How does banking competition affect financial stability? : the case of China

How does banking competition affect financial stability? The case of China

2003-2012

Bachelor Thesis

Sebastiaan Mullens

10220283

24-01-2014

1. Introduction

For long it has been debated among scholars and policymakers how banking competition affects bank’s risk-taking behaviour and financial stability. While in general it is assumed that competition is beneficial to society as it reduces prices and increases innovation and product availability, it has been questioned to what extent competition among banks is beneficial to society. This is particularly an important question when one considers the central role that the banking sector plays in the economy and the potential spill over to the real economy in the event of financial instability. Indeed, financial liberalization has often been cited as one of the main causes of financial instability and bank’s risk-taking behaviour. For example, the lifting of state branching restrictions in the early 1980s and the subsequent increase in competitive behaviour among banks in the United States has been identified as one of the main causes of the Savings and Loan (S&L) crisis at the end of the 1980s (Hellmann et al. (2000)). Also, the deregulation of the financial sector in many East-Asian countries in the early 1990s has been recognized as one of the main factors leading to the Asian Financial crisis, which erupted in 1997 (Hellmann et al. (2000)). Most recently, the 2008 financial crisis that originated in the United States has sparked the discussion what role financial deregulation and competition among banks have played in causing this crisis (Fu et al. (2014)). While the view that competition has a detrimental effect on financial stability has prevailed, in recent years evidence has emerged that supports the view that banking competition may have a positive effect on financial stability.

This study has the objective to answer the question how banking competition affects financial stability in China, considering the period from 2003 until 2012. During the last three decades the Chinese banking system has been greatly reformed and liberalized. Before 1978 the Chinese banking sector was meticulously controlled by the state and competition was virtually absent in the Chinese financial landscape. The banking system was subjected to lending quota’s, interest rates were repressed and regulations were in place that restricted the entry of new banks. From 1978 onwards the Chinese banking sector has been gradually restructured and liberalized. While reforms are still ongoing and some restrictions remain, a legion of banks now are allowed to compete with each other for market share.

Also, while the Chinese economy has become the second largest economy in the world, there exists surprisingly little research on banking competition in China. This study therefore also aims to add to the small amount of literature available on banking competition in China.

The rest of this paper is organized as follows. Section 2 provides a historical background of important banking reforms in China. Section 3 gives a literature review about the relationship between competition and financial stability in general, as well as specifically on China. Section 4 describes the research method, while section 5 displays the data used and the descriptive statistics. Section 6 will provide the results and a discussion of the results, and finally section 7 concludes.

2. Historical background

The year 1978 marks a turning point for the Chinese economy when the China Communist Party (CCP) started to gradually loosen its grips on the economy, which until then was subject to careful planning. Before this time nearly every aspect of the Chinese economy was controlled by the state, all in order to realize the economic development agenda of the CCP, which had the vision that China would be able to catch up with the West if it were to develop its heavy industry (Lin et al. (2003)). Also the financial system was completely controlled by the state and consisted of only a single mammoth banking organization, the People’s Bank of China (PBoC), which performed both the function of central and commercial bank (Wong and Wong (2001)). The banking system in reality was used as an extension of the government budget to finance government projects and enterprises, and commercial considerations and competition therefore were virtually absent in the Chinese financial landscape (Li (2009)). However, the new policy objective of the CCP to create a socialist market economy from 1978 onwards asked for a commensurate liberalization of the financial system so that it could support the new economy by allocating capital to the most innovative and fastest growing sectors of the economy in the same fashion financial institutions do in capitalist economies. Unlike many former Soviet countries who made a “big bang” transition to a market economy, China’s reform process has been gradual and is still ongoing.

The first major financial reforms came between 1978 and 1984 when the commercial banking and the central banking activities of the PBoC were separated from each other (Wong and Wong (2001)). While the PBoC continued to exercise its function as the central bank, four new commercial banks were established to take over the commercial banking activities (Wong and Wong (2001)). The Bank of China (BOC) focused on currency transactions and international business, the Agricultural Bank China (ABC) took over all rural banking activities, the China Construction Bank (CCB) became involved in financing construction projects and finally the Industrial and Commercial Bank of China (ICBC) assumed responsibility for all industrial and commercial banking activities (Wong and Wong (2001)). Although these four banks were formally separated into independent entities, the state remained the sole shareholder of these banks and competition was still absent as these four State-Owned Commercial Banks (SOCBs) were restricted to doing business in their assigned sectors, just like they had been doing when being part of the PBoC. However, the establishment of these four commercial banks can be seen as the groundwork for introducing elements of competition into the financial system in later years. In 1985, for example, the activity and geographic restrictions were lifted for the SOCBs, allowing them to compete with each other for the first time (Berger et al. (2009)).

In the 1990s, the financial sector was further restructured and liberalized, which opened up the possibility for Chinese banks to intensify competition. Firstly, in 1994 three new policy banks

were established by the central government to take over policy lending from the four SOCBs (Yuan (2006)). To further protect the commercial banks from government interference a Commercial Banking Law (CBL) was instituted in 1995, which established a legal framework for commercial banking for the first time in China. The 1995 CBL formally instituted the independence of the four SOCBs from the central and local governments, although it is questioned to what extent this independence holds in reality (Yuan (2006)). Another important development was the entry of new banking institutions in the financial sector. Towards the end of the 1990s, 12 Joint-Stock Commercial Banks (JSCBs) and more than 100 smaller City Commercial Banks (CCBs) had been established (García-Herrero and Santabárbara (2004)). These newly established institutions have been allowed to compete with the SOCBs for market share. Additionally, over the years many geographic and activity restrictions that these newly established institutions had been subjected to have been lifted to allow for greater competition. As a result, the market share of the SOCBs has been steadily decreasing, from about 72.1% in 1994 to 44.9% in 2012, while the market shares of the JSCBs and CCBs has increased from 5.1% and virtually zero in 1994 to 17.6% and 9.2% in 2012 respectively (Garcıa-Herrero et al. (2006)) (Appendix 1).

Financial reforms have continued in the 2000s. An important development was China’s accession to the World Trade Organization (WTO) in December 2001, which obliged it to provide foreign banks a level playing field from 2006 onwards. But again, economic reality may not completely correspond to the legal framework as it is well-known that foreign banks that wish to enter China’s financial markets must go through a tedious process of gaining approval to open new branches, a practice that continues until today and puts off foreign banks to develop branch networks in China. Nevertheless, China’s accession to WTO has prompted the government to further allow for competition in the financial sector in order to make domestic banks more competitive and more efficient as a preparation for the opening up of China’s financial sector to foreign banks (Xu et al. (2013)). In 2004, for example, the floor on deposit interest rates and the ceiling on lending rates were abolished, which gives Chinese banks more space to compete on prices, although other interest restrictions still remain (Garcıa-Herrero et al. (2006)).

3. Literature review

3.1 Banking competition and financial stability

There exists an extensive literature about the effect of banking competition on financial stability. In general, there are two different views, the competition-fragility view arguing that more competition leads to less financial stability and the competition-stability view arguing that more competition leads to more financial stability. Although the former perspective has prevailed, in the most recent years evidence has appeared in favour of the latter perspective (Schaeck (2009)).

The traditional competition-fragility or moral hazard view, argues that a higher degree of competition leads to more risk-taking behaviour and consequently leads to a more fragile financial system. Under this view, an increase in competition will lower prices and erode profit margins, which then again weakens a bank’s capital ratios and ultimately a bank’s charter value (Keeley (1990)). In order to mitigate the adverse effect that the lower profit margins have on a bank’s charter value and improve profit margins, a bank is willing to take on more risk (Keeley (1990)). However, when a banking sector is characterized by a lower degree of competition, for example as a result of entry restrictions, banks can earn monopoly rents. The ability to earn monopoly rents increases the value of the charter that a bank possesses. A bank would then be less inclined to take on risk since bankruptcy would imply the loss of its valuable charter (Keeley (1990)).

Empirical analysis has been conducted both as a cross-country analysis, where several countries are included, and as individual country analysis. Ariss (2010) for example, performs an empirical analysis with an unbalanced panel of 821 banks in 60 countries. Three different versions of the Lerner-index are used to gauge competition and the Z-score is used as a representative for banking risk. The results show a significant positive relationship between market power (inverse measure of competition) and financial stability. Jimenez et al. (2007) conduct an empirical analysis on the Spanish banking sector. They find that the Lerner-index is negatively correlated with bank’s risk-taking behaviour, that is to say, an increase in the Lerner-index negatively impacts the NPL-ratio, in line with the competition-fragility view.

In more recent years, an alternative competition-stability view has emerged, arguing that a higher degree of competition can lead to a more stable financial system. One of the channels through which an increase in competition may lead to a more stable banking system is that it lowers the interest rates being charged to customers. Lower interest rates then reduce the interest rate burden on the economy, lowers default rates and ultimately improves financial stability (Boyd and Nicoló (2005)). In addition, a lower degree of competition enables banks to charge higher lending rates, which decreases the profit margins of the lenders. which may induce more risk-taking behaviour on the part of lenders in order to maintain profit margins (Boyd and Nicoló (2005)). Another channel through which a lower degree of competition and a higher degree of concentration may lead to greater fragility is the perception on the part of banks that they are “”too big to fail” and that consequently the authorities will save them when they are threatened with bankruptcy (Beck et al. (2006)). When banks assume that they will be saved by the authorities they could start to take on more risk.

Empirical research that supports the competition-stability nexus includes Schaeck and Cihak (2007) who have used a sample of 2600 banks in 15 European countries. They find that banks are inclined to hold higher capital buffers in a more competitive environment. Also, the effect

competition has on the capital ratio is considerable, a 1 percent increase in the H-statistic increases the capital ratio by 5.6 to 5.9 percent. Liu et al. (2012) also find support for the competition-stability view when they analysed four South-East Asian countries.

Still, both views do not necessarily oppose each other. Berger et al. (2009), for example, test for the relationship between competition and stability using data of 8,235 banks in 23 developed countries. They use the Z-index, Equity/Total Assets and a Non-Performing Loan (NPL) ratio as dependent variables and the Lerner-index and Herfindahl-Hirschman Index (HHI) as independent variables. Their findings indicate that more market power is associated with a more risky loan portfolio, as measured by the NPL-ratio, which is in line with the competition-stability view. At the same time, however, market power is negatively related with the Z-score, indicating that less competition leads to saver banks, which is in line with the competition-fragility view. According to Berger et al. (2009), while a decrease in competition may lead to a more risky loan portfolio, overall risk need not increase as a bank can apply various risk-mitigating techniques to maintain overall risk exposure, for example by holding more equity capital or keeping a smaller loan portfolio.

A complication in these strands of literature is the problem of measuring competition. According to the Industrial Organization (IO) literature, concentration is an inverse measure of competition, also known as the Structure-Conduct-Performance (SCP) paradigm (Schaeck (2009)). The SCP paradigm concurs that the structure of the banking industry determines the conduct of the banks that are active in the industry, which then again determines the performance of these banks. The structure of the banking industry or its level of concentration therefore has a direct impact on bank’s risk-taking behaviour and financial stability. The level of concentration can be measured by for example the HHI, which takes into account the market shares of all firms in the industry, or alternatively by a concentration ratio like the C3-ratio, which adds up the market shares of the three largest banks in the industry.

In the most recent years, however, the view has emerged that concentration does not always properly measures competition, mainly because it has been shown that banking sectors with a higher degree of concentration do not necessarily need to show low levels of competition (Casu and Girardone (2006)). For example, merger waves in Europe in the last decades have increased overall concentration while at the same time competition has intensified (Casu and Girardone (2006)). Consequently, apart from the structural measures mentioned above, other measures that are based on firm-specific data are often included in studies that analyse competition in the financial sector. The Lerner-index, for example, measures the degree of pricing power firms have and is often included in competition studies. Also the H-statistic and PE-indicator are used.

3.2 Banking competition in China

While there is an extensive literature on competition and financial stability in general, the competition and financial stability literature on China is quite scarce. Most studies on competition in China focus on the relationship between competition and efficiency. For example, Fungacova et al. (2012), use the Lerner-index as a competition measure and conclude that competition between 2001 and 2011 has decreased. Xu et al. (2013), however, compare several competition measures and apply them to China, concluding that competition increased in the years after China’s accession to WTO. They further point out that the use of different competition measures yield different conclusions about the development of competititon. According to Xu et al. (2013), the Lerner-index is not the most appropriate measure for banking competition in China as interest rate restrictions remain in China which prevents banks from competing on prices.

Empirical analysis on the relationship between competition and financial stability is even scarcer. China is included in cross-country analysis on the effect of competition on stability (Fu et al. (2014)). However, there is only one very recent case where only China is considered (Lee and Hsieh (2013)). Lee and Hsieh (2013) use the standard deviation of ROA and ROE as dependent variables and the HHI and the C4-ratio as independent variables with the conclusion that the increase in competition, as measured by the decrease of the HHI and C4-ratio, is negatively related with risk since the standard deviation of ROA and ROE increases.

This study seeks to add to the small amount of literature that is available on banking competition and financial stability in China and extend the study of Lee and Hsieh (2013) by including a non-structural competition measure, the Lerner-index, as independent variable, as well as focusing on the Z–score and Equity/Total Assets as dependent variables.

4. Research method 4.1 Regression equations

The research method used in this study is a combination of elements that are used in several other studies on competition and financial stability (Berger et al. (2009)) (Lee and Hsieh (2013)). Equity/Total Assets and the Z-score are used as dependent variables to measure financial stability. The C5-ratio and the Lerner-index are used as independent variables to measure competition. The regression equations take the following forms:

Financial Stabilityit=ƒ(C5-ratiot, Control Variablesit) (1)

Financial Stabilityit=ƒ(Lerner-indexit, Control Variablesit) (2)

Financial Stabilityit=ƒ(C5-ratiot, Lerner-indexit, Control Variablesit) (3)

Following Lee and Hsieh (2013), the natural logarithm of Total Assets, GDP growth rate, Liquid Assets/Total Assets, Loans/Deposits and Net Loans/Total Assets are included as control variables. Furthermore, the regressions have been subjected to fixed effects to control for factors that are constant per year or per entity.

4.2 Dependent variables

For the dependent variables, two different estimators are used. The ratio 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑖𝑖𝑖𝑖

𝑇𝑇𝑇𝑇𝐸𝐸𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐸𝐸𝐴𝐴𝑖𝑖𝑖𝑖 is widely used

as a proxy for banking safety. Equity can be seen as a buffer which banks can use to withstand economic shocks, hence more equity makes a bank safer.

The Z-score is an alternative measure for banking safety and is also widely used in the literature. The interpretation is as follows: the part in the numerator, 𝑅𝑅𝑅𝑅𝑅𝑅𝐸𝐸𝐸𝐸+_{𝑇𝑇𝑇𝑇𝐸𝐸𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐸𝐸𝐴𝐴}𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑖𝑖𝑖𝑖 _{𝑖𝑖𝑖𝑖},

measures the total amount of funds available for equity holders. The Z-score, which divides the funds available for equity holders by the standard deviation of ROA, measures how many standard deviations away the bank is from insolvency. In this measure a risk element is included in the form of the standard deviation of ROA, the more volatile the ROA, the more risky the business is. Therefore, the higher the Z-score, the safer the bank.

Another often used measure in the literature is the 𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖𝑖𝑖

𝑇𝑇𝑇𝑇𝐸𝐸𝑇𝑇𝑇𝑇 𝑁𝑁𝑇𝑇𝑇𝑇𝐿𝐿𝐴𝐴𝑖𝑖𝑖𝑖 ratio, which divides the

amount of Non-Performing Loans (NPLs) by Total Loans. The NPL-ratio increases when customers are unable to pay back their loans and can therefore be interpreted as a measure of risk-taking behavior of banks. When banks apply more lenient credit conditions and lend money to more risky customers, this will ultimately be reflected in an increase of NPLs. This measure however is not included in this study because in the case of China it might be biased in some ways.

It has been recognized in the literature that the amount of NPLs in the books of financial institutions may not reflect the true amount of NPLs. The amount of NPLs might be understated because banks do not always recognize loans as non-performing but simply renew the loans and hence continue lending to companies that are actually not able to fulfill their obligations (Dobson and Kashyap (2006)) (Koons (2013)). Several dynamics are at work that cause banks to engage in this practice. Firstly, while commercial banks are formally independent entities and by law are supposed to make lending decisions on a commercial basis, state intervention still remains pervasive in the banking sector (Bailey et al. (2011)). Banks are often pressed by the central and local governments to finance State-Owned Enterprises (SOE), even when these SOE do not meet credit standards (Martin (2012)). The central and local governments are able to exert this pressure because the state until today maintains a large ownership stake in the Chinese banking sector and has the authority to appoint board members and other bank officials (Martin (2012)). Secondly, even if a bank would want to impose a hard budget constraint on SOE, for example by pushing SOE into liquidation, there

does not exist a proper bankruptcy law to support these actions (Peng (2007)). In this way, SOE are protected from bankruptcy. The best course of action for banks in this case is to continue lending in the hope that the financial condition of the lender improves over time.

4.3 Independent variables

In the competition literature, two different types of competition measures are used, namely structural and non-structural measures. In this study, both a structural and a non-structural measure are used.

Firstly, the C5-ratio is used as an independent variable, which is the sum of the market shares of the five largest banks in China. The C5-ratio measures the degree of concentration in the banking sector, where high values indicate a very concentrated banking industry and where low values indicate a lower degree of concentration. According to the Structure-Conduct-Performance (SCP) paradigm, the degree of concentration is directly linked to the degree of competition.

𝐶𝐶5𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝐸𝐸= � �_∑𝑇𝑇𝑟𝑟𝑟𝑟𝑟𝑟𝑇𝑇 𝑅𝑅𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟𝐴𝐴_{𝑇𝑇𝑟𝑟𝑟𝑟𝑟𝑟𝑇𝑇 𝑟𝑟𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟𝐴𝐴}𝐸𝐸𝐸𝐸 𝐸𝐸𝐸𝐸 𝐿𝐿 𝐸𝐸=1 � 5 𝐸𝐸=1 (4) The five largest banks in China are the Bank of China (BOC), Agricultural Bank of China (ABC), Industrial and Commercial Bank China (ICBC), China Construction Bank (CCB) and finally the Bank of Communications. This ratio is particularly interesting because it reflects the intensive restructuring of the banking sector in the last two decades. It is expected that the lifting of entry restrictions in recent decades will show a declining trend as newly-entered banks take away market share from the five SOCBs.

Secondly, due to the debate over the right way to measure competition, the Lerner index, which is a non-structural measure of competition, is included in this study. The Lerner-index in essence measures the market power of a firm, or in this case banking institutions in China. When a firm charges a price close to marginal cost a firm is said to have low market power and the Lerner-index is close to zero. However, when a firm can charge a high margin over marginal cost, the firm is said to have high market power and the Lerner-index will show higher values. The advantage of this measure over the concentration measure is that it provides a competition measure specific to each individual firm.

𝐿𝐿𝐴𝐴𝑟𝑟𝐿𝐿𝐴𝐴𝑟𝑟𝑟𝑟𝐿𝐿𝐿𝐿𝐴𝐴𝐿𝐿𝐸𝐸𝐸𝐸 =𝑃𝑃𝑟𝑟𝑟𝑟𝑃𝑃𝐴𝐴𝐸𝐸𝐸𝐸− 𝑀𝑀𝑟𝑟𝑟𝑟𝑀𝑀𝑟𝑟𝐿𝐿𝑟𝑟𝑇𝑇 𝑃𝑃𝑟𝑟𝐴𝐴𝑟𝑟_{𝑃𝑃𝑟𝑟𝑟𝑟𝑃𝑃𝐴𝐴} 𝐸𝐸𝐸𝐸

𝐸𝐸𝐸𝐸 (5)

In Equation (5), 𝑃𝑃𝑟𝑟𝑟𝑟𝑃𝑃𝐴𝐴𝐸𝐸𝐸𝐸 =_{𝑇𝑇𝑇𝑇𝐸𝐸𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐸𝐸𝐴𝐴}𝑅𝑅𝐴𝐴𝑅𝑅𝐴𝐴𝐿𝐿𝐸𝐸𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖_{𝑖𝑖𝑖𝑖} and 𝑀𝑀𝑟𝑟𝑟𝑟𝑀𝑀𝑟𝑟𝐿𝐿𝑟𝑟𝑇𝑇 𝑃𝑃𝑟𝑟𝐴𝐴𝑟𝑟𝐸𝐸𝐸𝐸 is calculated by taking the

derivative of the total cost function. The total cost function for each bank is estimated by the Trans-logarithmic Cost Function (TCF) with one output, which is approximated by total assets, and two

input prices, namely the price of capital and the price of borrowed funds. Furthermore, following the literature, homogeneity and symmetry restrictions are applied. Also, to estimate the cost function more accurately the following cost function will be calculated separately for every bank type (that is to say SOCB, JSCB, CCB and RCB):

ln 𝑇𝑇𝑟𝑟𝑟𝑟𝑟𝑟𝑇𝑇 𝑃𝑃𝑟𝑟𝐴𝐴𝑟𝑟 = 𝛼𝛼0+ 𝛼𝛼1ln 𝑇𝑇𝑅𝑅 + 𝛼𝛼2ln 𝑇𝑇𝑅𝑅2+ � 𝛽𝛽𝑘𝑘ln 𝑤𝑤𝑘𝑘 2 𝑘𝑘=1 + � � 𝛽𝛽𝑘𝑘𝑘𝑘ln 𝑤𝑤𝑘𝑘ln 𝑤𝑤𝑘𝑘 2 𝑘𝑘=1 2 𝑘𝑘=1 + � 𝛾𝛾𝑘𝑘ln 𝑇𝑇𝑅𝑅 ln 𝑤𝑤𝑘𝑘 2 𝑘𝑘=1 + 𝛿𝛿1𝑟𝑟𝑟𝑟𝑡𝑡𝐴𝐴 + 𝛿𝛿2𝑟𝑟𝑟𝑟𝑡𝑡𝐴𝐴2+ � 𝛿𝛿3𝑘𝑘𝑟𝑟𝑟𝑟𝑡𝑡𝐴𝐴 ∗ 𝑇𝑇𝐿𝐿𝑤𝑤𝑘𝑘 2 𝑘𝑘=1 + 𝛿𝛿4𝑟𝑟𝑟𝑟𝑡𝑡𝐴𝐴 ∗ 𝑇𝑇𝐿𝐿𝑇𝑇𝑅𝑅 (6)

In Equation (6), Total cost = non-interest expense + interest expense, TA=Total Assets, w1=price of capital = non-interest expense/fixed assets, w2=price of borrowed funds= interest expense/total funding. The price of labor, which is included in other studies when estimating the total cost function, is not included here because data about personnel costs is not readily available. However, the labor costs are included indirectly in Equation (6) because personnel costs are part of the non-interest expenses.

The estimated coefficients of the TCF (Equation (6)), which can be found in appendix 2 for the different bank types, are then used to estimate the marginal cost function (Equation (7)) for each individual firm. The marginal cost is equal to the derivative of the total cost function and looks as follows: 𝑀𝑀𝑟𝑟𝑟𝑟𝑀𝑀𝑟𝑟𝐿𝐿𝑟𝑟𝑇𝑇 𝐶𝐶𝑟𝑟𝐴𝐴𝑟𝑟𝐸𝐸𝐸𝐸 =_{𝑇𝑇𝑟𝑟𝑟𝑟𝑟𝑟𝑇𝑇 𝑅𝑅𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟𝐴𝐴}𝑇𝑇𝑟𝑟𝑟𝑟𝑟𝑟𝑇𝑇 𝐶𝐶𝑟𝑟𝐴𝐴𝑟𝑟𝐴𝐴𝐸𝐸𝐸𝐸 𝐸𝐸𝐸𝐸�𝛼𝛼1+ 𝛼𝛼2ln 𝑇𝑇𝑅𝑅 + � 𝛾𝛾𝑘𝑘ln 𝑤𝑤𝑘𝑘 2 𝑘𝑘=1 + 𝛿𝛿4𝑟𝑟𝑟𝑟𝑡𝑡𝐴𝐴� (7)

5. Data and descriptive statistics 5.1 Data

This study is using financial statement data from Bureau van Dijk’s Bankscope and the China Banking Regulatory Commission (CBRC). A panel of 118 banks is used which can be further divided into 5 State Owned Commercial Banks (SOCBs), 12 Joint Stock Commercial Banks (JSCBs), 88 City Commercial Banks (CCBs) and 12 Rural Commercial Banks (RCBs). The time period under consideration is 2003 until 2012, for which a total of 798 observations are used.

5.2 Descriptive statistics

Table 1 presents an overview of the main descriptive statistics. Firstly, the Equity/Total Assets ratio and the Z-score show a considerable improvement over time. The Equity/Total Assets ratio on average increases from 3.84 percent in 2003 to 6.18 percent in 2012. The Z-score also shows a

considerable increase over time, increasing from 2.03 times in 2003 to 3.74 times in 2012, showing that banks become more able to withstand shocks in the ROA.

The different independent variables that are included in this study show a different development of competition in China over the last decade. The C5-ratio has a steadily decreasing trend. In 2003 the five SOCBs owned 58 percent of total banking assets, while by 2012 the combined market share had decreased to just under 45 percent of total assets. According to the SCP paradigm the decline in market share of the largest banks in China indicates an increase in competition as smaller banks take over market share from the five SOCBs.

Table 1 – Descriptive statistics for dependent and independent variables, average per year Equity/TA Z-score C5-ratio Lerner-index

2003 3,841633 2,030075 0,580337 0,471897 2004 3,847869 2,028605 0,569059 0,47722 2005 4,400256 2,304475 0,560586 0,495065 2006 5,353535 2,46341 0,551454 0,496549 2007 5,557692 3,267867 0,536561 0,541498 2008 6,045632 3,583608 0,515825 0,571132 2009 6,396444 3,240732 0,513111 0,564476 2010 6,556875 3,611048 0,492043 0,612298 2011 6,999255 3,823271 0,476078 0,605261 2012 6,187179 3,740559 0,449327 0,608378

However, the non-structural measure, which is based on firm-specific data, shows a somewhat different picture of the development of competition in China, as is depicted in Table 2. The Lerner-index shows an increase from 0.471897 in 2003 to 0.608378 in 2012, which means that Chinese banks have been gaining market power to set prices. This indicates that price competition in China is declining. Furthermore, the magnitude of the Lerner-index is relatively high compared with Lerner-indices from other countries, which indicates that price competition is weak in China. Carbo et al. (2009), for example, find a mean Lerner-index for the EU of 0.16, which is considerably lower than China’s range of 0.47 to 0.61.

Looking more closely to the development of the output price and marginal costs provides an explanation of the development of the Lerner-index. Table 3 shows that the price of banking output, as measured by Total Revenues/Total Assets, increases considerably. The price charged by banks on average increases from 0.034121 in 2003 to 0.047812 in 2012. When looking at the price

development of the different bank types, the price of banking output also increases. However, when looking at the marginal costs, as depicted in Table 4, an identical increase is not observed, which explains why the Lerner-index shows an overall increase over time. The marginal costs overall are relatively stable, amounting to 0.01794 in 2003 and 0.018398 in 2012. Also, the marginal costs of the SOCBs are the lowest of all bank types and has declined further between 2003 and 2012, indicating that the SOCBs are among the most cost-efficient banks in China. This result is quite remarkable when one considers the huge amount of articles claiming that the SOCBs are the least efficient bank type. However, it can be explained by the fact that the deposit insurance scheme is only explicitly guaranteed for SOCBs, while it remains vague whether other banks are also covered under the scheme (Wong and Wong (2001)). This possibly gives the SOCBs an advantageous position vis-à-vis other banks and enables SOCBs to pay lower interest rates on deposits, which ultimately results in lower marginal costs.

Table 2 - Lerner-index, average per year All observations SOCB JSCB CCB RCB 2003 0,471897 0,518614 0,504069 0,453542 0,462707 2004 0,47722 0,54106 0,531143 0,465082 0,368815 2005 0,495065 0,530463 0,505029 0,501023 0,400964 2006 0,496549 0,569462 0,569338 0,495887 0,457356 2007 0,541498 0,644511 0,584549 0,518394 0,598005 2008 0,571132 0,640177 0,565612 0,551462 0,639216 2009 0,564476 0,631205 0,488144 0,569314 0,536495 2010 0,612298 0,679135 0,600987 0,59741 0,621572 2011 0,605261 0,688242 0,601937 0,593348 0,642646 2012 0,608378 0,699291 0,610121 0,593103 0,515744

Table 3 – Price of output, average per year All observations SOCB JSCB CCB RCB 2003 0,034121 0,033534 0,035521 0,03437 0,025423 2004 0,040961 0,03386 0,056335 0,03857 0,033517 2005 0,042216 0,03329 0,03621 0,046128 0,030089 2006 0,044714 0,036149 0,046941 0,046349 0,033424 12

2007 0,047616 0,043239 0,05524 0,047784 0,040772 2008 0,052233 0,044718 0,04885 0,054222 0,049302 2009 0,04021 0,033889 0,034975 0,042171 0,037746 2010 0,041573 0,03438 0,037442 0,041592 0,048591 2011 0,052223 0,039213 0,057743 0,052906 0,048019 2012 0,047812 0,043006 0,051389 0,046936 0,046685

Table 4 – Marginal cost of output, average per year All observations SOCB JSCB CCB RCB 2003 0,01794 0,016094 0,017127 0,018056 0,024478 2004 0,020526 0,015434 0,020748 0,021255 0,019591 2005 0,021372 0,014894 0,020979 0,022446 0,018947 2006 0,021031 0,015191 0,019137 0,022416 0,015875 2007 0,02117 0,015341 0,019385 0,0228 0,015194 2008 0,022489 0,01611 0,022095 0,024317 0,016859 2009 0,017082 0,01244 0,017336 0,017594 0,016036 2010 0,015662 0,010988 0,014609 0,016416 0,014452 2011 0,019324 0,012153 0,019287 0,020409 0,016117 2012 0,018398 0,01285 0,019674 0,018607 0,022623

6. Results and discussion 6.1 Results

Table 5 and 6 show the regression results with the C5-ratio and Lerner-index as independent variables and the Z-score and Equity/Total Assets ratio as dependent variables, controlling for various factors that are likely to influence financial stability. The first two columns have the C5-ratio as independent variable. For both regressions the C5-ratio has a negative relationship with financial stability. This result is in line with the competition-stability view as a higher degree of competition is related to safer banks. Furthermore, the effect competition has on financial stability is considerable, -45.1774 when Equity/Total Assets is used as dependent variable and -22.4793 when the Z-score is used as dependent variable, both results being significant at the one percent level.

The estimated effect that the Lerner-index has on financial stability, however, is more ambiguous. When the Z-score is used as dependent variable the coefficient has a positive sign, whereas with the Equity/Total Assets ratio as dependent variable the sign of the coefficient is

negative. With the Z-score as independent variable, the coefficient of the Lerner-index is 0.169748, which is significant at the one percent level but not very large in magnitude, indicating that the Lerner-index does not contribute much to financial stability. When Equity/Total Assets is regressed on the Lerner-index, the estimated coefficient is -0.4045, which is again significant at the one percent level but not very large in magnitude. Also when both the C5-ratio and the Lerner-index are included simultaneously in the regression, the same ambiguity remains. The estimated coefficients of both the Lerner-index and the C5-ratio remain more or less the same as compared with the regression results when only one of them is included.

Table 5 – Dependent variable: Z-score - Equation (1), (2) and (3) from section 4.1

C5-ratio -22,4793 *** -22,0031 *** 1,690672 1,716587 Lerner-index 0,169748 *** 0,099898 ** 0,047076 0,042482 ln (Total Assets) 0,012789 0,8618 *** 0,027001 0,080551 0,053036 0,080588 GDPgrowth 0,067512 *** 0,000233 0,064439 *** 0,017018 0,018083 0,016941

Liquid assets/Total assets -0,02766 1,869812 *** -0,18699

0,294457 0,284349 0,300844

Loans/Deposits -0,00317 -0,00751 *** -0,00364

0,002234 0,002472 0,002233

Net loans/Total assets 0,007206 0,017541 *** 0,005438

0,005481 0,006007 0,005458

Constant 13,83045 *** -7,57136 *** 13,54707 ***

1,804299 0,848489 1,814137

Table 6 – Dependent variable: Equity/Total Assets - Equation (1), (2) and (3) from section 4.1

C5-ratio -45,1774 *** -48,3691 *** 5,234174 5,361242 Lerner-index -0,4045 *** -0,55803 *** 0,139406 0,132694 ln (Total Assets) -1,05349 *** 0,66241 *** -1,17331 *** 0,249082 0,156993 0,251721 14

GDPgrowth 0,073998 -0,04777 0,093167 **

0,052419 0,053527 0,052889

Liquid assets/Total assets -2,61362 *** 2,514783 *** -2,00729 ***

0,9118 0,842029 0,939704

Loans/Deposits 0,005309 -0,00286 0,005652

0,006925 0,00732 0,006974

Net loans/Total assets -0,0119 0,020163 -0,00649

0,01691 0,017784 0,017046

Constant 41,2365 *** -2,53219 43,88621 ***

5,577773 2,510004 5,664399

6.2 Discussion

This study uses two different measures of competition, a structural measure and a non-structural measure, but argues that in the case of China the structural measure, or C5-ratio, is the most appropriate. The C5-ratio is the most appropriate measure of competition in China as it reflects the extensive restructuring of the banking sector in the last two decades and the lifting of most geographic and entry restrictions. It has been recognized that banking competition in China is largely competition on quality and service (Wong and Wong (2001)). Ferry (2007), for example shows that newer established banks have a competitive advantage over the older SOCBs. Their competitive advantage is that they are more responsive to financial needs and have mainly situated themselves in the faster-growing coastal regions, where there is a greater demand for financial services. Indeed, as is depicted in Table 1, the combined market share of the five largest banks has decreased from 58 percent in 2003 to 45 percent in 2012, while the market shares of the JSCBs and CCBs has increased from 5.1% and virtually zero in 1994 to 17.6% and 9.2% in 2012 respectively.

The Lerner-index would be a less appropriate measure of competition as competition on prices is largely absent in China. In China there exists a legal ceiling on deposit rates and hence banks cannot increase deposit rates above a certain level to attract more customers. According to Anderson (2009) the current ceiling on deposit rates is repressing interest rates and without the ceiling deposit rates would be much higher. Indeed, the Lerner-index, as is depicted in Table 2, ranges between 0.47 in 2003 to 0.61 in 2012, which is high as compared to other countries, indicating that price competition is weak in China.

The regression results show that the C5-ratio has a negative relation to financial stability, both when the Z-score and the Equity/Total Assets ratio are used as dependent variables. Furthermore, the estimated coefficients of the C5-ratio are considerable, amounting to -22,4793

when the Z-score is used as a dependent variable and -45,1774 when the Equity/Total Assets ratio is used as dependent variable, both results being significant at the one percent level. These result are maintained when both the C5-ratio and the Lerner-index are included simultaneously in the regression.

The results therefore support the competition-stability view as the increase in competition, as measured by the decrease of the C5-ratio, is positively related to financial stability. The theoretical explanation might be that because of competition on service and quality JSCBs and CCBs have been gaining market share at the expense of the older SOCBs. At the same time, because of the absence of price competition banks have also been able to maintain and even improve financial stability. Hellmann et al. (2000) concur that with increased competition, the presence of deposit interest rate controls improve financial stability. When deposit rates are restricted, banks are able to earn higher profits, which increases a bank’s charter value. Other literature that covers the Chinese financial system also point out that deposit rate controls protect the profit margins of Chinese banks (García-Herrero et al. (2006)).

The conclusion of this study, that is to say that China’s financial situation fits the competition-stability view, is somehow at odds with the conclusion of Lee and Hsieh (2013), who found a negative relationship between competition and risk in China between 1993 and 2007. However, this may largely be the result of the use of different measures for financial stability. Lee and Hsieh focus on the standard deviation of the Return On Assets (ROA) and the Return On Equity (ROE) as risk measures, which leads them to the conclusion that risk has increased in China. This study, as opposed to the study of Lee and Hsieh, focuses on the Equity/ Total Assets ratio and the Z-score, leading to the conclusion that financial stability has increased.

7. Conclusion

The Chinese financial system in the last three decades has undergone far-reaching reforms, thereby changing from a monobanking system to a banking system where a legion of banks are competing for market share. A whole strand of literature is devoted to analysing the effects such periods of liberalization have on financial stability. There are two different views on the impact competition has on financial stability, the competition-fragility view and the competition-stability view. It is the aim of this study to determine how competition affects financial stability in China, considering the period from 2003 until 2012.

The results of this study support the competition-stability view, as the increase in competition, as measured by the decrease in C5-ratio, has a positive effect on financial stability. The theoretical explanation might be that while competition on service and quality has increased, it is just because the lack in pricing competition that banks have been able to increase their profit

margins and therewith ultimately financial stability. This study also argues that in the case of China the C5-ratio would be a more appropriate measure of competition than the Lerner-index as price competition is largely absent.

Finally, as is common in studies on competition and financial stability, this study might have an important policy implication for China. The reforms in recent decades that allowed for the entry of new financial institution and resulted in the deterioration of the market shares of the once dominant SOCBs have at least not harmed financial stability in China. However, the absence of price competition may play an important role in safeguarding financial stability. Future reforms that would allow for more competition on prices should therefore be made cautiously so as not to have a detrimental effect on overall financial stability, with all its possibly dire consequences for the Chinese economy.

Appendix

Appendix 1. Market shares of SOCBs, JSCBs, CCBs and RCBs – 2003-2012 (Source: CBRC website)

Appendix 2. Estimation of Trans-logarithmic Cost Function for SOCBs, JSCBs, CCBs and RCBs

The coefficients in bold below are used to calculate the marginal cost function, where lnTA is the natural logarithm of Total Assets, lnw1 is the cost of capital, lnw2 is the cost of borrowed funds and time refers to the year, where 1=2003, 2=2004 etc.

SOCB Trans-logarithmic Cost Function Coefficient Standard Error T-value 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Market Shares

lnTA -2,66795 2,276611 -1,17 lnTA^2 0,230434 0,155573 1,48 lnw1 0,935489 0,799658 1,17 lnw2 0,064511 0,799658 0,08 lnw1*lnw1 0,032828 0,093976 0,35 lnw1*lnw2 -1,36E-16 (constrained) lnw2*lnw1 -9,06E-17 (constrained) lnw2*lnw2 -0,03283 0,093976 -0,35 lnTA*lnw1 -0,01148 0,04966 -0,23 lnTA*lnw2 0,01148 0,04966 0,23 time 0,919489 0,424752 2,16 time^2 0,014446 0,005026 2,87 time*lnw1 -0,01479 0,010389 -1,42 time*lnw2 -0,00034 0,016361 -0,02 time*lnTA -0,064 0,027932 -2,29 JSCB Trans-logarithmic Cost Function

Coefficient Standard Error T-value lnTA 1,172163 0,355094 3,30 lnTA^2 -0,03461 0,032905 -1,05 lnw1 1,010383 0,279762 3,61 lnw2 -0,01038 0,279762 -0,04 lnw1*lnw1 0,019672 0,014875 1,32 lnw1*lnw2 -1,36E-16 (constrained) lnw2*lnw1 -9,06E-17 (constrained) lnw2*lnw2 -0,01967 0,014875 -1,32 lnTA*lnw1 -0,00677 0,020229 -0,33 lnTA*lnw2 0,006773 0,020229 0,33 time 0,008019 0,125718 0,06 time^2 0,005636 0,003813 1,48 time*lnw1 -0,07294 0,01027 -7,1 time*lnw2 0,037186 0,009807 3,79 18

time*lnTA -0,01036 0,010147 -1,02 CCB Trans-logarithmic Cost Function

Coefficient Standard Error T-value lnTA 1,740146 0,185322 9,39 (lnTA)^2 -0,1129 0,020653 -5,47 lnw1 0,436139 0,183354 2,38 lnw2 0,563861 0,183354 3,08 lnw1*lnw1 0,02364 0,01435 1,65 lnw1*lnw2 -1,36E-16 (constrained) lnw2*lnw1 -9,06E-17 (constrained) lnw2*lnw2 -0,02364 0,01435 -1,65 lnTA*lnw1 0,019045 0,012642 1,51 lnTA*lnw2 -0,01904 0,012642 -1,51 time -0,23874 0,047306 -5,05 time^2 0,002199 0,002007 1,1 time*lnw1 -0,01235 0,006068 -2,04 time*lnw2 -0,00222 0,004643 -0,48 time*lnTA 0,022999 0,004868 4,72 RCB Trans-logarithmic Cost Function

Coefficient Standard Error T-value lnTA -1,10328 0,593344 -1,86 lnTA^2 0,177019 0,060149 2,94 lnw1 1,111876 0,524875 2,12 lnw2 -0,11188 0,524875 -0,21 lnw1*lnw1 0,071061 0,03374 2,11 lnw1*lnw2 -1,36E-16 (constrained) lnw2*lnw1 -9,06E-17 (constrained) lnw2*lnw2 -0,07106 0,03374 -2,11 lnTA*lnw1 0,001654 0,034879 0,05 19

lnTA*lnw2 -0,00165 0,034879 -0,05 time 0,423573 0,125544 3,37 time^2 0,013262 0,003152 4,21 time*lnw1 -0,01438 0,011496 -1,25 time*lnw2 0,006976 0,008018 0,87 time*lnTA -0,04185 0,010384 -4,03

Bibliography

Anderson, J. (2009). The China Monetary Policy Handbook. In J. R. Barth, J. A. Tatom, & G. Yago, The

Milken Institute Series on Financial Innovation and Economic Growth (pp. 167-265). New

York: Milken Institute.

Ariss, R. T. (2010). On the implication of market power in banking: Evidence from developing countries. Journal of banking & Finance, 34, 765-775.

Bailey, W., Huang, W., & Yang, Z. (2011). Bank Loans with Chinese Characteristics: Some Evidence on Inside Debt in a State-Controlled Banking System. Journal of financial and Quantitative

Analysis, 46(6), 1795-1830.

Beck, T., Kunt, A. D., & Levine, R. (2006). Bank concentration, competition, and crises: First results.

Journal of Banking & Finance, 30, 1581-1603.

Berger, A. N., Hasan, I., & Zhou, M. (2009). Institutional Development, Ownership Structure, and Business Strategies: A New Era in the Chinese Banking Industry. In J. R. Barth, J. A. Tatom, & G. Yago, The Milken Institute Series on Financial Innovation and Economic Growth (pp. 369-401). New York: Milken Institute.

Berger, A. N., Klapper, L. F., & Turk-Ariss, R. (2009). Bank Competition and Financial Stability. J Financ

Serv Res, 35, 99-118.

Boyd, J. H., & Nicoló, G. D. (2005). The Theory of Bank Risk Taking and Competition Revisited. The

Journal of Finance, 60(3), 1329-1343.

Carbo, S., Humphrey, D., Maudos, J., & Molyneux, P. (2009). Cross-Country Comparisons of Competition and Pricing Power in European Banking. Journal of International Money and

Finance, 28, 115-134.

Casu, B., & Girardone, C. (2006). Bank competition, concentration and efficiency in the single European market. The Manchester School, 74, 441-468.

Dobson, W., & Kashyap, A. K. (2006). The Contradiction in China's Gradualist Banking Reforms.

Brookings Papers on Economic Activity, 2, 103-162.

Ferri, G. (2009). Are New Tigers supplanting Old Mammoths in China’s banking system? Evidence from a sample of city commercial banks. Journal of Banking & Finance, 33, 131-140.

Fu, X., Lin, Y., & Molyneux, P. (2014). Bank Competition and Financial Stability in Asia Pacific. Journal

of Banking and Finance, 38, 64-77.

García-Herrero, A., & Santabárbara, D. (2004). Where is the Chinese Banking System Going With the Ongoing Reforms. Banco de Espana Documentos Ocasionales. N.º 0406, 1-44.

García-Herrero, A., Gavilay, S., & Santabárbara, D. (2006). China’s Banking Reform: An Assessment of its Evolution and Possible Impact. CESifo Economic Studies, 52(2), 304-363.

Hellmann, T. F., Murdock, K. C., & Stiglitz, J. E. (2000). Liberalization, Moral Hazard in Banking, and Prudential Regulation: Are Capital Requirements Enough? American Economic Review, 90(1), 147-165.

Jimenez, G., Lopez, J. A., & Saurina, J. (2007). How does competition impact bank risk-taking?

Working Paper - Federal Reserve Bank of San Fransisco, 1-36.

Keeley, M. C. (1990). Deposit Insurance, Risk, and Market Power in Banking. The American Economic

Review, 80(5), 1183-1200.

Koons, C. (2013, December 3). Skepticism on China's Nonperforming Loans. The Wall Street Journal. Lee, C.-C., & Hsieh, M.-F. (2013). Beyond Bank Competition and Profitability: Can Moral Hazard Tell

Us More? Journal of Financial Services Research, 44, 87-109.

Li, T. (2009). China’s Nonperforming Loans: A $540 Billion Problem Unsolved. In J. R. Barth, J. A. Tatom, & G. Yago, China’s Emerging Financial Markets: Challenges and Opportunities (pp. 403-422). New York: Springer New York.

Lin, J. Y., Cai, F., & Li, Z. (2003). The China Miracle: Development Strategy and Economic Reform. Hong Kong: The Chinese University Press.

Liu, H., Molyneux, P., & Nguyen, L. H. (2012). Competition and risk in South East Asian commerical banking. Applied Economics, 44(28), 3627-3644.

Martin, M. F. (2012). China’s Banking System: Issues for Congress. Congressional Research Service -

CRS Report for Congress, 1-47.

Peng, Y. (2007). The Chinese Banking Industry: Lessons from history for today's challenges. New York: Routledge.

Schaeck, K. (2009). Bank Market Structure, Competition, and Stability: Issues and Concepts. In P. Alessandrini, M. Fratianni, & A. Zazzaro, The Changing Geography of Banking and Finance (pp. 133-153). New York: Springer.

Schaeck, K., & Cihak, M. (2007). Banking Competition and Capital Ratios. IMF Working Paper, 1-40. Wong, R. Y., & Wong, S. M. (2001). Competition in China's Domestic Banking Industry. Cato Journal,

21(1), 19-41.

Xu, B., Rixtel, A. v., & Leuvensteijn, M. v. (2013). Measuring bank competition in China: a comparison of new versus conventional approaches applied to loan markets. BIS Working Papers No 422, 1-62.

Yuan, Y. (2006). The state of competition of the Chinese banking industry. Journal of Asian

Economics, 17, 519-534.