Assessing the determinants of bank performance in vietnam empirical evidence from domestic commercial banks in the years 2005 2011

While bank size, capitalization, the growth rate of GDP per capita and annual inflation rate are found to have significant impact in JSCBs, the profitability of SOCBs are mainly relied o

(OUM OPEN UNIVERSITY MALAYSIA

EMPIRICAL EVIDENCE FROM DOMESTIC

COMMERCIAL BANKS IN THE YEARS

ACKNOWLEGEMENT

I would like to express my sincere gratitude to my thesis and academic Advisor, Phan Thi Giac Tam, Ph.D, for her encouragement, guidance and patience, time and invaluable input throughout the preparation of this work Thank for her dedication and expertise, I am able to complete the research so effectively in a short period of time Grateful thanks are therefore due

to my Advisor, Phan Thi Giac Tam, Ph.D

Furthermore, I would like to thank my family and friends for their love, encouragement and

support

ABSTRACT

This paper aims to identify the deteminants of commercial bank profitability in Vietnam and

to discuss the extent to which these determinants have influenced response variable In order

to examine the relationship between bank performance and its control factors, the ordinary least squares (OLS) estimation with the fixed effects model are employed Data set is derived from 15 commercial banks over the period 2005-2011 and divided into two groups according

to their proprietorship (SOCB and JSCB) The empirical results show that both internal factors (bank-specific characteristics) and external factors (macroeconomic conditions and market structure) have effects on return on average assets Notably, the influence of determinants in state-owned banks is dissimilar to private banks While bank size, capitalization, the growth rate of GDP per capita and annual inflation rate are found to have significant impact in JSCBs, the profitability of SOCBs are mainly relied on the level of capitalization, fund source and credit risk Finally, with regard to the importance of determinants of bank performance but scant sources, suggestions for further research are provided in the last section

TABLE OF CONTENTS

Page Title Page

Acknowledg€rT€TIE << 5< s25 x4 0 T1 Hàn Hán 141801441411014.1 7811118 i ASTTACE cà HH HH HH TH ng g4 1 14841111101001101110020.1000010 1n ii Tabl€ OŸ COTIE€TIẨS Hà nh nh TH nàn HH0 18.1011414102101011721172000700 iii List of Abbreviations 2.0.0.0 seesssssscscssssnenssssessusesenscsevensessnseteasenesersenssensatoneneneseesessnsaeneneees iv List of Tables and EÏEUT€S - sa cà HH HH H091 0101 14003131340118120100070141204e V Chapter One: INTRODUCTION .sssssssssesssseseccessersrserenceserssserserernsnssonsnsnensssceneneasennens 01 Chapter Two: LITERATURE REVIEW .ccccssssscscssesssssscesenserersnsneserensaees 03 2.1 The studies outside of Vietnam .ccceccsssssssssssscsstsnserenesceseesesensereassnsenseeesseeetes 03 2.2 The studies of Vietnamese banking s€CfOT che 16 2.3 Summary and remarks ccesssecssscecseescssessssssensesesscsssenseneseesssssnsessnsassansessssneeeeeeres 19 Chapter Three: METHODOLOGY AND DA TA « eeessseseseeseesesessesersresee 23

3.1 Methodology and econometric mOd€ÏL - «- se s3 t S1 1411111 rirexxe 23

3.2 Data Set 26 3.2.1 Selection of samples c.cccccscseecescecseeesesesssssssssssssusnsrsssessneseseeneeesneasesesnsesenes 26 3.2.2 Selection of the variabÌ€S - «ch HH HH HH2 1c 1010301801110 28 3.2.2.1 Dependent var1abÌ€ « cưng HH Hàn nà HH 08010101951 28 3.2.2.2 Independent variabbÌ€s - + + nen n0 0111410318411801171070015e 29 Chapter Four: EMPIRICAL RESULTS AND DISCUSSION .<<.< 37

AL Findings 37 4.2 Compare and contrast to existing SEUI€S - che rereereeerrrke 45

Return on Average Equity

State Bank of Vietnam Stochastic frontier approach State-owned Commercial Bank

State-owned Enterprises

United Nations Development Programme Vietcombank Securities

Vietnam Dong

LIST OF TABLES AND FIGURES

Table 1: Summary of cross-country studies on bank performance

Table 2: The comparison between GMM and OLS estimators

Table 3: The details of sample banks studied

Table 4: List of the variables studied in the study

Table 5: Descriptive statistics

Table 6: Correlation matrix for the explanatory variables

Table 7: Multiple regression results

Figure 1: The number of banking institutions in Vietnam (2005-2011)

Figure 2: The proportion of deposits and credit market share according to bank ownership

Chapter One INTRODUCTION Banking industry has been widely considered as the most performing and contributing service sector in an economy as it encourages people to save and invest in productive streams and become contributing individuals of a nation and it also provides security and ensure safety

of investments and savings (Raza et al., 2013) Similar to the economy in Vietnam, although banking sector does not directly create goods or products, the growth in this industry is an incentive for developments in other sectors (PNS, 2013) In other words, the strong and stable position of banking system has become more and more important, especially in a world of global financial system as recent Hoffmann (2011) states that in the liberalization markets, the foreign bank involvement in domestic banking markets has increased, it tends to result in intensifying competition and reduction in margins So as to operate effectively in such severe

environment, it is essential to answer the following questions: what are the factors that can

help banks achieve better performance? To what extent do these control variables have impact

on bank profitability? The findings of these questions are widely believed not only important

to bank managers but also crucial banking regulators in enhancing the policy (Berger and Humphrey, 1997; Vu and Nahm, 2013) As far as ] am aware, the relationship between bank profitability and its indicators has been conducted in the developed countries for a while but it has just been concerned in the developing and emerging markets from last two decades In spite of the recent appearance, this topic has been an exciting phenomenon studied by a great number of analysts such as Guru et al (2002), Sufian and Chong (2008), Flamini et al (2009), Gardener et al (2011), Reddy and Nirmala (2013) However, to the best of my knowledge, there are only few researches in the Vietnamese banking system and almost all of them are focused on bank efficiency’ instead of profitability As being acutely conscious of the importance of the objective but insufficiency of existing resources, I decide to choose

“Assessing the determinants of bank performance in Vietnam: Empirical evidence from domestic commercial banks in the years 2005-2011.” as my thesis topic

' Bank efficiency is comprised of cost efficiency and profit efficiency Together with bank profitability, they are proxies for bank performance.

As the first study which examines the determinants affecting bank performance measured by profitability ratios in Vietnam, the paper is expected to achieve two following aims The principal target is to identify the correlation between explanatory indicators and bank profitability on the basis of a fixed effects model The second target is to compare and contrast the effect of these factors on two groups of commercial banks: state-owned banks and joint-stock banks Employing the ordinary least squares (OLS) and data set for the period '

2005-2011, the results indicate that bank specific characteristics, macroeconomic conditions

and market structure altogether have affected to bank performance but in disparate magnitude and sign It is found that the factors which have significant impact on joint-stock banks’ performance, for instance bank size, the growth rate of economy, and annual inflation rate are insignificantly related to profitability of state-owned banks Conversely, the significant

indicators in state-owned banks, such as credit risk and the effect of fund source are :

insignificant in joint-stock banks Moreover, while the performance of the latter group seems reliant on the macroeconomic conditions, the regression results reveal that there is no significant correlation among macroeconomic risk factors and state-owned banks’ performance The further details are exhibited in chapter four

The rest of this paper is organized as following Chapter two reviews the literatures relating to the determinants of bank performance from various banking markets and then narrowing to the target market Chapter three introduces the methodology and the selection of the variables applied in the analysis Next, chapter four exhibits and discusses on empirical results Finally, chapter five is going to summarize and remark findings for the whole context

Chapter Two LITERATURE REVIEW This chapter is going to review the existing papers that are related to determinants of bank profitability For a clear structure, this section is mainly divided into three parts: the first part is the general review of studies in different countries, except Vietnam The second part is going to focus on the analyses in Vietnam banking system Finally, based on the experience from various literature and cross-reference studies, the direction for the current analysis is

exhibited

2.1 The studies outside of Vietnam

For the origin of the objective, it seems that the researches on bank performance and its determinants have been the interest of a large number of authors for many decades ago (Alhadeff, 1954; Schweiger and McGee, 1961; Grebler and Brigham, 1963; Benston, 1972)

In these early studies, the examination of production efficiency and contro] factors are generally for the purpose of testing the presence of the economies or diseconomies of scale”

At first, the observation of scale economies is mainly based on the relation among size,

organizational form and cost efficiency Then, it is proved that beyond the influence of size factor, the cost efficiency is also explained by other explanatory indicators (Smith, 1955) Investigating for the different control variables in scale economies is the way how these early studies closely related to the current papers on assessing influencing factors of bank performance (Spathis et al., 2002) Schweiger and McGee (1961), Gramley (1962), Grebler and Brigham (1963) are regarded as the pioneering work on diversifying explanatory indicators in economies of scale Apart from bank size variable captured by deposits, Schweiger and McGee (1961) select the ratio of time to total deposits; the level of earning asset structure captured by the ratios of business loans to assets, consumer loan to assets, and farm loans to assets; the rate of asset growth; and other structure variables as independent variables affecting cost efficiency In line with the statement that there are many of the

? Economies of scale is a long-run concept used to express the statement that an organization tends to perform better, especially when an equiproportional increase in inputs leads to a greater than equiproportional increase in outputs, Heffernan (2010) defined At the beginning, scale economies were regarded as the cost-size relations Then it is proved that the size is not the dominant factor to assess production efficiency (Smith, 1955).

determinants which influence bank costs other than total assets, Gramley (1962) take more independent variables into account, such as percentage growth of assets, the ratios proxy for earning asset structure (total loans to total assets, non-government securities to total assets, and consumer to total loans) Employing multiple regression technique, Schweiger and McGee (1961), Gramley (1962) testing for the all member banks in Chicago and Tenth District respectively, both studies conclude that the model design highly presents economies of scale

In other words, it indicates that the additional variables also affect cost efficiency Taking Grebler and Brigham (1963) as another example, more independent variables are also added to assess the cost efficiency in California, such as average deposit size, the rate of assets growth, the ratio of buildings to total assets However, the findings of Grebler and Brigham show no evidence for the presence of economies of scale So a raising question here is that applying similar functional forms why the results are completely conflicting among studies With regard to these above studies, Hester and Zoellner (1966) state that the findings obtained from model employed are unreliable Explaining for the weaknesses, Hester and Zoellner assert that

as the model was not formulated to derive estimates of net rates of return so in actually, the regressions did not take into account all earning assets and deposit liabilities as explanatory variables Another limitation of the regression models is the multicollinearity problem among variables, for example the close correlation between dependent variable (measured by the ratio

of total operating costs to total assets) and size variable (total assets) may lead to unexpected misstatement In support of the remaining issues in Schweiger and McGee (1961), Gramley

(1962), Grebler and Brigham (1963), it is criticized that:

The assumed functional relationship among the output variables is not based on any production theory or other rationale .[and] single point stock data were used to

estimate service flow, which introduces an unknown bias into the coefficients

(Berger, 1972, p 323)

In spite of the disadvantages above, these scale economies studies are still appreciated as they initiate the new consciousness that bank profitability is not only computed by size and organizational form but also other factors In the following sections, the findings for

Based on the experience from Gramley (1962) as mentioned above, Hester and Zoellner (1966) do a research on the relation between bank portfolios and earnings in Kansas City and Connecticut over the period 1957-1963 This study is marked as one of the earliest papers in the U.S with the aim of examining the determinants of bank profits instead of economies of scale Analyzing the influence of balance sheet items on profits, Hester and Zoellner (1966) find that almost all of items on the debit side have significant positive impact on bank profits, whereas the items on credit side is negatively related to bank earnings To some extent, although the structure of bank is found to have importance relation to its earnings, the explanatory power of the model design is quite low In other words, it is quite probable that bank performance is likely affected by other factors beyond the internal resources

In support this statement, Haslem (1968, 1969) expands the scope of determinants with management quality, size, and location of commercial banks in California Using 64 operating ratios to assess the management of banks, classifying size and location features into different groups, it is found that all these explanatory factors (management, size, location) have significant relationship with bank profitability (Haslem, 1968& 1969) Whereas, in another study conducted in Texas, Fraser and Rose (1971) point out the opposite results With the data from 1966 to 1967, it is found that the bank profitability in Texas is not influenced by loan

composition, bank size, and bank costs

Due to the considerably contradictory findings, Rhoades (1977) makes an observation to compare the relative studies in the U.S In the total 39 papers reviewed, it is found there were

9 cases without evidence of the concerned relation and 30 cases presenting the significant correlation To explain for the difference, the geographic scope of the studies is believed the main cause (Rhoades, 1977; Wheelock and Neely, 1997) In the 9 cases without bank structure-performance relation, they observe the individual banks instead of whole market as a unit As a result of smal! sample, it just concerns “little room for variation in the degree of concentration” and that cause conflicting results among studies (Rhoades, 1977)

The studies of assessing the determinants and bank performance continue to be conducted in many aspects, for example the increasing scope of sample tested, the expansion

in explanatory factors with inclusion of market structure variables Taking the analysis of

Heggestad (1977) as an example, if in the preceding studies, the independent variables are generally restricted to bank compositions, the industry-specific and market structure variables are additional concerns Using the ratio of net income after taxes to total assets (a flexible form of return on assets) as proxy for bank profitability, it is found that the relation between market structure variables and bank profitability is more statistically significant than between bank concentration and bank behaviour (Heggestad, 1977) In line with this statement, Smirlock (1985) also highlights the considerable influence of market structure on return on assets, return on equity, and return on total capital For further researches, Smirlock (1985) also posits that beyond bank concentration and market structure, there may be other potential

Tesources

Beside the familiar profitability ratios, such as return on assets, and return on equity,

bank performance is also measured by net interest margins (Ho and Saunders, 1981) Four principal explanatory factors are believed to have strong effect on banker’s mark-up are the implicit interest rate, the opportunity cost of required reserve, default premiums on loans and market structure Lerner (1981) comments on Ho and Saunders (1981) that even though the model design is highly effective in assessing the determinants of net interest margins, the difficulty of the analysis is the variability of independent variables In spite of the limitation,

there are many studies (Alle, 1988; Angbazo, 1997; Naceur, 2003; Ongore and Kusa, 2013) is

developed based on the relationship of net interest margins and its determinants

The determinants of bank performance are not only testing for profitability ratios, such

as return on assets, return on equity, and net interest margins as mentioned above but also for the score of bank efficiency For instance, Akhigbe and McNulty (2003, 2005, and 2011) have used profit efficiency to investigate the actual financial performance of banks in the U.S over three periods of time 1990-1996, 1995-2001, and 1999-2005 respectively In accordance with this measure, Akhigbe and McNulty also adopt similar explanatory indicators of bank profitability to the efficiency score, including bank specific characteristics and macroeconomic conditions However, due to the nature of dependent variable, there is no specific formula to calculate the exact amount of efficiency level Hence, in the studies of

bank efficiency, the analyses usually consist of two stages The estimation of efficiency score

is going to be conducted first and then the evaluation of relationship between bank efficiency and its determinants is concerned According to Akhigbe and McNulty, the techniques employed are stochastic frontier analysis’ and Tobit regression model respectively The empirical results indicate that bank size, bank age, fee-based services, credit risk, and location have significant and positive effect on bank efficiency By contrast, bank management ability and state ownership are found to highly affect profit efficiency in negative trend (Akhigbe and McNulty, 2011)

Together with the level of bank efficiency, the researches on determinants of bank performance measured by financial ratios continue to be concerned One of the most recent studies of the US market is Hoffmann (2011) examining the determinants of bank profitability over the period 1995-2007 The analysis adopts three different types of method: fixed effect estimations, Generalised Method of Moments (GMM) with system estimator, and OLS estimations Consequently, the findings in fixed effect approach and ordinary least squares technique are fairly similar to each other but quite different from GMM For example, the business capacity ratio is found to be significantly and negatively related to return on equity in the former approaches, conversely, in the latter approach, the coefficient indicates significantly positive relation Taking the ratio of interest expense to total equity as another example, while the correlation between this ratio and bank profitability is significant and positive in fixed effects and ordinary least squares models but considerably negative in GMM

estimation

In the European region, the examinations of determinants of bank profitability are also the interest to study Molyneux and Thorton (1992) analysed this subject on a sample of 18 countries for the period from 1986 to 1989 It is found that there are positive associations between bank profitability and bank concentration, government ownership Arguably, Goddard et al (2004a) focus on six developed markets in the EU (Denmark, France, Germany, Italy, Spain, the UK) and find that ownership is insignificantly related to return on equity The leverage of banks significantly and positively correlates with profitability, this finding implies

* Stochastic frontier approach (SFA) is one of parametric approaches usually used to measure bank efficiency score For further details, please refer to Berger and Humphrey (1997).

that the higher the ratio of capital to total assets, the more profitable the banks In the terms of bank size, Goddard and his colleagues state that the relationship between bank size and profitability is still unconvincing In accordance with off-balance sheet activities, only the UK indicates positive relation, by contrast in other markets, the non-traditional — profitability relationship is neutral or negative (Goddard et al., 2004) Another typical study also analysing the determinants of bank profitability in European countries as a group is introduced by Athanasoglou et al (2006) Using least squares methods of fixed effect and random effect models, Athanasoglou and his partners reveal that bank specific characteristics, industry specific determinants and macroeconomic conditions altogether have strong effect on bank profitability Collecting the data in South Eastern European bank during the period 1998 to

2002, capital adequacy ratio, bank size, foreign ownership and inflation rate are found to have positive and significant influence on return on assets and return on equity The factors which may reduce considerably the profitability are credit risk, the overheads efficiency ratio One of the key points of the study is that it not only identifies the correlations between bank performance and explanatory indicators but also provides the solutions for negative ones to the purpose of maximizing profits For example, the increase in credit risk is declared to be one of main causes of earnings decrease Hence, it is postulated that South Eastern European banks should pay more attentions to bad-debt management Otherwise, the increasing proportion of non-performing loans might become a serious problem (Athanasoglou et al., 2006)

In addition to the studies using financial ratios as proxy bank profitability, there are a number of papers preferring to bank efficiency, such as Girardone et al (2004), Kasman and Yildirim (2006), Yildirim and Philippatos (2007) In line with the methodology applied in the U.S cases, they all employ the two-stage process to examine the relationship between bank efficiency and its indicators Girardone et al (2004) use Fourier-flexible stochastic cost frontier and logistic functional form to analyse the bank inefficiency in Italy from 1993 to

1996 The findings indicate that the inefficiency level is affected positively by assets quality and negatively by capital adequacy ratio In the cases of Kasman and Yildirim (2006), Yildirim and Philippatos (2007), stochastic frontier approach and the distribution free

found that when bank size, credit risk, capitalization level, and the proportion of off-balance

sheet items to total assets increase, bank efficiency is also rising, especially with cost efficiency Conversely, the levels of cost and profit efficiency significantly decrease when non-performing provision ratio increases (Yildirim and Philippatos, 2007)

Apart from the cross-countries analyses, there are a great number of specific country

studies in the EU and other developed countries, such as Ho and Tripe (2002), Williams

(2003), Naceur and Goaied (2008), Kosmidou (2008), Alper and Anbar (2011), etc Taking Greece as typical example, there are at least three researches on this subject (Spathis et al., 2002; Pasiouras and Kosmidou, 2007; Kosmidou, 2008; Athanasoglou et al., 2008) First of all, Spathis et al (2002) employ multicriteria methods to assess the profitability factors in Greece banks during 1990 to 1999 period Secondly, the profitability factors in European banks from 1995 to 2001 are examined by fixed effects regression model Next, Kosmidou (2008) applies a fixed regression model with pool time series dataset in 23 Greece banks over the period from 1990 to 2002 Last of all, Athanasoglou and his collaborators analyse the influence of profitability determinants from 1985 to 2001 by GMM Although three studies are different in methodology and testing period, they provide fairly consistent findings It is point out that the profitability in commercial Greece banks is dependent on both internal and external factors Among variables tested, the level of capitalization, productivity growth and Gross Domestic Product are the most significant factors which improve bank performance On the other hand, credit risk, operating expense management, and stock market capitalization have significant and negative effect on profitability Interestingly, the annual inflation rate indicates the significantly positive correlation with bank performance in the study of Athanasoglou et al (2008) but in reverse sign (significantly negative) in the study of Kosmidou (2008) The difference between two studies is explained by the nature of inflation factor On the one hand, if the increase in inflation rate is anticipated, it will encourage bank profitability On the other hand, the increases in costs will be higher than the rises in revenues that results in negative correlation, if the growth of inflation rate is unanticipated (Kosmidou, 2008) In another aspect, Pasiouras and Kosmidous (2007) classify the samples into two categories based on the ownership (domestic banks and foreign banks) and there are some noticeable differences defined between two groups The empirical results indicate that only

domestic banks gain benefits from the increases in inflation rate and growth rate of real GDP per capita increase Conversely, foreign banks seem to suffer from statistical reduction in earnings From this point of view, Pasiouras and Kosmidou (2008) point out that state ownership also has particular impact on bank profitability and it can be regarded as a separate explanatory factor in further researches

In the Switzerland banking system, Dietrich and Wanzenried (2011) evaluate the determinants of profitability from 1999 to 2009 The testing period is divided into before and after the crisis (1999-2006, 2007-2009) Employing the GMM estimation, the empirical results depict that bank performance is mainly explained by operational efficiency, the growth of total loans, funding costs and business model (Dietrich and Wanzenried, 2011) From the general view of findings, the Switzerland banking system presents inefficiency in productivity even for whole period tested The relationship of return on average assets with a majority of explanatory variables, such as operating efficiency, the growth of deposits, bank size, the level

of capitalization, and asset quality, almost all of them tend to reduce the profits For instance,

the significant and negative coefficient of operating efficiency implies that the increases in administration costs as well as wages and salary expenses are going to reduce the profits rather than improve bank management ability (Dietrich and Wanzenried, 2011) Furthermore, the analysis is one of the scant studies presenting the changes in correlations before and after the peak of global crisis, especially with bank’s capital, credit quality, funding costs and Herfindahl index Taking the capital adequacy ratio as example, before the crisis 2007, the correlation between this control variable and bank profitability was insignificant When the crisis occurs, the relationship turns to significant and negative trend In line with this

statement, the relation of credit risk with earnings is also found to transform from insignificant

to significantly negative It implies the fact that the increases in provision for bad-debts are in parallel with the rises in monitoring expenditures

Apart from the U.S and the EU analyses, the studies of the determinants of bank performance are widely explored in Asian from the beginning of the twenty first century In line with the experience from the U.S and EU studies, though the analyses in the Asian banking market may vary in methodology, the compositions of explanatory indicators seem to

be standardized Specifically, bank profitability is generally measured by the financial ratios of

return on assets, return on equity and net interest margins The determinants of these ratios are also divided into two categories: internal factors (bank specific characteristics) and external factors (market structure and macroeconomic conditions) Although the relative studies in Asian countries appear quite late comparing to the U.S and EU countries, it has been an inspiration to study with a great number of papers, for instance Guru et el (2002), Chantapong (2005), Vong and Chan (2009), Garcia-Herrero et al (2009), Sufian and Habibullal (2009), Alper and Anbar (2011), Raza et al (2013), Reddy and Nirmala (2013)

One of the remarkable papers on the Asian market is the examination of Guru and his colleagues (2002) Assessing the determinants of commercial bank profitability in Malaysia, Guru et al (2002) find that bank profitability is strongly affected not only by bank specific

characteristics, such as size, capital adequacy, credit risk, bank management ability but also by

market factors, for example inflation rate and interest rate For more details, bank size,

capitalization level, credit risk, inflation and interest rate have positive effect on bank

profitability while the correlation of management efficiency with profits is considerably negative From the coefficients of explanatory indicators, bankers might adjust their financial structure better in order to maximize the profitability For example, both loans and investments in securities and subsidiaries encourage increasing returns However, Guru et al (2002) suggest that the statistical significance of the former is higher than the latter so bankers

should focus on loans rather than investments

Chantapong (2005) analyses the relation between return on assets and its indicators in Thailand during the period 1995 to 2000 Using the generalized least squares technique, it is found that the most effective factors are credit risk, quality of assets, and the level of bank

diversification into non-traditional services On the one hand, the significant and positive

correlation between the ratio of loans to total assets and bank profitability encourages the Thai banks to increase loans to their customers On the other hand, the results also indicate that bank managers should focus on controlling non-performing loans as the influence of bad-debts

is significantly negative on bank revenues The significant and positive relation between the earnings from non-traditional activities and bank profitability highlights the important contribution of these activities in improving bank performance In addition, due to the increasing participation of foreign banks in Thai banking market, Chantapong (2005) makes a

~11~

comparison between foreign banks and domestic banks according to the determinants studied The results point out that the foreign banks obtain the higher profitability than domestic ones From this point of view, it is quite probable that state ownership should be considered as the separate indicator in the analysis, especially with the countries that have a great number of foreign participations Also examining Thai banking sector, Sufian and Habibullah (2010) analyse technical efficiency score and its determinants in the years 1999-2008 The analysis consists of two stages: firstly, the data envelopment analysis’ is employed to assess the level

of efficiency Then, to evaluate the correlation between technical efficiency score and its determinants, both Tobit regression model and OLS are adopted Surprisingly, the empirical results exhibit that there is no significant difference between two regression models.° It is

pointed that bank size, credit risk, and bank diversification towards non-interest income are

significant and negative factors which lower the efficiency of banks By contrast, the significant and positive correlation is found between bank capitalization and technical efficiency score Based on the findings, Sufian and Habibullah (2010) conclude that the inefficiency is mainly dependent on scale of banks rather than pure technical efficiencies

From this point of view, small banks are the most efficient, whereas medium-sized banks are

the least efficient group However, the limitation of this statement is that the analysis seems to have no concern with the largest banks which may be benefits from economies of scale (Akhigbe and McNulty, 2003; Yildirim and Philippatos, 2007) Furthermore, Sufian and Habibullah (2010) remark that domestic banks seem to be more efficient than foreign counterparts This conclusion is quite conflicting to the study of Chantapong (2005) above It

is fairly certain that difference results from two main reasons: the method employed and the objective studied as Tregenna (2009) state that high profitability may not be parallel to high efficiency Therefore, the correlation of determinants with bank efficiency may be dissimilar

to that with bank profitability

? Data envelopment analysis (DEA) is one of the non-parametric techniques generally used to rank bank

efficiency score For further details, please refer to Yue (1992)

® According to Hoff (2007) and Andries (2011), the ordinary least squares may be unbiased in assessing the

determinants of bank efficiency A Tobit regression model is more appropriate because efficiency measures range between 0 and 1

Sufian and Chong (2008) analyse the determinants of bank profitability in the Philippines during the period from 1990 to 2005 Adopting the linear regression model for 24 commercial banks, it is found that the variations in bank performance are not only explained

by endogenous factors but also exogenous indicators In terms of bank specific characteristics, bank profitability on the one hand suffers negative effects from bank size, credit risk, and the overhead expense ratio, on the other hand is supported significantly by the increase in capital adequacy ratio and fee-based services In terms of macroeconomic conditions, only annual inflation rate has significant and negative impact on bank profitability The other external indicators, such as economic growth, the growth of money supply, and the developments of stock-market though encourage bank profitability but their impact is insignificant

The determinants of Macao commercial banks have been examined by Vong and Chan (2009) With the combination of fixed effects model and panei data over the period from 1993

to 2007, the linear regression analysis indicates that bank specific attributes have significant influence on earnings The findings in Macao are quite similar to the Philippines case above Particularly, it is also found that bank size, credit risk, and quality of assets are negative determinants of bank profitability By contrast, Macao banks seem to achieve a higher level of profitability when they increase the level of capitalization and improve the bank management

ability In accordance with macroeconomic conditions, annual inflation rate is the most

important factor comparing to the real interest rate, the growth of economy, the size of the banking sector in the whole economy, and the Lerner Monopoly Index, that have positive impact on bank returns

In the China banking system, the determinants of commercial banks have just been considered for last few years (Garcia-Herrero et al., 2009; Sufian and Habibullah, 2009a) Although the analyses are conducted by different estimations (Generalized Method of Moments and linear regression model respectively), the conclusions seem similar to each other Both of studies conclude that almost all of bank specific characteristics, industry specific factors and macroeconomic conditions have statistically significant influence on bank profitability Garcia-Herrero et al (2009), Sufian and Habibullah (2009a) both conclude that bank profitability is significantly and positively related to capitalization level and quality of assets However, Sufian and Habibullah (2009a) declare that the influence of explanatory

~13~

indicators and bank performance is inconsistent among different type of banks To support this statement, the authors run multivariate regressions four times (all banks, state-owned commercial banks, joint-stock commercial banks and City banks) Interestingly, the empirical

results illustrate dissimilar correlations For example, in the aspect of credit risk, the “all

banks” coefficient indicates positive but insignificant association, in line with that is joint- stock banks and City banks However, credit risk is a significant factor to optimize profits in state-owned banks In terms of quality of assets, the coefficients in “all banks” case and foreign banks indicate that increase in non-performing loan provision will reduce bank profits but insignificantly On the contrary, this connection is significant and positive in the other types of bank (Sufian and Habibullah, 2009a) Put another way, returns in state-owned banks and joint-stock banks are growing significantly when the ratio of assets quality rises As a result of disparate correlations, it is essential to divide all banks into smaller groups so as to provide the more accurate relations

Taking India as another typical example for the Asian region, Reddy and Nirmala (2013) point out statistically significant associations between profit inefficiency and explanatory indicators Employing the stochastic frontier approach and technical inefficiency effects model, it is revealed that profit efficiency of Indian commercial banks tends to increase when banks expand their size, the level of capitalization, the proportion of total loans over total assets, the ratio of demand deposit over total deposits, and off-balance sheet activities The other determinants relating to industry specific attributes, such as Herfindahl index, bank ownership (foreign banks or domestic banks) are found to have positive correlations with bank inefficiency (Reddy and Nirmala, 2013)

Sufian and Habibullah (2009b) assess the determinants of commercial banks in Bangladesh, one of the developing markets in Asian region recently In terms of response variables, return on average assets, return on average equity and net interest margins are simultaneously proxied for bank profitability In terms of explanatory variables, credit risk, bank size, quality of assets, fee-based services, bank management efficiency, capital adequacy ratio, growth rate of economy, and inflation are adopted Employing a linear regression model for the data set during the period 1997-2004, Sufian and Habibullah (2009b) find that the performance of Bangladesh commercial banks mainly depends on bank-specific determinants

Macroeconomic conditions though have impact on bank profitability, the influence of these variables are insignificant Based on the empirical analysis, it is noticed that in the correlation between bank profitability and its determinants, the opportunity of a fixed effects model has been rather than random effects model (Sufian and Habibullah, 2009b) This statement was also supported in the Philippines study Moreover, the results exhibit that the correlation of each independent variable with different profitability measures are dissimilar from each other For instance, when the level of capitalization increases, it will affect three profitability ratios

in different ways as follow: return on average assets increases but insignificantly, net interest margins significant increases, and return on average equity slightly decreases It is in line with the suggestion in previous study that the choice of profitability ratio is dependent on the objective of measure (Guru et al., 2002)

One of other successful markets in Asia is Korea, Sufian (2011) analyses factors influencing bank profitability in this country over the period 1992 to 2003 In line with the existing studies, bank performance in Korea is highly affected by internal and external factors

In the analysis, least squares method of the fixed effects model is adopted It is pointed that assets quality and bank management efficiency are the most significant factors in endogenous aspect that have negative effect on bank profitability On the contrary, fee-based services highly encourage bank performance In accordance with exogenous factors, annual inflation rate, the level of concentration measured by the ratio of three largest banks’ assets, and the development of stock market are found to have significant and positive influence on improving efficiency of assets management (Sufian, 2011)

In short, the studies relating to the determinants of performance in commercial banks

have been conducted worldwide The topic has attracted a large number of analysts not only in the developed countries but also in the developing nations A review of literature and cross- reference studies, excluding Vietnamese banking system shows that either financial ratios, such as return on assets, return on equity, net interest margins or the level of efficiency can be proxies for this response variable Based on the objective studied, the techniques employed are different from each other Particularly, in the cases of bank efficiency, the procedure consists

of two stages: the non-parametric approach (DEA) or parametric approach (SFA) is applied to estimate efficiency score and the Tobit model is used to assess the determinants of bank

~15~

efficiency In the cases of financial ratios as proxy bank profitability, the GMM estimator or OLS method is preferable technique to evaluate the influencing factors From the findings of each literature, it is likely that the impact of explanatory indicators on bank performance is inconsistent among countries or even at the same country The inconsistency highly results from different macroeconomic environments, the unequal size and operation of banking sector across countries (Demirguc-Kunt and Huizinga, 1998) As a result of that, it is worthwhile to focus on the target market to review how the determinants of bank performance have been

concerned

2.2 The studies of Vietnamese banking sector

In this section, the studies of evaluating bank performance in Vietnamese market are mainly regarded To my knowledge, the investigation of the efficiency in banks has conducted from the 2000s, such as Hung (2007), Ngo (2010), Vu and Turnell (2010) However, these studies just focus on measuring the value of cost and profit efficiency but not introducing their specific control factors yet The findings from these studies just provide the general statements such as the total factor productivity is mainly achieved from the advancement in allocative efficiency, meaning regulatory procedures; and technical efficiency, meaning managerial capacity (Hung, 2007) Ngo (2010) employs DEA approach to analyse the efficiency of commercial banks in the year 2008 Based on the inputs variables (wages, cost, expenses, etc.) and outputs variables (quantity, revenues, profit), the empirical result indicates that the efficiency score of Vietnamese banks is close to optimal score Vu and Turnell (2010) postulate that the mean of bank cost efficiency in the years 2000-2006 is 0.8721 and the average banks should reduce the costs by 12.8 percent to catch up the most efficient

associations Nonetheless, the study has not mentioned the solutions to obtain that target

Until 2011, the first study on the relation between bank efficiency and its determining variables is attempted by Gardener and his partners Nevertheless, this study is an analysis for

a group of five countries in South East Asia’, not Vietnam individually Combining the non- parametric approach (DEA) with Tobit regression model, it is found that microeconomic conditions, such as the level of capitalization, and bank private credit are significantly and

7 The analysis comprises of Malaysia, the Philippines, Indonesia, Thailand and Vietnam

positively related to bank efficiency On the contrary, the increases in bank size tend to lower bank efficiency score considerably Additionally, the coefficient of ownership dummy variable reveals that foreign banks are superior to local counterparts in efficiency In terms of macroeconomic conditions, the influence of the rate of economic growth and annual inflation rate are selected to review but the empirical results exhibit no significant correlation between these external factors and bank efficiency Arguably, Ngo (2012) focuses whole attention on the influence of market structure on Vietnamese bank efficiency but comes to contradictory findings against Gardener et al (2011) In accordance with exogenous variables studied, it is pointed out that bank efficiency is significantly affected by external factors Particularly, the nominal interest rate, government expenditures, and the concentration level of the banking system (measured by the total assets of three largest banks to all banks) are found to be positively related to bank productivity The correlations of nominal exchange rate and inflation rate with bank efficiency are significant and negative The conflicts appear between two studies although both of them employ the same methodology The disagreement in sample size may be the possible explanation for the conflicts Gardener et al (2011) is a cross-country study, whereas, Ngo (2012) is a single country paper Consequently, the findings of Gardener

et al (2011) are for the whole group As Berger and Mester (1997) claim that the study whose data set including many countries may have the difficulties in employing the best efficiency concepts and measurement techniques Furthermore, there are always the gaps in accounting standards, managerial policy, economic conditions and financial structures among different nations (Demirguc-Kunt and Huizinga, 1999) From these points of view, it implies the notion that the accuracy of the results for each member country is just at moderate level

The more details of explanatory factors in Vietnamese banking system have recently been introduced by Minh et al (2013), Vu and Nahm (2013) Once again, the level of efficiency is the main interest in both studies and they both employ DEA approach with Tobit regression model However, the control variables studied in these papers are disparate On the former paper, Minh and his collaborators hypothesize that the potential determinants of efficiency score are bank size, state ownership, capital intensity (captured by the ratio of net capital to number of labours), quality of labour (estimated by the average salary expense of each bank), and market share (measured by the ratio of each bank’s loans to the total loans of

~17~

all banks in the market) According to Minh et al (2013), it is found that bank size and market share have significant and positive effect on profit efficiency, whereas ownership structure is negatively and significantly related to productivity The latter paper, Vu and Nahm (2013) is seen as the improvement of the previous one There are more explanatory indicators to be concerned in the model and they can be divided into four following groups: (1) bank specific characteristics, (2) ownership, (3) transitional indicators, and (4) macroeconomic conditions Assessing balance panel data during the period from 2000 to 2006, Vu and Nahm (2013) find

that bank size, management quality, ownership, the development of the stock market, growth

rate of real GDP per capita, interest margin have strong positive impact on bank profit efficiency By contrast, the empirical results also indicate that the other indicators such as capital adequacy, asset quality, and inflation rate greatly affect profit efficiency in negative way The advantage of Vu and Nahm (2013) is that the analysis covers not only a large number of familiar factors from previous studies but also takes into account the effect of some typical factors, such as transitional variables including reform effect and international

commitments effect In which, the international-commitments variable is found to have

significant and negative impact on productivity Furthermore, Vu and Nahm (2013) divide capital adequacy ratio and ownership indicator into lesser extent so that they can gain more efforts for these variables The demand for more attention on both determinants results from the assumptions that the relationship between the level of capitalization and profit efficiency is significant but it may be non-linear, the different origins of foreign banks can lead to dissimilarity in profits (Vu and Nahm, 2013) Interestingly, it is pointed that capitalization does not always negatively affect bank productivity If bank managers maintain the capital adequacy ratio between 4% and 14%, they even gain more profit efficiency In terms of ownership structure, not all of foreign banks have higher profit efficiency than their counterparts but only foreign banks from the developed countries, for example the U.S, Europe, Australia and Japan banks

In brief, it is highly likely that bank performance in Vietnam is examined on the basis of cost and profit efficiency The number of researches on this subject is fairly insufficient as it has just been attracted since last decade While the majority of papers still focus on the estimation of efficiency score, there are only a few studies relating to the reasons of changes in

the efficiency As far as I am concerned, the most relevant study assessing the determinants of bank performance in Vietnam is merely Vu and Nahm (2013) On this observation, the response variable is also profit efficiency In other words, there has not been any literature on the determinants of bank performance through financial accounting ratios yet As the

experience from other countries mentioned above, it is worthwhile and feasible to examine the

correlation of bank productivity and its explanatory factors in Vietnam Table 1 below will recap the main points of studies which have been reviewed so far

2.3 Summary and Remarks

To sum up the whole section, the relationship between bank performance and its potential determinants have been conducted for a few decades (Spathis et al., 2002) Starting from the observations of scale economies in the U.S banks to assess the correlation of bank efficiency with inputs, outputs data, the earliest studies illustrate the correlation between bank size and bank performance Recently, this relation has been observed in more details with a wider range of explanatory variables and more appropriate methodologies A review of literature has shown that bank productivity is usually measured by either bank efficiency or profitability ratios Based on the response variable studied, the methodologies employed differ

from each other For example, to assess the relation of bank efficiency and its control factors,

parametric (SFA) or non-parametric (DEA) and Tobit regression model are generally applied, whilst the GMM estimation or OLS approach is preferable technique to examine the impact of various factors on bank profitability However, the control variables seem similar no matter which dependent variable explored From various studies, it is widely accepted that both internal factors, captured by bank-specific characteristics and external factors, identified by industry-specific features together with macroeconomic conditions have effects on bank profitability or efficiency In particular, the prevalent endogenous indicators are bank size (measured by total assets or deposits), the level of bank capitalization, assets quality, credit risk, fund source, management capacity, and bank ownership The exogenous indicators usually perceived are the real rate of gross domestic product per capital, annual inflation rate,

real interest rate, and stock-market capitalization level For further details, the next chapter

will introduce the methodology employed and the explanatory indicators studied

~ 19 ~

Chapter Three METHODOLOGY AND DATA 3.1 Methodology and Econometric Model

This chapter is going to present three main points: first, briefly reviewing various methodologies employed in recent studies to choose the most appropriate technique Second, establishing the econometric model applied in the analysis In the final part, the data set and the complete model included selected variables are performed

This study is employing the balanced panel data to assess the determinants of bank profitability A large number of papers, such as Goddard et al (2004b), Alper and Anbar (2011), and Hoffmann (2011) have considered panel data as the most suitable tool According

to Vong and Chan (2009), this technique is preferable to other estimations because of two main reasons Firstly, panel data provide more informative facts as the sample comprises of both cross-sectional and time-series data Put another way, it is believed that panel data model

is effective to identify a common group of characteristics and to take into account the heterogeneity The second advantage of this technique is that it allows analysts to evaluate separately the influence of microeconomic factors on bank productivity Then, industry- specific factors and macroeconomic conditions can be examined in combination with less collinearity among variables and greater efficiency (Vong and Chan, 2009)

As mentioned in the preceding chapter, the reasonable methodologies employed to examine the influencing factors of bank profitability are the GMM estimation and the OLS technique A review of literature also shows that the results from both approaches can be similar to each other (Raza et al., 2013) or contrasting to each other (Hoffmann, 2011) Following that, the comparison between these techniques is essential before proceeding to the details of the empirical analysis It is quite probable that there is no dominant model to approach the topic Each method consists of both advantages and disadvantages (Hansen,

1982; Stock and Wright, 2000; Baum et al., 2003; Roodman, 2006)

~23~

In terms of GMM technique, it is defined as:

[a] part of a broader historical trend in econometric practice toward estimators that fewer assumptions about the underlying data-generating process and use more complex techniques to isolate useful information

(Roodman, p.16, 2006)

On the one hand, the advantage of this estimator has been implied in a great number of studies is the possibility to deal with potential endogeneity of regressors (Goddard et al.,

2004a; Athanasoglou et al., 2008; Dietrich and Wanzenried, 2011; Hoffman, 2011; Raza et al.,

2013) It is widely believed that there is likely endogeneity issue in the nature of the economic relationships being tested So as to avoid the problem, GMM method is introduced as a highly reliable tool (Tregena, 2009).To support for the usage of this methodology, it is declared that the GMM estimator is particularly optimal in the following cases: (1) “small T, large N” panels, in which there are a large number of observations but only a few time periods covered; (2) a linear functional relationship; (3) dynamic single left-hand-side variable depending on its own past realizations; (4) non strictly endogenous independent variables, meaning correlated with past and possibly current realizations of the error; (5) fixed effects model; and (6)

heteroskedasticity and autocorrelation within individuals but not across them (Roodman,

2006) On the other hand, the GMM system consists of disadvantages, such as it may require a large amount of samples to obtain reasonable estimates of fourth moments (Newey, 1985; Stock and Wright, 2000) or it is too sophisticated and easily generates invalid estimates

(Roodman, 2006)

In regard to OLS approach, Baum et al (2003) declare that this technique perform ideally under conditional homoscedasticity and the assumption that all the regressors are exogenous Short (1979) and Bourke (1989) posit that the linear regression approach is not only simple to apply but also provides the reasonable results In comparison between OLS and other techniques, such as GMM estimator, multivariate regression, Attanasio et al (2000) assert that even simple OLS model may be preferable if the sample size is big enough In support of this, Xue and Harker (1999), Guru et al (2002) also assert that the linear regression analysis is an appropriate technique to investigate the relation between bank profitability and its determinants By contrast, Bond et al (2001) point out that the disadvantage of OLS

estimation is that it may exclude the variables which are correlated with one of the regressors Although OLS is regarded as a flexible form of GMM, Baum et al (2003) find that it is efficient in homoscedastic conditions rather than heteroskedastic estimations The main points

of GMM and OLS techniques are summarized in the table as follows:

Table 2: The comparison between GMM and OLS estimators

the parameter of interest (Levine, 2004)

— a superior estimation in the presence of

heteroskedasticity cases (Baum et al.,

~25~

the relationship among variables (Xue and

— OLS is a flexible form of GMM estimator