Brown, K. (2005), Banking Efficiency in China: Application of DEA pre- and post- Deregulation Eras, , China Economic Review 16 (3): Demirgüç-Kunt,A., and H. Huizinga (2012) Do we need big banks?evidence on performance, strategy and market discipline BIS paper Dobson, W. and A.K. Kashyap (2006), The Contradiction in China s Gradualist Banking Reforms, Brookings Papers on Economic Activity 2, Fan, G.(2003) China s Nonperforming Loans and National Comprehensive Liabilities, Asian Economics Paper 2:1, Fenech, J.P., Yap, Y.K., Shafik, S., (2014), Can the Chinese banking system continue to grow without sacrificing loan quality? Journal of International Financial Markets, Institutions and Money 31, Fiordelisi, F., Marques-Ibanez, D. Molyneaux, P., (2011), Efficiency and risk in European banking, Journal of Banking and Finance 35, Foo, J. and Witkowska, D. (2014) An efficiency comparison of Chinese banks: a multidimensional analysis, International Journal of Business 19(1), Fu, X. (2009) Competition in Chinese Commercial Banking, working paper in 22nd Australasian Finance and Banking Conference. 24

25 Fu, X. and Heffernan, S. A.(2007), Cost X-efficiency in China's Banking Sector, China Economic Review 18, Fu, X. and Heffernan, S. A.(2008), Economies of Scale and Scope in China's Banking Sector, Applied Financial Economics18, Fu, X and Heffernan, S.(2009), The effects of reform on China s bank structure and performance, Journal of Banking & Finance 33(1), Guarda P., Rouabah, A. and Vardanyan, M. (2013), Identifying bank outputs and inputs with a directional technology distance function, Journal of Productivity Analysis 40(2), Greene, W. (2007) LIMDEP 9.0 Handbook. Hasan, I., Kobeissi, N., Wang, H. and Zhou, M. (2015), Banking Structure, Marketization, And Small Business Development: Regional Evidence From China, Pacific Economic Review, 20: 3, He, L.-P.(2001), Facing the WTO Accession: Problems and Challenges in China s Banking Industry, China Macroeconomic Information Network, July 24, Hefferman, S. and Fu, X. (2010) Determinants of financial performance in Chinese banking, Applied Financial Economics 20(20): Huang X., and Xiong Q. (2015), Bank capital buffer decisions under macroeconomic fluctuations: Evidence for the banking industry of China, International Review of Economics and Finance 36 (2015) IIF Research Note (2014) International Expansion of Chinese Banks, df. Jagtiani, J., Nathan, A. and Sick, G. (1995), Scale economics and cost complementarities in commercial banks on-and off-balance-sheet activities, Journal of Banking & Finance, 19, pp Jiang C., S. Yao, G. Feng, (2013) "Bank Ownership, Privatization, and Performance: Evidence from a Transition Country", Journal of Banking and Finance (37),

26 Knaack, P. (2014). From laggard to primus -Why is China exceeding global banking standards? Working paper presented in the 2014 annual meeting of the American Association for Chinese Studies, Washington D.C., and available at Leung,M. K.; D. Rigby; T. Young (2003), Entry of foreign banks in the People's Republic of China: a survival analysis, Applied Economics 35( 1), Lieu,P T, T LYeh and Y-H Chiu, (2005), Off-Balance Sheet Activities and Cost Inefficiency in Taiwan s Banks, Service Industries Journal, 25(7), Li,Q., Zeng,Y., and Liu, B. (2014) Asymmetric Information, Foreign Entry and Multiperiod Credit Competition in Banking Industry, Quarterly Review of Economics and Finance 54(2), Li, Y., Xu, T. and,yuan, H. (2015), Spillover Effects of Foreign Bank Entry in China's Banking Sector, International Business Research, 8 (1), Li, K.-W.and Ma, J. (2004), The Economy Intricacies of Banking Reform in China, Chinese Economy, 37 (4): pp Lin, W.T., Lin, H.-J., and S. K. Mohanty,(2009), The Cost and Profit Efficiencies and Scope Economies of Commercial Banks: Evidence from Taiwan and China, International Journal of Finance 21(2), Lin, X, and Y. Zhang (2009), Ownership reform and bank performance in China, Journal of Banking & Finance 33(1), Lin, J. Y.; Sun, X.; Wu, H. X. (2015) Banking Structure and Industrial Growth: Evidence from China. Journal of Banking and Finance 58, Lu, D., S. M. Thangavelu and Q. Hu (2005), Biased Lending and Non-performing Loansin China s Banking Sector, Journal of Development Studies 41(6), Molyneux, P., H. Liu and C. Jiang (2014), Bank capital, adjustment and ownership: Evidence from China, BOFIT Discussion Papers 16 available at 26

27 Shen, C.-H.and C.-H. Lu (2008) Is There A Silver Lining in the Cloudy Performance of Chinese Banks? -- An Empirical Investigation Into The Determinants of Profitability and Risk, In Emerging Topics in Banking and Finance. Edited by E.J. Fuchs and F. Braun, Nova Science Publishers, Inc. ISBN: Wang, H.J. (2002), "Heteroscedasticity and Non-Monotonic Efficiency Effects of a Stochastic Frontier Model," Journal of Productivity Analysis 18, pp Wang, H.-J.and Schmidt, P. (2002), One-Step and Two-Step Estimation of the Effects of Exogenous Variables on Technical Efficiency Levels, Journal of Productivity Analysis, 18: Wang K, Huang W, Wu J, Liu YN. (2014) Efficiency measures of the Chinese commercial banking system using an additive two-stage DEA. OMEGA, 44, Yao, S., C. Jiang, G. Feng and D. Willenbocke (2007), WTO challenges and efficiency of Chinese banks, Applied Economics, 39, Yu, H.-C.; Tong, D. P. (2015), Banking Relationships, R&D Investment, and Growth Opportunities in China. Banks and Bank Systems 10(2), Zhang, J., C. Jiang, B.I. Qu, and P. Wang, (2013) "Market concentration, risk-taking, and bank performance: evidence from emerging economies", International Review of Financial Analysis (30), Zhu, W., and Yang, J. (2016), State ownership, cross-border acquisition, and risk-taking: Evidence from China s banking industry, Journal of Banking & Finance 71,

28 Table 1 The List of Banks and Time Periods under StudyYear Obs Type Obs

29

30 Type=1 State-owned Type=2 Joint Stock Type=3, Regional Commercial Banks Notes: The data set is collected from BankScope. This is an asymmetric panel data set. Among 124 banks, 10 are state-owned, 82 are joint stock banks, and 32 are regional (including city, township and rural) commercial banks 30

31 Table 2 Summary Statistics All Banks Variable TC NI Y1 Y2 Y3 P1 P2 TA ET TIER1 NPL Mean Std.Dev. 21, , , , , , ,282 50, , ,006,860 1,478,360 6,391, ,853, ,677 State-Owned Banks Variable TC NI Y1 Y2 Y3 P1 P2 TA ET TIER1 NPL Mean 97, , ,995,600 2,645,880 1,140, ,522, ,414 Std.Dev. 82, , ,634,990 2,663,170 1,102, ,151, ,469 Joint Stock Banks Variable TC NI Y1 Y2 Y3 P1 P2 TA ET TIER1 NPL Mean 6, , , , , , ,837 Std.Dev. 13, , , ,904 6,816, , ,472 Regional Commercial Banks Variable TC NI Y1 Y2 Y3 P1 P2 TA ET TIER1 NPL Mean 1, ,162 54, , , ,126 Std.Dev. 2, ,248 54,323 7,851, , ,409 Note: Dependent variables: TC: total costs of each bank including overheads (i.e., operating costs and personnel wages and benefits) plus interest costs in thousand Renminbi. ROA: return on assets in percentage. Output variables (Y): y1=tl: Total loans in million Renminbi, including all long term loans and short term loans. Non-performing loans and other receivables are not included. 31