Efficiency, growth and market power in the banking industry: New approach to efficient structure hypothesis

North American Journal of Economics and Finance 42 (2017) 531–545

Contents lists available at ScienceDirect

North American Journal of Economics and Finance j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / e c o fi n

Efficiency, growth and market power in the banking industry: New approach to efficient structure hypothesis Habib Hussain Khan a, Ali M. Kutan b,⇑, Iram Naz c, Fiza Qureshi d a

Faculty of Business & Accountancy, University of Malaya, Malaysia Department of Economics and Finance, Southern Illinois University Edwardsville, USA c Faculty of Management Sciences, Capital University of Science and Technology, Islamabad, Pakistan d Institute of Business Administration, University of Sindh, Jamshoro, Pakistan b

a r t i c l e

i n f o

Article history: Received 10 April 2017 Received in revised form 2 August 2017 Accepted 14 August 2017

JEL classification: G01 G21 G28 Keywords: Efficient structure hypothesis Cost efficiency Bank growth Bank concentration Data Envelopment Analysis

a b s t r a c t We extend the work of Homma, Tsutsui, and Uchida (2014) to provide empirical evidence on nexus of relationships in efficient structure (ES) hypothesis. In this framework, we test causality from cost efficiency to bank growth and then from bank growth to market concentration. We apply this approach to banking industry in Association of South East Asian (ASEAN) over the period of 1999–2014. The efficiency scores have been estimated by employing Slack Based Measurements Data Envelopment Analysis (SMB DEA). We apply Two-step system Generalized Method of Moments (GMM) and Panel Vector Auto Regression (PVAR) to account for endogeneity in estimation models. The results show that cost efficiency enables the banks to grow and obtain higher market share. The resultant growth then leads to higher market concentration/bank market power. There is also some evidence to support for quiet life (QL) hypothesis. Therefore, both ES and QL hypotheses may coexist in ASEAN banking industry. Ó 2017 Elsevier Inc. All rights reserved.

1. Introduction The efficient structure (ES) hypothesis (Demsetz, 1973) suggests that efficient firms grow, obtain higher market share and become larger; consequently, the market becomes more concentrated. The policy implications of ES hypothesis thus follow that the concentrated markets are dominated by efficient firms/banks. Therefore, antitrust policies/anti-concentration measures can bring unwarranted distortions in the market. The traditional approach to test ES hypothesis has not been very convincing. For instance, empirical studies relate firm/bank efficiency to firm/bank performance to test ES hypothesis [see for example (Berger, 1995; Berger & Hannan, 1989; Smirlock, 1985; Smirlock, Gilligan, & Marshall, 1984; Weiss, 1974)]. However, Homma, Tsutsui, and Uchida (2014) argue that ES hypothesis is a composite hypothesis that predicts stages of causal relationships form efficiency to firm growth and then to market structure (concentration/market power). In their study, Homma et al. (2014) test ES hypothesis through a causal relationship from efficiency to firm growth, but they do not test subsequent causal relationship from growth to market structure. ⇑ Corresponding author. E-mail addresses: [email protected] (H.H. Khan), [email protected] (A.M. Kutan), [email protected] (I. Naz), [email protected] (F. Qureshi). http://dx.doi.org/10.1016/j.najef.2017.08.004 1062-9408/Ó 2017 Elsevier Inc. All rights reserved.

532

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545

In this study, we extend the work of Homma et al. (2014) and test ES hypothesis considering all relationships in efficiency-growth-market structure nexus. Unlike earlier literature, we examine the causality from efficiency to growth and then from growth to market structure. In comparison to the methodology employed by existing studies, this is more direct approach to test ES hypothesis. We apply the proposed methodology to commercial banks in Association of South East Asian (ASEAN) region over the period of 1999–2014. In doing so, we also apply a non-structural measure of market structure i.e. Lerner Index in addition to the traditional concentration indices i.e. Hirschman Herfindahl Index (HHI) and Concentration Ratio (CR). Application of Lerner Index enables us to overcome the criticism on use of traditional concentration indices for their inability to assess the level of competition. Our study contributes to banking literature in several ways. First, although, this new test was proposed by Homma et al. (2014), our study is first to put complete set of relations into practice using empirical data. Second, we consider the ASEAN economies where the banking industry is moving towards more concentration. For instance, bank concentration in ASEAN – as represented by five bank concentration ratio (CR5) and Hirschman Herfindahl Index (HHI), increased from 0.55 and 0.11 in 1999 to 0.82 and 0.23 in 2014. Similarly, the bank profitability – as represented by return on assets (ROA) and return on equity (ROE) increased from 4% and 9% in 1999 to 17% and 14% respectively in 2014.1 These developments have important implication for policy makers in context of market structure-efficiency relationship.2 The results from empirical analysis show that efficiency enables the banks to grow and obtain higher market share. The resultant growth in banks’ loans, assets, and deposits leads to higher concentration/market power. The results are robust to alternative measures of growth and market structure variables and the estimation techniques. We also find some support for coexistence of efficient structure and quiet life hypothesis. Rest of the study is structured as follows: Section 2 discusses the methodological issues in literature concerning ES hypothesis. Section 3 outlines the methodology and construction econometric model for analysis of ES hypothesis. Section 4 reports the estimation results and their discussion. Finally, Section 5 concludes the study with discussion on policy implications and direction for future research. 2. Literature review The ES hypothesis is considered as alternative to the structure conduct performance (SCP) hypothesis which suggests that concentration of market share promotes collusive behavior among market players and allows them to earn abnormal profits through monopoly pricing. Although, both SCP and ES predict a positive relationship between concentration and profitability, the underlying mechanism is different under these hypotheses. Accordingly, both SCP and ES recommend contradictory policy measures. For instance, the SCP hypothesis favors antitrust/anti-concentration policies, whereas, the ES hypothesis suggests that such policies may bring inefficiency in the market. The empirical studies on ES are not separable from those on SCP because most of the time they are tested simultaneously. Therefore, most of these studies correspond to both SCP and ES hypotheses. In this section, we review some relevant literature with focus on methodology applied to test ES hypothesis. Earlier studies directly regress profitability measures on market share along with concentration to test ES and SCP hypothesis. In these studies, market share is considered as proxy for relative efficiency of firms. If market share has positive effect on firms’ profit, it supports the ES hypothesis but if market concentration is positively related to firms’ profit, the SCP hypothesis is supported (Berger, 1995; Berger & Hannan, 1989; Smirlock, 1985; Smirlock et al., 1984; Weiss, 1974). Subsequently, a large number of studies applied a similar methodology to test ES and SCP hypotheses and found mixed evidence.3 Although this approach is simple and easy to apply, it has some technical drawbacks. For instance, market share as an indicator of relative efficiency is not logical. Moreover, it is very uncertain that relationship between market share and profitability actually supports ES hypothesis. In this regard, a very influential article by Demsetz (1973) suggests that correlation between market concentration/market share and profitability is not enough evidence to support either ES or SCP hypothesis. Moreover, Shepherd (1986) argues that market share can be an indicator of market power and if it is related to profitability then it supports SCP instead of ES hypothesis. Another approach is proposed by Berger and Hannan (1989) who use interest rate paid on deposits instead of profitability to test SCP and ES hypotheses. The argument is that, the relationship between market concentration and profit is similar (positive) for SCP and ES. Nonetheless, the SCP and the ES differ in their implications regarding concentration-price relationship. For instance, the SCP hypothesis predicts that firms in a concentrated market have higher monopoly power and they can set higher prices. On the other hand, the ES hypothesis suggests that efficient firms dominate the concentrated market 1 The values of CR5, HHI, ROA and ROE represent the yearly averages for five ASEAN countries – i.e. Indonesia, Malaysia, Philippines, Singapore and Thailand. Data on CR5 has been extracted from Global Financial Development Database (GFDD), World Bank. The data for calculation of HHI and that on ROA and ROE has been extracted from BankScope provided by Fitch-IBCA. 2 Apart from the historical rise in bank concentration, the liberalization of the banking markets under the Banking Integration Framework (BIF) will allow the banks in ASEAN to operate across the member countries. Although BIF is expected to increase the level of competition among regional banks, it will lead to more consolidations as each country would like to fortify the domestic banks. 3 The primary methodology in these studies is similar but they sometimes use alternative measures of profitability such as Tobin’ q, net interest and interest rates. Some of these studies include Gale and Branch (1982), Smirlock et al. (1984), Shepherd (1986), Smirlock, Gilligan, and Marshall (1986), Stevens (1990), Smirlock (1985), Lloyd-Williams, Molyneux, and Thornton (1994), Samad (2008), Bhatti and Hussain (2010), Rhoades (1985), Evanoff and Fortier (1988), Martin (1988) among others.

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545

533

and set lower prices.4 It is more likely that, higher prices are related to other characteristics of market structure – i.e. product differentiation and research and development, instead of concentration. For instance, the product differentiation view suggests that firms with well differentiated and high quality may charge higher prices and earn high profits (Mueller, 1983; Ravenscraft, 1983; Ravenscraft, 1984; Shepherd, 1982). Similarly, the negative relationship between concentration and prices may not be the necessary condition to supports ES hypothesis. Homma et al. (2014) argue that it could be norm for efficient firms to set lower prices in short run to compete in the market, but the efficient firms may also set higher prices and enjoy monopoly profits if efficient firms enjoy a competitive advantage over their rivals. Few other studies that use a similar methodology include Bourke (1989), Molyneux and Thornton (1992) and Brewer and Jackson (2006). A little improvement in methodology is associated with Berger (1995) and Berger and Hannan (1997) who use market concentration, market share and measures of efficiency (X-efficiency and Scale-efficiency) to explain banks’ profitability. They also employ two additional regressions examining the causality from cost efficiency to market share and market concentration. A significant coefficient on market concentration is enough to validate SCP hypothesis. For ES hypothesis, the efficiency has to be positively related to profitability, market share and market concentration.5 Although, the approach used by Berger (1995) and Berger and Hannan (1997) does provide some improvements but it is not perfect. For instance, keeping in mind how these hypotheses are originally defined, it not logically clear that the effect of cost efficiency on profitability, concentration and market share supports the ES hypothesis. There are other studies that apply non-structural structure measures of market structure – i.e. Lerner Index and PanzarRosse (PR) H-statistics, and relate them profitability or efficiency.6 These studies however follow the traditional way of relating market structure measures to profitability and/or efficiency. Similarly, Casu and Girardone (2006, 2009) use multiple econometric techniques i.e. vector auto regression (VAR) and Granger causality, to test causalities among concentration, competition (Lerner index) and efficiency. Nonetheless, they do not use growth variable in their analysis to examine its role in ES hypothesis. Homma et al. (2014) in a recent study argue that the legitimate test of ES hypothesis should examine the causality from efficiency to growth and then from growth to concentration. However, Homma et al. (2014) only test the relationship from efficiency to growth. In this study we extend the work of Homma et al. (2014) and provide empirical evidence on nexus of all relationship in ES hypothesis i.e. from efficiency to growth and then from growth to concentration/market power. Our approach differs from Homma et al. (2014) in sense they only focus on causal relationship from efficiency to growth and do not consider complete nexus of relationship in the ES hypothesis i.e. efficiency to growth and then from growth to concentration. 3. Methodology We extend the work of Homma et al. (2014) to study the causality from efficiency to growth and then from growth to bank market structure. Accordingly, the empirical models are developed to identify the indirect relationship from cost efficiency to market structure through bank growth. These models are discussed in the following section. 3.1. Empirical models We follow an approach proposed by Baron and Kenny (1986) to identify the indirect relationship (mediating relationship) among variables of interest. Under this approach, if cost efficiency causes banks to grow and the resultant growth subsequently causes the banks to gain more market power, then cost efficiency, growth and market structure should be related in such a way that: (a) cost efficiency significantly affects growth, (b) growth significantly affects market structure, (c) cost efficiency affects market structure in absence of growth, and (d) the effect of efficiency on market structure reduces with inclusion of growth variable in the model. Accordingly, Eqs. (1), (2) and (3) have been developed to empirically test these relationships:

Growthi;j;t ¼ x0 þ x1 Efficiencyj;t1 þ km

n n X X X i;j;t þ sk Z j;t þ ei;j;t m¼1

MSj;t ¼ x0 þ x1 Efficiencyj;t1 þ km

n X m¼1

X i;j;t þ sk

ð1Þ

k¼1 n X Z j;t þ ei;j;t

ð2Þ

k¼1

4 This approach is criticized by Jackson (1992) for linearity assumption in price-concentration relationship because he finds that the negative priceconcentration relationship is not consistent over the full range of observed market concentration values. 5 There are several other studies that apply methodology proposed by Berger (1995) and Berger and Hannan (1997). These include Molyneux and Forbes (1995), Goldberg and Rai (1996), Maudos (1998), Mendes and Rebelo (2003), Papadopoulos (2004), Papadopoulos (2004) Park and Weber (2006), Tregenna (2009) Al-Muharrami and Matthews (2009), Seelanatha (2010) and Hsieh and Lee (2010) among others. For more recent use of similar or alternative methodologies, see, among others, Chan, Koh, and Kim (2016), Zhu (2016), and Chao, Yu, Lee, and Hsiao (2017). 6 These studies include Calem and Carlino (1991), Shaffer and DiSalvo (1994), De Bandt and Davis (2000), Bikker and Haaf (2002), Pilloff and Rhoades (2002), Coccorese (2009), Berger, Rosen, and Udell (2007), Turk Ariss (2010), Soedarmono and Tarazi (2016), and Kim, Park, and Song (2016), among others.

534

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545

MSj;t ¼ x0 þ x1 Efficiencyj;t1 þ x2 Growthi;j;t þ km

n X

X i;j;t þ sk

m¼1

n X Z j;t þ ei;j;t

ð3Þ

k¼1

where Efficiencyi;t1 is the cost efficiency bank ‘‘i” at time t-1; subscript t-1 shows that efficiency is assumed to affect firm profitability with one year lag. Growthi;t1 , refers to growth in loans and assets bank ‘‘i” at time t-1. MSj;t1 is market structure of banking sector for country ‘‘j” at time t-1. Xi;t , is vector of bank level controls; ‘‘Zj;t ” is vector of country level control variables and term ‘‘i;t ” in all the equations is random error term. We further verify the indirect effect of cost efficiency on market structure by applying the statistical tests introduced by Goodman (1960), Sobel (1982), MacKinnon and Dwyer (1993) and MacKinnon, Warsi, and Dwyer (1995). This approach follows the computation of test score (z-value) to check the significance/insignificance of indirect (mediation) relationship. The test scores are calculated as follows:

  z ¼ a  b=SQRT b2  SE2a þ a2  SE2b

Sobel Statistics ) Aroian Statistics )

  z ¼ a  b=SQRT b2  SE2a þ a2  SE2b þ SE2a  SE2b

Goodman Statistics )

  z ¼ a  b=SQRT b2  SE2a þ a2  SE2b  SE2a  SE2b

ð4Þ ð5Þ ð6Þ

where a is the coefficient on independent variable (IV) when the mediating variable (MV) is regressed on the IV, SEa is the standard error of a. The b is the coefficient on MV when the dependent variable (DV) is regressed on both IV and MV, while SEb is the standard error of b. The null hypothesis underlying each test is that the indirect effect is not significantly different from zero. Additionally, we also employ Panel Vector Auto Regressive (PVAR) in generalized method of moments (GMM) environment (Abrigo & Love, 2016) to cost efficiency, bank growth and market structure variables for robustness of findings. The empirical model in PVAR is given as under:

Growthi;t ¼ a3 þ

2 2 2 X X X b5 Growthi;t1 þ b6 Efficiencyi;t1 þ c3 MSi;t1 þ e3i;t i¼1

MSi;t ¼ a1 þ

n X

b1 MSi;t1 þ

i¼1

Efficiencyi;t ¼ a2 þ

i¼1

n n X X b2 Growthi;t1 þ c1 Efficiencyi;t1 þ e1i;t i¼1

ð8Þ

i¼1

n n 2 X X X b3 Efficiencyi;t1 þ b4 MSi;t1 þ c2 Growthi;t1 þ e2i;t i¼1

ð7Þ

i¼1

i¼1

ð9Þ

i¼1

Eqs. (7)–(9) show the system of equations where each variable i.e. bank growth, market structure and cost efficiency is represented as a function of its own lags and the lags of other variables. 3.2. Variables of the study 3.2.1. Cost efficiency Following Chan, Koh, Zainir, and Yong (2015), we use Slack Based Measures DEA (SBM DEA) to estimate efficiency scores. The SBM DEA is preferable over traditional DEA for two reasons. First, SBM DEA considers input and output slacks which are ignored by traditional DEA. Ignoring input and output slacks makes it difficult to compare efficient decision making units (DMUs) that use operational slacks. Second, SBM DEA assumes that DMUs can simultaneously optimize inputs and outputs to increase efficiency. However, DEA assumes that DMUs can opt for either inputs or outputs to create efficient frontier. If there are ‘‘n” DMUs with ‘‘m” inputs vectors of X ¼ ðxij Þ 2 Rmn , and ‘‘s” outputs vectors of Y ¼ ðyij Þ 2 Rsn , then the efficiency equation for DMUs using SBM DAE is as follows:

Min q ¼

P 1  1=m m s =xi0 Psi¼1 þi 1  1=s i¼1 si =yi0

Subject to x0 ¼ Xk þ s ; i ¼ 1; . . . ; m y0 ¼ Yk þ sþ ; i ¼ 1; . . . ; s k P 0; j ¼ 1; . . . ; n s P 0; i ¼ 1; . . . ; m sþ P 0; i ¼ 1; . . . ; s

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545

535

where q s , sþ represent the efficiency, input slack and output slack respectively. The value of q ranges between 0 and 1, a value of 1 indicates the DMU is efficient and s , sþ are zero. For choice of inputs and outputs, we use intermediation approach following Altunbas, Evans, and Molyneux (2001), Casu, Girardone, and Molyneux (2004) and Chan et al. (2015). The inputs and outputs used for SBM DEA are selected following Sturm and Williams (2004), Avkiran and Thoraneenitiyan (2010) and Chan et al. (2015). Accordingly, the inputs include personnel expenses, interest expenses and non-interest expenses. The outputs consist of interest income from loans, and investments, fee income from off balance sheet items and other non-interest income. 3.2.2. Growth The growth variable represents the growth in banks’ market share. According to ES hypothesis, cost efficiency enables banks to increase their share of assets, loans and deposits. Following Homma et al. (2014), growth is measured as the annual percentage change in banks’ assets and loans. As an addition robustness check, we also considers growth in deposits, which is measured as annual percentage change in banks’ deposits. 3.2.3. Market structure We use the Lerner index (Lerner, 1934) as main measure of market structure. However, we also use two concentration indices – i.e. CR5 and HHI for robustness check. The choice of Lerner index is based on two important considerations. First, the Lerner index is calculated at bank level and provides observation to observation match with other bank level variables. Second, few earlier studies i.e. Casu and Girardone (2006) and Casu and Girardone (2009), have used the Lerner index to test causality between cost efficiency and market structure. In this sense, our results can be compared with those studies. The Lerner index represents the ratio of mark up7 to output price i.e. L = (P-MC)/MC, where ‘‘P” and ‘‘MC” are the price of the output and the marginal cost of producing an additional unit of output respectively.8 The Lerner Index ‘‘ranges from 0 to 1. The higher values of Lerner indicate more market power and less competitive conditions. The marginal cost can be obtained from following translog cost function:

ln Cost i;t ¼ b0 þ b1 ln Q i;t þ þ

2 X

3 3 3 X 3 X X b2 1X ck;t ln W k;i;t þ uk ln Q i;t ln W k;i;t þ di;j ln W k;i;t ln W j;i;t ln Q 2it þ 2 k¼1 j¼1 2 k¼1 k¼1

gk trendk þ

k¼1

3 X

xi ln W j;i;t trend þ v ln Q t;j trend þ ei

ð10Þ

i¼1

where Cost i;t and Q i;t represent the total cost and output for bank ‘‘i” in time ‘‘t” respectively, and W 1 , W 2 and W 3 are the input prices of deposit funds, labor and capital. The marginal cost is the first derivative of the cost function with respect to the level of output as represented by Eq. (11):

MC i;t ¼

" # 3 X Cost i;t b1 þ b2 ln Q i;t þ hk ln W k;i;t þ d3 Trendi;t Q i;t k¼1

ð11Þ

Once the marginal cost is estimated, it is used to calculate the Lerner Index for individual banks through the formula L = (P-MC)/MC. The conventional Lerner Index has been criticized for its profit and cost efficiency assumptions. For instance, conventional Lerner Index assumes that banks are able to achieve full efficiency. Therefore, it fails to consider the possibility that banks may not fully exploit the opportunities to price their output because of their market power (Koetter, Kolari, & Spierdijk, 2012). To address these issues, we also calculate the adjusted Lerner index following the procedure laid down in Koetter et al. (2012) using Eq. (12).9

LernerðAdjustedÞ ¼

pi þ tci  mci  qi pi þ tci

ð12Þ

where pi , tci , mci and qi respectively show the profit, total cost, marginal cost and output of bank ‘‘i”. The value of the adjusted Lerner Index also ranges between 0 and 1 (like the conventional Lerner) with higher values implying more market power. 3.2.4. Other variables Following earlier studies – i.e. Gardener, Molyneux, and Nguyen-Linh (2011), Homma et al. (2014), Chan et al. (2015, and Chavarín (2016)), we also include in the estimation models several bank level and macroeconomic level variables to control cross sectional differences among banks and countries. The bank level variables include bank profits, bank size, bank capitalization, bank z-score and ownership structure. Macroeconomic factors include GDP, inflation, stock market development, regulatory framework and institutional characteristics. 7

Mark up is the difference between output price and marginal cost. Output of a banking firm is represented by total assets, therefore the price of total assets is calculated as total revenue/total assets. 9 Issues with the conventional approach to calculate the Lerner Index and the calculation of the efficiency adjusted Lerner Index are discussed in detail in Koetter et al. (2012). 8

536

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545

Table 1 Variables, definition/description and sources. Variables

Description/Definition

Sources

Cost Efficiency

Efficiency scores based on data envelopment analysis (DAE). See Section 3.2.1 for more details. The ratio of mark up (difference between output price and marginal cost) to output price. Its value ranges from 0 to 1 with higher values indicating more market power and less competition. Conventional Lerner Index adjusted for efficiency. See Section 3.2.3 for more details. Total assets held by five largest banks of a country to the total assets of all banks in that country. Sum of squared market shares of all the banks in a country in a year.

Authors’ Calculations

Conventional Lerner Index

Adjusted Lerner Index Five Bank Concentration Ratio (CR5) Hirschman Herfindahl Index (HHI) Growth in Loans Growth in Assets Growth in Deposits Return on Average Assets (ROAA) Net Interest Margin (NIM) Bank Capitalization Overhead Bank Size Z-Score Dummy (Merger) Dummy (Ownership) Inflation GDP Growth Stock Market Turnover Institutional Characteristics Regulatory Framework

Annual percentage increase in amount of total loans Annual percentage increase in amount of total assets Annual percentage increase in amount of total deposits Profit before tax scaled by average total assets The difference between the interest income generated by banks and the amount of interest paid out to their lenders divided by interest earning assets. Ratio of equity to total assets The ratio of overheads to total assets Natural log of total assets Z-score is an inverse proxy for the firm’s probability of failure i.e. Z = [ROA + EQ/ TA]/SD (ROA). Larger values of Z-score indicate less overall risk. Dummy variable for merged banks which equals 0 if bank merged during the sample period and 0 otherwise. Dummy variable, equals 1 if more than 51% share is held by foreign shareholders and 0 otherwise. Inflation based on consumer price index Inflation adjusted growth rate for GDP Total value of shares traded during the period divided by the average market capitalization for the period. Institutional Characteristics includes an overall index of economic freedom from Heritage Foundation. Regulatory Framework is an overall governance index based on indices of Rule of Law, Regulatory Quality, Government Effectiveness and Political Stability from World Governance Indicators provided by Kaufmann, Kraay, and Mastruzzi (2015).

Authors’ Calculations

Authors’ Calculations World Bank Authors’ Calculations BankScope BankScope BankScope BankScope BankScope BankScope BankScope BankScope BankScope BankScope BankScope World Bank World Bank World Bank Heritage Foundation World Governance Indicators

Note: The Table shows the variables of the study, their description/definition and the sources of data on these variables.

3.3. Sample and data The data on bank level variables (including those used in calculation of cost efficiency and Lerner Index) has been collected from Bank Scope – managed by Fitch International Bank Credit Analysis Ltd. (IBCA) – over the period of 1999– 2014. The data on country level variables has been extracted from Global Financial Development Database (GFDD) and World Governance Indicators (WGI) provided by the World Bank and Index of Economic Freedom compiled by Heritage Foundation. Table 1 reports the main variables, their description/definition and their sources. The sample consists of commercial banks from ASEAN—for numerous changes have taken place in structure of ASEAN banking industry after Asian financial crisis 1997–1998 and Global financial crisis 2008–2009 which have important implications for regulatory authorities i.e. central banks.10 The choice of commercial banks instead of all banks makes the group of observations more consistent and allows for better comparisons and efficiency estimates (Bikker, 1999; Punt & Van Rooij, 2003). We follow earlier studies – i.e. Turk Ariss (2010) and Arena, Reinhart, and Vazquez (2007), to filter the original sample of commercial banks. The filtering criteria includes dropping out banks that have less than five year consecutive observations, and those having missing data on important variables. Consequently, the final sample consists of unbalanced panel of 133 commercial banks with 1507 observations.11 Tables 3 and 4 summarize the descriptive analysis for overall sample and for individual countries respectively. The cost efficiency of banks for entire sample averages around 66.12% which is almost equivalent to 66.72% as determined by Chan et al. (2015). The average efficiency score for ASEAN indicates these banks can increase their efficiency by 33.88%. Table 3 indicate that commercial banks do not differ significantly in terms of cost efficiency across individual countries. However, Philippian banks are the most efficient in the region with average efficiency score of 73.1% which is concurrent with findings

10 Complete coverage of ASEAN is constrained by data unavailability, therefore, the sample has been restricted to five prime countries i.e. Malaysia, Indonesia, Singapore, Philippines and Thailand. 11 Distribution of the sample is presented in Table 2.

537

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545 Table 2 Number of banks per year. Years

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

No. of Banks

41

53

53

53

57

75

85

93

105

112

115

133

133

133

133

133

Table 3 Complete sample-descriptive analysis.

Cost Efficiency Lerner Index Adjusted Lerner Index CR5 HI Growth (Loan) Growth (Assets) Growth (Deposits) ROAA NIM Bank Capitalization Overhead Bank Size Z-Score Inflation GDP Growth Stock Market Institutional Characteristics Regulatory Framework

Mean

Median

S.D

Minimum

Maximum

0.661251 0.260296 0.217163 0.686652 0.130321 0.239137 0.453327 0.568885 0.010168 0.041377 0.125457 0.027812 7.859239 12.05041 0.050795 0.051257 3.795007 65.9875 57.07404

0.811635 0.260984 0.217776 0.666392 0.110641 0.209048 0.449218 0.550301 0.012125 0.036941 0.103505 0.024517 7.833244 4.946843 0.045404 0.055848 3.791618 11.74585 50.06193

0.337266 0.138627 0.123378 0.103585 0.053345 0.167764 0.318519 0.365957 0.048656 0.034251 0.106109 0.024724 1.837859 11.62284 0.037135 0.022941 0.522743 137.9649 22.94514

0.000735 0.05538 0.04529 0.452684 0.05533 0.09952 0.09906 0.19988 0.72445 0.23024 1.29214 0.000544 3.061047 2.90965 0.00846 0.0233 2.635872 89.40000 20.40199

1.00000 0.519565 0.462413 0.795745 0.436633 0.429447 0.781541 0.696096 1.199891 0.393541 0.854111 0.380405 12.66896 31.78235 0.204891 0.152404 4.804121 65.98750 97.98464

Note: The Table reports the descriptive account of the variables for overall sample.

of Gardener et al. (2011) and Chan et al. (2015). On the other hand, Indonesian and Malaysian banks have lowest mean score for cost efficiency i.e. Indonesia (59.7%) and Malaysia (67.2%). The average value of CR5 for entire sample is 68.66% which indicates that banking industry in ASEAN is highly concentrated. It may seem that sample average is skewed upward due to Singapore’s highest concentration ratio (97.9%), however, the average concentration in all other countries is relatively closer to the sample average. The statistics are grossly similar to those observed by Chan et al. (2015). The Lerner index which measures the market power has an average value of 0.2602 implying that banks in ASEAN do enjoy some level of market power. The country averages on Lerner index show that commercial banks in Singapore exercise more market power as compared to rest of the countries in the sample. This is expected under SCP hypothesis which predicts that banks in concentrated markets enjoy more market power. The average value of growth (loans) for total sample is 23.91% suggesting that there has been healthy lending activity in ASEAN from 1999 to 2014. Indonesia is at top in terms of growth in loan with average value of 25.4% followed by Philippines (17.4%), Thailand (16.4%), Singapore (14.3%) and Malaysia (9.4%). The country level statistics for growth in lending activity in ASEAN are consistent with those observed by Khan, Ahmad, and Gee (2016). Table 5 reports pairwise correlations among variables of the study. There are two important implications of these correlations. First, the correlations among main variables of the study i.e. efficiency, growth and market structure variables are significant. Second, the correlations among variables that will appear on right hand side of equation (independent) are not so high to create multicollinearity problem. Although, correlations among few variables are high but they are not problematic as they are alternative proxies and do not enter the estimation model together. 4. Results and discussion In order to test causality from efficiency to growth and then growth to efficiency, we follow the methodology proposed by Baron and Kenny (1986).Under this methodology, if cost efficiency affects bank growth and bank growth in return causes market structure, then following relationships must exist. First, the cost efficiency should affect growth; second, growth has to affect market structure; third, cost efficiency should affect market structure in absence of growth; and fourth, the effect of efficiency on market structure must reduce with inclusion of growth variable in the model. To estimate these relationships, we apply Two-step system GMM to Eqs. (1), (2) and (3) separately. We further verify the indirect relationship between cost efficiency and market structure by applying the test statistics proposed by Goodman (1960), Sobel (1982), MacKinnon and Dwyer (1993) and MacKinnon et al. (1995). The estimation results are discussed in the next section.

538

Variables

Efficiency Lerner Index Adj. Lerner Index CR5 HI Growth (Loan) Growth (Assets) Growth (Deposits) ROAA NIM Bank Capitalization Overhead Bank Size Z-Score Inflation GDP Growth Stock Market Institutional Character Reg. Framework

Indonesia

Malaysia

Philippines

Singapore

Thailand

Avg.

SD

Min

Max

Avg.

SD

Min

Max

Avg.

SD

Min

Max

Avg.

SD

Min

Max

Avg.

SD

Min

Max

0.597 0.190 0.169 0.636 0.108 0.254 0.356 0.360 0.015 0.050 0.117 0.032 7.058 2.434 0.074 0.056 0.486 54.988 31.862

0.369 0.054 0.048 0.054 0.027 0.279 0.257 0.601 0.010 0.031 0.115 0.018 1.448 1.842 0.030 0.007 0.132 2.579 6.554

0.001 0.042 0.038 0.576 0.080 0.128 0.098 0.291 0.001 0.230 1.292 0.002 5.006 0.025 0.043 0.045 0.315 51.900 20.402

1.000 0.255 0.227 0.766 0.172 0.981 0.902 1.767 0.034 0.171 0.642 0.273 9.484 3.700 0.131 0.063 0.713 61.500 44.185

0.672 0.324 0.288 0.767 0.159 0.094 0.235 0.259 0.012 0.030 0.118 0.014 8.341 13.978 0.021 0.052 0.324 3.064 67.178

0.320 0.171 0.152 0.100 0.040 0.214 0.183 0.185 0.004 0.017 0.075 0.007 1.374 1.854 0.008 0.019 0.062 59.900 3.187

0.003 0.012 0.011 0.453 0.055 0.324 0.094 0.195 0.004 0.000 0.000 0.001 6.395 11.825 0.010 0.005 0.227 69.600 61.714

1.000 0.520 0.462 0.890 0.220 0.591 0.842 0.932 0.018 0.139 0.679 0.069 10.203 18.069 0.036 0.074 0.446 58.381 73.338

0.731 0.278 0.259 0.668 0.128 0.174 0.341 0.316 0.014 0.052 0.136 0.043 7.871 18.237 0.046 0.053 0.213 2.524 40.075

0.294 0.115 0.103 0.068 0.033 0.259 0.245 0.281 0.007 0.061 0.077 0.049 1.298 4.780 0.016 0.020 0.031 54.700 4.567

0.036 0.023 0.020 0.571 0.080 0.299 0.081 1.000 0.003 0.072 0.060 0.012 5.972 12.559 0.030 0.011 0.162 62.500 33.649

1.000 0.386 0.343 0.875 0.198 0.872 0.997 1.610 0.027 0.394 0.640 0.380 9.831 30.079 0.083 0.076 0.261 87.700 51.166

0.699 0.457 0.440 0.977 0.301 0.143 0.478 0.877 0.010 0.021 0.127 0.014 9.689 20.736 0.029 0.062 0.761 0.818 94.790

0.312 0.069 0.061 0.011 0.039 0.168 0.575 0.898 0.004 0.013 0.052 0.008 2.062 5.423 0.020 0.046 0.269 86.100 2.089

0.013 0.038 0.034 0.964 0.269 0.161 0.190 0.288 0.002 0.003 0.043 0.007 6.523 6.266 0.006 0.006 0.404 89.400 91.238

1.00 0.554 0.526 0.996 0.437 0.522 0.982 1.472 0.015 0.069 0.351 0.064 12.272 25.871 0.065 0.152 1.220 64.794 97.985

0.706 0.189 0.168 0.662 0.108 0.164 0.308 0.488 0.006 0.030 0.142 0.022 8.790 2.556 0.026 0.042 0.883 2.115 51.466

0.012 0.013 0.011 0.001 0.000 0.273 0.148 0.265 0.012 0.001 0.008 0.001 1.385 1.187 0.014 0.026 0.172 54.988 6.956

0.006 0.887 0.789 0.046 0.012 0.153 0.938 0.338 0.021 0.121 0.871 0.098 6.325 1.120 0.003 0.001 0.642 2.579 44.346

1.000 0.455 0.405 0.640 0.102 1.031 0.999 1.000 0.021 0.023 0.017 0.004 10.506 0.761 0.046 0.077 1.154 51.900 63.447

Note: This Table shows country-wise descriptive analysis of the variables.

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545

Table 4 Country-wise descriptive analysis.

(1) Cost Efficiency (2) Lerner Index (3) Adj. Lerner Index (4) CR5 (5) HI (6) Growth (Loan) (7) Growth (Assets) (8) Growth (Deposits) (9) ROAA (10) NIM (11) Bank Capitalization (12) Overhead (13) Bank Size (14) Z-Score (15) Inflation (16) GDP Growth (17) Stock Market (18) Ins. Character (19) Reg. Quality

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

1 0.16** 0.24** 0.13** 0.17** 0.21** 0.11** 0.19** 0.41** 0.09* 0.07**

1 0.89** 0.42** 0.33** 0.31** 0.27** 0.23** 0.17** 0.39** 0.09**

1 0.27** 0.22** 0.21** 0.15** 0.17** 0.12** 0.31** 0.11**

1 0.95** 0.016** 0.10** 0.08** 0.05* 0.24** 0.04

1 0.09** 0.14** 0.11** 0.03** 0.21** 0.03*

1 0.77** 0.63** 0.48* 0.29* 0.31*

1 0.93** 0.29** 0.18** 0.13

1 0.17** 0.06* 0.03

1 0.25** 0.36**

1 0.28**

**

**

*

0.07 0.15** 0.12* 0.03* 0.07* 0.11 0.04* 0.08**

*

0.10 0.14** 0.31** 0.28* 0.10** 0.15** 0.05** 0.06**

0.09 0.16** 0.26** 0.27** 0.11** 0.13** 0.07** 0.05**

*

0.04 0.25** 0.15** 0.25** 0.09* 0.06* 0.08* 0.06*

*

0.20 0.22** 0.25** 0.18** 0.04 0.09** 0.07** 0.09*

*

0.12 0.24* 0.16* 0.13* 0.14* 0.06* 0.05* 0.11*

0.03 0.03** 0.07 0.04 0.11* 0.03* 0.06* 0.07*

0.02 0.25* 0.02 0.07 0.012* 0.04* 0.08* 0.13*

0.12 0.01 0.08** 0.08** 0.07** 0.09** 0.06** 0.10**

(11)

0.48 0.011** 0.12** 0.09** 0.13** 0.08** 0.12** 0.14**

(12)

(13)

(14)

(15)

(16)

(17)

(18)

(19)

1 0.17** 0.14** 0.16** 0.02 0.10** 0.07** 0.05**

1 0.25** 0.11 0.01 0.06** 0.11** 0.08**

1 0.18** 0.09** 0.08** 0.06** 0.15**

1 0.08** 0.04 0.05 0.03

1 0.12** 0.16** 0.09**

1 0.10** 0.13*

1 0.08**

1

1 0.02 0.34** 0.03 0.13** 0.05* 0.02 0.03 0.02

Note: This table shows pairwise correlations among important variables of the study. Subscripts **, * denote the significance of relationships at 5% and 10% levels respectively.

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545

Table 5 Pairwise correlations.

539

540

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545 Table 6 Cost efficiency and bank growth. Bank Growth in Terms of Loans, Assets and Deposits

Efficiency (t-1) ROA Bank Size Bank Capitalization Bank Z-Score Ownership Structure D(Merger) GDP Growth Inflation Stock Market Development D(Crisis) Regulatory Framework Institutional Characteristics Country Dummy Time Dummy AR(1) AR(2) Sargan/Hensen No. of Instruments No. of Groups

Model (1)

Model (2)

Model (3)

0.139*** (0.041) 0.064** (0.031) 0.125*** (0.041) 0.125** (0.057) 0.013** (0.005) 0.151** (0.073) 0.328** (0.161) 0.299** (0.143) 0.276 (0.329) 0.023* (0.012) 0.141** (0.066) 0.083** (0.038) 0.098** (0.047) Yes Yes 0.013 0.139 0.239 115 133

0.113*** (0.027) 0.127** (0.059) 0.097** (0.047) 0.122** (0.058) 0.016** (0.007) 0.156** (0.075) 0.339** (0.166) 0.306** (0.155) 0.089* (0.045) 0.031* (0.016) 0.096** (0.046) 0.061** (0.028) 0.072** (0.034) Yes Yes 0.027 0.127 0.247 115 133

0.183*** (0.051) 0.130** (0.062) 0.079** (0.038) 0.157** (0.069) 0.026** (0.012) 0.158** (0.076) 0.332** (0.163) 0.317** (0.166) 0.067* (0.034) 0.047* (0.024) 0.089** (0.043) 0.073** (0.034) 0.085** (0.041) Yes Yes 0.042 0.131 0.257 118 133

Note: The table shows the results from estimation of Eq. (1) where bank growth is regressed on bank efficiency and other control variables. Dependent variables are growth in loans (model 1), growth in assets (model 2) and growth in deposits (model 3). The coefficient are estimated using Two-step system GMM with Windmeijer (2005) corrected standard errors and small sample options. The diagnostics tests show that GMM is correctly specified and there are no identification issues. For instance, significant values of AR (1) indicate that null hypothesis of no autocorrelation among error terms in first difference is rejected. AR (2) is insignificant indicating that error terms in level regressions are not correlated. Values of Sargan/Hensen are insignificant indicating that instruments are valid. Corrected standard errors are reported in the parenthesis. Subscripts ***, **, * denote the significance of relationships at 1%, 5% and 10% levels respectively.

4.1. Cost efficiency, bank growth and market structure The estimation results for growth equation i.e. Eq. (1), are reported in Table 6. The growth variable is represented by growth in loans (Model 1), growth in assets (Model 2) and growth in deposits (Model 3). The coefficients on efficiency variable are significant and positive in all three models implying that cost efficiency allows banks to grow. This is in support of first part of ES hypothesis which states that efficient banks grow. This finding is consistent with that of Homma et al. (2014) who also find a positive impact of efficiency on bank growth. The behavior of coefficients on control variables is also according to the theory. For instance, profitability, bank size and bank capitalization have positive impact on bank growth. Moreover, banks with foreign ownership are likely to experience higher growth than domestic banks. The coefficient on financial crisis is negative indicating that bank growth has been slower during the global financial crisis of 2008–2009.12 The results for control variables are grossly similar to those from earlier studies such as Homma et al. (2014), Chan et al. (2015) and Khan et al. (2016). Table 7 reports the estimation results of market structure equation (Eq. (2)). The market structure variable is represented by conventional Lerner index (model 1), adjusted Lerner index (model 2), five-bank concentration ratio (model 3), and Hirschman Herfindahl index (model 4). The coefficients on cost efficiency are significant and positive in all models suggesting that cost efficiency leads to more market power/bank concentration. These findings are in contrast to Casu 12 For recent studies of financial crises on banking sector and financial development, see, among others, Ari and Cergibozan (2016), Laidroo (2016), Lee and Huang (2016), Mulder, Perrelli, and Rocha (2016), and Süer, Levent, and Sß en (2016).

541

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545 Table 7 Cost efficiency and market structure. Dependent Variables are Market Structure measures (Conventional Lerner Index, Adjusted Lerner Index, CR5 and HHI)

Efficiency (t-1) Number of Banks ROA (t-1) Bank Size (t-1) Inflation GDP D(Crisis) Ownership Structure Regulatory Framework Institutional Characteristics Country Dummy Time Dummy AR(1) AR(2) Sargan/Hensen No. of Instruments No. of Groups

Model (1)

Model (2)

Model (3)

Model (4)

0.031** (0.013) 0.051** (0.025) 0.019** (0.009) 0.031** (0.014) 0.009 (0.012) 0.229* (0.115) 0.227** (0.063) 0.062** (0.029) 0.041* (0.022) 0.077** (0.035) Yes Yes 0.024 0.193 0.267 99 133

0.041** (0.019) 0.036** (0.017) 0.022** (0.011) 0.034** (0.016) 0.016 (0.019) 0.231* (0.116) 0.211** (0.103) 0.067** (0.032) 0.046 (0.027) 0.069** (0.034) Yes Yes 0.016 0.233 0.356 99 133

0.029** (0.014) 0.017* (0.009) 0.018** (0.008) 0.039** (0.019) 0.028 (0.029) 0.206* (0.104) 0.235** (0.115) 0.061** (0.027) 0.059 (0.033) 0.065 (0.039) Yes Yes 0.031 0.227 0.326 99 133

0.026** (0.012) 0.026* (0.014) 0.015** (0.007) 0.041** (0.020) 0.025 (0.029) 0.189** (0.094) 0.197** (0.096) 0.058** (0.026) 0.051* (0.026) 0.063** (0.031) Yes Yes 0.027 0.241 0.328 103 133

Note: The table shows the results from estimation of Eq. (2) where market structure variables are regressed on bank efficiency and other control variables. Dependent variables are Conventional Lerner index (model 1), Adjusted Lerner (model 2), CR5 (model 3) and HHI (model 4). The coefficient are estimated using Two-step system GMM with Windmeijer (2005) corrected standard errors and small sample options. The diagnostics tests show that GMM is correctly specified and there are no identification issues. For instance, significant values of AR (1) indicate that null hypothesis of no autocorrelation among error terms in first difference is rejected. AR (2) is insignificant indicating that error terms in level regressions are not correlated. Values of Sargan/Hensen are insignificant indicating that instruments are valid. Corrected standard errors are reported in the parenthesis. Subscripts ***, **, * denote the significance of relationships at 1%, 5% and 10% levels respectively.

and Girardone (2006) and Casu and Girardone (2009) who do not find evidence of causality running from efficiency to market structure. The analysis provided in Tables 6 and 7 suggests that cost efficiency positively influences the bank growth and market power/concentration. However, as discussed in Section 3.1, bank growth should also significantly influence market power/ concentration in order for ES hypothesis to be valid. Additionally, the inclusion of growth variable in market structure regression (Eq. (3)) should reduce the coefficient on cost efficiency. Table 8 reports the results of analysis when market structure variables are regressed on both efficiency and bank growth. Panel A, B and C report the results for different proxies of bank growth. The dependent variable is market structure represented by Lerner index (column 1, 5, 9), adjusted Lerner index (column 2, 6, 10), CR5 (column 3, 7, 11) and HHI (column 4, 8, 12). There are two important findings from this analysis. First, the coefficients on growth variables are significant and positive across all the models. Second, the coefficients on cost efficiency are still significant but the magnitude of the coefficients is lower than those reported in Table 7. Although, there is no earlier evidence on causality running from growth to market structure, this study has provided empirical evidence on such a relationship. Taken together, the analysis reported in Tables 6–8 fulfills the criterion set for indirect relationship by Baron and Kenny (1986). At least for our sample, there are four important findings. First, the cost efficiency has significant positive impact on bank growth represented by loans, assets and deposits. Second, bank growth significantly contributes to bank market power and bank concentration. Third, cost efficiency is also related to increased market power and bank concentration. Fourth, the impact of cost efficiency on market power/bank concentration decreases in magnitude when bank growth variable is added to the estimation model. However, it is interesting that the coefficient on efficiency variable becomes very low but it does not become insignificant. This implies that there are some other variables that may also explain the relationship between efficiency and market structure. These findings provide sufficient support to the existence of the nexus of relationship in ES hypothesis as proposed by Homma et al. (2014). To further verify the indirect relationship between cost efficiency and market structure – i.e. from cost efficiency to bank growth and then from bank growth to market power/bank concentration – we follow a procedure introduced by Goodman (1960), Sobel (1982), MacKinnon and Dwyer (1993) and MacKinnon et al. (1995). This approach involves calculation of test statics using Eqs. (4), (5) and (6). The coefficients of cost efficiency based on Eq. (1) (Table 6) and bank growth based on

542

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545

Table 8 Bank efficiency, growth and market structure. Dependent Variables are Market Structure measures (Conventional Lerner Index, Adjusted Lerner Index, CR5 and HHI) Panel A: Loan Growth (1) Efficiency (t-1) Bank Growth (t-1) No. of Banks ROA (t-1) Bank Size (t-1) Inflation GDP D(Crisis) Ownership RFW IC Country Dummy Time Dummy AR(1) AR(2) Sargan/Hensen No. of Instruments No. of Groups

(2) **

0.032 (0.015) 0.047*** (0.016) 0.143 (0.136) 0.021** (0.009) 0.036** (0.017) 0.019 (0.012) 0.211* (0.106) 0.272** (0.133) 0.062** (0.028) 0.038 (0.024) 0.092** (0.043) Yes Yes 0.021 0.237 0.268 93 133

Panel B: Asset Growth

(3) **

0.039 (0.019) 0.041** (0.019) 0.138 (0.117) 0.023** (0.011) 0.029** (0.013) 0.013 (0.012) 0.219* (0.110) 0.261** (0.1128) 0.068** (0.032) 0.037 (0.025) 0.086** (0.041) Yes Yes 0.025 0.219 0.346 93 133

(4) **

0.027 (0.013) 0.032** (0.015) 0.134 (0.129) 0.022** (0.010) 0.031** (0.015) 0.007 (0.010) 0.206* (0.102) 0.215** (0.105) 0.063** (0.029) 0.035 (0.021) 0.077** (0.038) Yes Yes 0.032 0.254 0.371 93 133

(5) **

0.025 (0.012) 0.029** (0.014) 0.153* (0.077) 0.018** (0.009) 0.039** (0.019) 0.019 (0.018) 0.214* (0.106) 0.247** (0.121) 0.056** (0.026) 0.048 (0.034) 0.069 (0.038) Yes Yes 0.012 0.227 0.375 93 133

(6) **

0.035 (0.017) 0.051** (0.019) 0.225* (0.113) 0.016** (0.008) 0.041** (0.018) 0.017 (0.011) 0.181** (0.088) 0.165** (0.131) 0.055** (0.027) 0.039 (0.027) 0.063* (0.032) Yes Yes 0.011 0.236 0.264 93 133

Panel C: Deposit Growth

(7) **

0.037 (0.018) 0.043** (0.021) 0.148* (0.075) 0.015** (0.006) 0.042** (0.021) 0.019 (0.018) 0.214* (0.108) 0.250** (0.124) 0.052** (0.025) 0.051 (0.031) 0.069 (0.057) Yes Yes 0.013 0.282 0.348 93 133

(8) **

0.026 (0.012) 0.037** (0.018) 0.133 (0.129) 0.018** (0.008) 0.033** (0.015) 0.007 (0.007) 0.208* (0.105) 0.218** (0.108) 0.056** (0.026) 0.036 (0.022) 0.077** (0.037) Yes Yes 0.016 0.261 0.371 93 133

(9) **

0.023 (0.011) 0.036** (0.017) 0.143 (0.139) 0.020** (0.009) 0.036** (0.016) 0.014 (0.012) 0.234* (0.118) 0.269** (0.134) 0.052** (0.025) 0.031 (0.019) 0.079** (0.038) Yes Yes 0.021 0.214 0.343 93 133

(10) **

0.08 (0.013) 0.044*** (0.013) 0.151 (0.144) 0.023** (0.011) 0.033** (0.016) 0.014 (0.012) 0.232* (0.115) 0.266** (0.131) 0.059** (0.028) 0.028 (0.022) 0.084** (0.041) Yes Yes 0.013 0.265 0.339 93 133

(11) **

0.036 (0.017) 0.043** (0.021) 0.217* (0.109) 0.014** (0.005) 0.041** (0.019) 0.018 (0.031) 0.183** (0.091 0.166** (0.127) 0.049** (0.023) 0.038 (0.024) 0.059* (0.030) Yes Yes 0.012 0.252 0.368 93 133

(12) **

0.024 (0.011) 0.033** (0.016) 0.146* (0.074) 0.015** (0.006) 0.038** (0.018) 0.016 (0.019) 0.218* (0.111) 0.248** (0.123) 0.053** (0.025) 0.049 (0.031) 0.066 (0.053) Yes Yes 0.019 0.253 0.341 93 133

0.031** (0.014) 0.045** (0.022) 0.232* (0.117) 0.019** (0.009) 0.053** (0.026) 0.023 (0.016) 0.192** (0.094) 0.192** (0.143) 0.057** (0.027) 0.039 (0.023) 0.067* (0.034) Yes Yes 0.019 0.235 0.317 93 133

Note: The table shows the results from estimation of Eq. (3) where market structure variables are regressed on bank efficiency, bank growth and other control variables. Panel A, B and C report the estimation results when growth variables is represented by growth in loans, growth in assets and growth in deposits respectively. Dependent variables are Conventional Lerner index (model 1, 5, 9), Adjusted Lerner (model 2, 6 and 10), CR5 (model 3, 7 and 11) and HHI (model 4, 8 and 12). RFW and IC respectively represent regulatory framework and institutional characteristics. The coefficient are estimated using Twostep system GMM with Windmeijer (2005) corrected standard errors and small sample options. The diagnostics tests show that GMM is correctly specified and there are no identification issues. For instance, significant values of AR (1) indicate that null hypothesis of no autocorrelation among error terms in first difference is rejected. AR (2) is insignificant indicating that error terms in level regressions are not correlated. Values of Sargan/Hensen are insignificant indicating that instruments are valid. Corrected standard errors are reported in the parenthesis. Subscripts ***, **, * denote the significance of relationships at 1%, 5% and 10% levels respectively.

Eq. (3) (Table 8) are used for calculation of Sobel, Aroian and Goodman test statistic (z). The results of these tests are reported in Table 9. The statistic (z) under Sobel, Aroian and Goodman test are significant thus rejecting the null hypothesis of no indirect relationship between cost efficiency and market structure. 4.2. Robustness analysis The relationship between efficiency and market structure is also explained by quiet life (QL) hypothesis which suggests that ‘‘in a concentrated market firms do not minimize costs because of insufficient managerial effort, lack of profitmaximizing behavior, wasteful expenditures to obtain and maintain monopoly power, and/or survival of inefficient managers” (Berger & Hannan, 1989). The QL hypothesis thus suggests the causality from market power/bank concentration to efficiency. In Section 4.1, we use Two-step system GMM and treat cost efficiency and market structure as endogenous variables in estimation. However, as a robustness check, we also employ PVAR in GMM environment (Abrigo & Love, 2016) to cost efficiency, bank growth and market structure variables. PVAR also allows us to examine the reverse causality between cost efficiency and market structure. We consider this possibility following Homma et al. (2014) who argue that both ES and QL may coexist at least in short run. For instance, ‘‘if the banking market becomes more concentrated due to the growth of efficient banks, the finding for the quiet-life hypothesis then implies that the banks lose efficiency”. The estimation results from PVAR are reported in Table 10. Model, 2 and 3 respectively represent the equations corresponding to growth, efficiency and market structure.13 There are 13 Although, we apply PVAR using each proxy of bank growth and market structure, we only report results from cost efficiency, loan growth and Lerner index for sake of brevity. Results from other proxies of bank growth and market structure are qualitatively similar to those reported in Table 10.

543

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545 Table 9 Sobel, Aroian and Goodman test. Indirect relationship from cost efficiency to market structure through bank growth

Coefficients on Efficiency Variable based on Eq. (1) (Table 6) Coefficients on Growth based on Eq. (3) (Table 8) Sobel Test Aroian Test Goodman Test

1

2

3

0.139*** (0.041) 0.047*** (0.016) 2.117** (0.0039) 2.140** (0.0027) 2.108** (0.0043)

0.113*** (0.027) 0.051** (0.019) 2.157** (0.0038) 2.147** (0.0026) 2.212** (0.0042)

0.183*** (0.051) 0.044*** (0.013) 2.151** (0.0037) 2.151** (0.0025) 2.218** (0.0041)

Note: The Table reports the mediation analysis based on Sobel, Aroian and Goodman tests. The coefficients on cost efficiency (row 1) and the growth variable (row 2) are used to calculate the z statistics for mediation analysis on the basis of Sobel, Aroian and the Goodman tests which are reported in the last three rows. The null hypothesis underlying each test is that the indirect relationship between efficiency and concentration/market power (from efficiency to growth and then from growth to concentration/market power) is not significantly different from zero. Rejection of null hypothesis thus implies that such effect is present. Standard errors are reported in the parenthesis. Subscripts ***, **, * denote the significance of relationships at 1%, 5% and 10% levels respectively.

Table 10 Panel Vector Auto Regression (PVAR).

Cost Efficiency (t-1) Cost Efficiency (t-2) Wald Test P Value Growth (t-1) Growth (t-2) Wald Test P Value Market Structure (t-1) Market Structure (t-2) Wald Test P Value

Model 1 Growth

Model 2 Efficiency

Model 3 Market Structure

0.062** (0.029) 0.035** (0.016) (0.013)** 0.063** (0.031) 0.054 (0.047) – 0.063 (0.351) 0.087 (0.253) (0.142)

0.179** (0.078) 0.026 (0.057) – 0.015 (0.018) 0.011 (0.019) (0.133) 0.168 (0.136) 0.091** (0.044) (0.047)**

0.129** (0.063) 0.024* (0.013) (0.041)** 0.112** (0.052) 0.014** (0.007) (0.007)*** 0.036** (0.017) 0.084 (0.063) –

Note: This Table shows the estimation results from application of Panel Vector Auto Regression (PVAR) in generalized method of moment (GMM) environment proposed in Abrigo and Love (2016) to cost efficiency, bank growth, and market structure variable (Lerner index). Wald P-value shows the probability of joint significance of coefficients under Granger causality. Variables are tested for unit root using Fisher type augmented Dickey-Fuller and the Phillips-Perron unit root tests. The lag selection is based on the Bayesian Information Criteria (BIC), the Akaike Information Criteria (AIC) and the Hannan-Quinn Information Criteria (HQIC). The diagnostic tests indicated that model is good fit. Standard errors are reported in the parenthesis. Subscripts ***, **, * denote the significance of relationships at 1%, 5% and 10% levels respectively.

several important findings from the estimation of PVAR. First, the cost efficiency positively affects bank growth (model 1) and market structure (model 3). Second, the bank growth positively influences the market structure (model 3). Third, there is no causality from bank growth to cost efficiency (model 2). Fourth, there is no causality from market power to bank growth (model 1). Finally, there is some evidence of causality from market structure to cost efficiency (model 2), as coefficient on second lag of market structure is significant and negative. These causalities are also supported by Wald p-values reported below the coefficients on each variable. This last finding supports the co-existence of QL hypothesis as suggested by Homma et al. (2014). Overall evidence from analyses suggests that cost efficiency enables banks to grow and increase their market share. Subsequently, the growth in terms of loans, assets and deposits leads the banks to gain more market power/leads the market to become more concentrated. These findings support the ES hypothesis for banking industry in ASEAN. However, there is also some evidence to support QL hypothesis implying that the ES and QL hypotheses may coexist.

544

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545

5. Conclusion and policy implications The efficient structure (ES) hypothesis suggests that efficient firms grow, obtain higher market share and become larger; consequently, the market becomes more concentrated. The policy implications of ES hypothesis thus follow that the concentrated markets are dominated by efficient firms/banks. Therefore, antitrust policies/anti-concentration measures can bring unwarranted distortions in the market. These implications depend on a sound and acceptable empirical test of the ES hypothesis. However, the traditional tests of the ES hypothesis have been controversial. For instance, Homma et al. (2014) propose that the legitimate test of ES hypothesis should examine the causalities from efficiency to growth and then from growth to market structure. The new test proposed by Homma et al. (2014) is however incomplete as they only test the relationship from efficiency to growth. This study extends the work of Homma et al. (2014) and provides empirical evidence on complete set of relationships in their proposed method – i.e. efficiency to growth and then growth to market structure. We apply Two-step system GMM and PVAR to panel of commercial banks in ASEAN-5 over the period of 1999–2014. The results show that efficient banks grow and the resultant growth leads to banks’ market power and market concentration. This empirical evidence corresponds to the nexus of relationships predicted by the ES hypothesis. There is also some evidence to support QL hypothesis, which implies that ES and QL hypothesis may coexist for ASEAN banking industry. The interesting finding is that the relationship between efficiency and market structure is not completely defined by bank growth suggesting that there are some other variables that may also explain this relationship. Investigation of these variables is however beyond scope of this study and we leave this topic for future research. Findings of this study provide important implications for policy makers in ASEAN. For example, the implications of ES hypothesis are that the concentrated markets are dominated by efficient firms/banks. Therefore, antitrust policies/anticoncentration measures can bring unwarranted distortions in the market. However, the implications of QL hypothesis are quite contrary to those of ES hypothesis. Since, this study has provided empirical evidence on co-existence both these hypothesis, the policy makers need to seek a balance between level of competition and bank concentration.

References Abrigo, M. R., & Love, I. (2016). Estimation of panel vector autoregression in stata. Stata Journal, 16(3), 778–804. Al-Muharrami, S., & Matthews, K. (2009). Market power versus efficient-structure in arab gcc banking. Applied Financial Economics, 19(18), 1487–1496. Altunbas, Y., Evans, L., & Molyneux, P. (2001). Bank ownership and efficiency. Journal of Money, Credit and Banking, 926–954. Arena, M., Reinhart, C. M., & Vazquez, F. (2007). The lending channel in emerging economies: Are foreign banks different? IMF Working Papers, 1–52. Ari, A., & Cergibozan, R. (2016). The twin crises: Determinants of banking and currency crises in the Turkish economy. Emerging Markets Finance and Trade, 52(1), 123–135. Avkiran, N. K., & Thoraneenitiyan, N. (2010). Purging data before productivity analysis. Journal of Business Research, 63(3), 294–302. Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173. Berger, A. N. (1995). The profit-structure relationship in banking-tests of market-power and efficient-structure hypotheses. Journal of Money, Credit and Banking, 404–431. Berger, A. N., & Hannan, T. H. (1989). The price-concentration relationship in banking. The Review of Economics and Statistics, 291–299. Berger, A. N., & Hannan, T. H. (1997). Using efficiency measures to distinguish among alternative explanations of the structure-performance relationship in banking. Managerial Finance, 23(1), 6–31. Berger, A. N., Rosen, R. J., & Udell, G. F. (2007). Does market size structure affect competition? The case of small business lending. Journal of Banking & Finance, 31(1), 11–33. Bhatti, G. A., & Hussain, H. (2010). Evidence on structure conduct performance hypothesis in Pakistani commercial banks. International Journal of Business and Management, 5(9), p174. Bikker, J. A. (1999). Efficiency in the European banking industry: An exploratory analysis to rank countries. De Nederlandsche Bank NV. Bikker, J. A., & Haaf, K. (2002). Competition, concentration and their relationship: An empirical analysis of the banking industry. Journal of Banking & Finance, 26(11), 2191–2214. Bourke, P. (1989). Concentration and other determinants of bank profitability in Europe, North America and Australia. Journal of Banking & Finance, 13(1), 65–79. Brewer, E., & Jackson, W. E. (2006). A note on the ‘‘risk-adjusted” price–concentration relationship in banking. Journal of Banking & Finance, 30(3), 1041–1054. Calem, P. S., & Carlino, G. A. (1991). The concentration/conduct relationship in bank deposit markets. The Review of Economics and Statistics, 268–276. Casu, B., & Girardone, C. (2006). Bank competition, concentration and efficiency in the single European market. The Manchester School, 74(4), 441–468. Casu, B., & Girardone, C. (2009). Testing the relationship between competition and efficiency in banking: A panel data analysis. Economics Letters, 105(1), 134–137. Casu, B., Girardone, C., & Molyneux, P. (2004). Productivity change in European banking: A comparison of parametric and non-parametric approaches. Journal of Banking & Finance, 28(10), 2521–2540. Chan, S., Koh, E. H. Y., & Kim, Y. (2016). Effect of foreign shareholdings and originating countries on banking sector efficiency. Emerging Markets Finance and Trade, 52(9), 2018–2042. Chan, S.-G., Koh, E. H., Zainir, F., & Yong, C.-C. (2015). Market structure, institutional framework and bank efficiency in asean 5. Journal of Economics and Business, 82, 84–112. Chao, C., Yu, M., Lee, Y., & Hsiao, B. (2017). Measurement of banking performance in a dynamic multiactivity network structure: Evidence from banks in Taiwan. Emerging Markets Finance and Trade, 53(4), 786–805. Chavarín, R. (2016). Profitability in banks affiliated to a business group: Evidence from Mexico. Emerging Markets Finance and Trade, 52(8), 1892–1909. Coccorese, P. (2009). Market power in local banking monopolies. Journal of Banking & Finance, 33(7), 1196–1210. De Bandt, O., & Davis, E. P. (2000). Competition, contestability and market structure in European banking sectors on the eve of emu. Journal of Banking & Finance, 24(6), 1045–1066. Demsetz, H. (1973). Industry structure, market rivalry, and public policy. Journal of Law and Economics, 1–9. Evanoff, D. D., & Fortier, D. L. (1988). Reevaluation of the structure-conduct-performance paradigm in banking. Journal of Financial Services Research, 1(3), 277–294. Gale, B. T., & Branch, B. S. (1982). Concentration versus market share: Which determines performance and why does it matter. Antitrust Bulletin, 27, 83.

H.H. Khan et al. / North American Journal of Economics and Finance 42 (2017) 531–545

545

Gardener, E., Molyneux, P., & Nguyen-Linh, H. (2011). Determinants of efficiency in South East Asian banking. The Service Industries Journal, 31(16), 2693–2719. Goldberg, L. G., & Rai, A. (1996). The structure-performance relationship for European banking. Journal of Banking & Finance, 20(4), 745–771. Goodman, L. A. (1960). On the exact variance of products. Journal of the American Statistical Association, 55(292), 708–713. Homma, T., Tsutsui, Y., & Uchida, H. (2014). Firm growth and efficiency in the banking industry: A new test of the efficient structure hypothesis. Journal of Banking & Finance, 40, 143–153. Hsieh, M.-F., & Lee, C.-C. (2010). The puzzle between banking competition and profitability can be solved: International evidence from bank-level data. Journal of Financial Services Research, 38(2–3), 135–157. Jackson, W. E. (1992). The price-concentration relationship in banking: A comment. The Review of Economics and Statistics, 373–376. Kaufmann, D., Kraay, A., & Mastruzzi, M. (2015). The worldwide governance indicators (wgi) project. Retrieved 02 November, 2015 http://info.worldbank. org/governance/wgi/index.aspx#home. Khan, H. H., Ahmad, R. B., & Gee, C. S. (2016). Bank competition and monetary policy transmission through the bank lending channel: Evidence from asean. International Review of Economics & Finance, 44, 19–39. Kim, H., Park, K., & Song, S. (2016). Banking market size structure and financial stability: Evidence from eight Asian countries. Emerging Markets Finance and Trade, 52(4), 975–990. Koetter, M., Kolari, J. W., & Spierdijk, L. (2012). Enjoying the quiet life under deregulation? Evidence from adjusted Lerner indices for us banks. Review of Economics and Statistics, 94(2), 462–480. Laidroo, L. (2016). Bank Ownership and Lending: Does Bank Ownership Matter? Emerging Markets Finance and Trade, 52(2), 285–301. Lee, C., & Huang, T. (2016). Productivity changes in pre-crisis Western European banks: Does scale effect really matter? The North American Journal of Economics and Finance, 36, 29–48. Lerner, A. P. (1934). The concept of monopoly and the measurement of monopoly power. The Review of Economic Studies, 1(3), 157–175. Lloyd-Williams, D. M., Molyneux, P., & Thornton, J. (1994). Market structure and performance in Spanish banking. Journal of Banking & Finance, 18(3), 433–443. MacKinnon, D. P., & Dwyer, J. H. (1993). Estimating mediated effects in prevention studies. Evaluation Review, 17(2), 144–158. MacKinnon, D. P., Warsi, G., & Dwyer, J. H. (1995). A simulation study of mediated effect measures. Multivariate Behavioral Research, 30(1), 41–62. Martin, S. (1988). Market power and/or efficiency? The Review of Economics and Statistics, 331–335. Maudos, J. (1998). Market structure and performance in Spanish banking using a direct measure of efficiency. Applied Financial Economics, 8(2), 191–200. Mendes, V., & Rebelo, J. (2003). Structure and performance in the Portuguese banking industry in the nineties. Portuguese Economic Journal, 2(1), 53–68. Molyneux, P., & Forbes, W. (1995). Market structure and performance in European banking. Applied Economics, 27(2), 155–159. Molyneux, P., & Thornton, J. (1992). Determinants of European bank profitability: A note. Journal of Banking & Finance, 16(6), 1173–1178. Mueller, D. C. (1983). The determinants of persistent profits. Washington, DC: Bureau of Economics, Federal Trade Commission. Mulder, C., Perrelli, R., & Rocha, M. D. (2016). The role of bank and corporate balance sheets on early warning systems of currency crises—An empirical study. Emerging Markets Finance and Trade, 52(7), 1542–1561. Papadopoulos, S. (2004). Market structure, performance and efficiency in European banking. International Journal of Commerce and Management, 14(1), 79–100. Park, K. H., & Weber, W. L. (2006). Profitability of Korean banks: Test of market structure versus efficient structure. Journal of Economics and Business, 58(3), 222–239. Pilloff, S. J., & Rhoades, S. A. (2002). Structure and profitability in banking markets. Review of Industrial Organization, 20(1), 81–98. Punt, L. W., & Van Rooij, M. (2003). The profit-structure relationship and mergers in the European banking industry: An empirical assessment. Kredit und Kapital, 36(1), 1–29. Ravenscraft, D. J. (1983). Structure-profit relationship at the line of business and industry level. The Review of Economics and Statistics, 22–31. Ravenscraft, D. J. (1984). Collusion vs. superiority: A Monte Carlo analysis. International Journal of Industrial Organization, 2(4), 385–402. Rhoades, S. A. (1985). Market share as a source of market power: Implications and some evidence. Journal of Economics and Business, 37(4), 343–363. Samad, A. (2008). Market structure, conduct and performance: Evidence from the Bangladesh banking industry. Journal of Asian Economics, 19(2), 181–193. Seelanatha, L. (2010). Market structure, efficiency and performance of banking industry in Sri Lanka. Banks and Bank Systems, 5(1), 20–31. Shaffer, S., & DiSalvo, J. (1994). Conduct in a banking duopoly. Journal of Banking & Finance, 18(6), 1063–1082. Shepherd, W. G. (1982). Causes of increased competition in the us economy, 1939–1980. The Review of Economics and Statistics, 613–626. Shepherd, W. G. (1986). Tobin’s q and the structure-performance relationship: Comment. The American Economic Review, 1205–1210. Smirlock, M. (1985). Evidence on the (non) relationship between concentration and profitability in banking. Journal of Money, Credit and Banking, 69–83. Smirlock, M., Gilligan, T., & Marshall, W. (1984). Tobin’s q and the structure-performance relationship. The American Economic Review, 1051–1060. Smirlock, M., Gilligan, T., & Marshall, W. (1986). Tobin’s q and the structure-performance relationship: Reply. The American Economic Review, 1211–1213. Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural equations models. Sociological Methodology. San Francisco: Jossey-Bass. Soedarmono, W., & Tarazi, A. (2016). Competition, financial intermediation, and riskiness of banks: Evidence from the Asia-Pacific Region. Emerging Markets Finance and Trade, 52(4), 961–974. Stevens, J. L. (1990). Tobin’s q and the structure-performance relationship: Comment. The American Economic Review, 618–623. Sturm, J.-E., & Williams, B. (2004). Foreign bank entry, deregulation and bank efficiency: Lessons from the Australian experience. Journal of Banking & Finance, 28(7), 1775–1799. Süer, Ö., Levent, H., & Sßen, S. (2016). Foreign entry and the Turkish banking system in 2000s. The North American Journal of Economics and Finance, 37, 420–435. Tregenna, F. (2009). The fat years: The structure and profitability of the US banking sector in the pre-crisis period. Cambridge Journal of Economics, 33(4), 609–632. Turk Ariss, R. (2010). On the implications of market power in banking: Evidence from developing countries. Journal of Banking & Finance, 34(4), 765–775. Weiss, L. W. (1974). The concentration-profits relationship and antitrust. Industrial concentration: The new learning, 184. Zhu, N., Wang, B., Yu, Z., & Wu, Y. (2016). Technical efficiency measurement incorporating risk preferences: An empirical analysis of Chinese commercial banks. Emerging Markets Finance and Trade, 52(3), 610–624.