# Liquidity Has Been Known As The Ability Finance Essay

## 1.0) Introduction

Liquidity has been known as the ability of bank to fulfill their financial obligations when they come due. Basel Committee on Banking Supervision (1997) defined liquidity risk as the inability of bank to decrease in liabilities or fund up assets. A bank with insufficient liquidity is unable convert its asset into fund promptly without a huge reduction on the assets price.

A bank which maintaining high liquidity is always welcomed by regulators as the illiquidity of bank assets has been considered as causes of banking failure (Financial Services Authority, 2002). Indeed, the increases in liquidity make banks less vulnerable to liquidity shock and therefore promote the creditworthiness of banks to higher level. A higher creditworthiness of bank could consider as an advantage to attract more demand deposit and accept more loans.

Nevertheless, an earlier study of Wagner (2007) provided that an increasing in liquidity of banks not only change their stability but change their behavior toward risk taking as well. The increasing in liquidity does improve the stability and profit of banks initially. However, it does not affect the probability of default as overall. The main reason is when a bank increases liquidity, it make a crisis less costly for bank as well. A higher liquidity could be considered as reduction in the discount if bank loan need to be sold in crisis. Therefore, banks tend to increase risk taking in order to gain higher profit. Meanwhile, the high stability advantage has been offset by new risks.

Diamond and Rajan (1999) explained the bank role of bank as a liquidity provider to the depositor through liquidity creation. Liquid creation is essential in order to fulfill demand of liquidity by depositor at an inconvenient time and prevent a fire sale due to inadequate of liquidity. Banks lend out the fund it received from depositor and lend out to entrepreneur to earn interest profit. However, loans are considered as partial liquidity assets due to its low sale price before maturity. The liquidity risk arise when bank need funds to fulfill their financial obligation. In order to finance a large project with huge amount of money, bank could become intermediaries, whom called as banker. The banker can transform the illiquid loans into more liquid demand deposit (Diamond and Dybvig, 1983). The project would be financed by many banks or depositors. If any depositor has liquidity needs, the banker can refinance by issuing new demand deposits. Further, the conversion of illiquid loans into demand deposit could be considered as a shield to borrowers since it provides liquidity for both lender and borrower.

Previous studies seldom focused on the impact of liquidity towards bank performance. This research study is therefore focus on the impact of liquidity towards performance of Malaysia Commercial bank.

## 1.1) The Malaysia Banking System

According to the Banking and Financial Institutions Act 1989 (BAFIA), Banking sector are licensed institution namely finance companies, commercial banks, money brokers and discount houses which supervised by Bank Negara Malaysia (BNM).

As primary source of financing domestic economy, bank sector plays a role as intermediary, accounting for about 70% of total assets of Malaysia financial system. Currently, there are more than 2000 branches of banking institutions operated across whole country. Further, there are also 14 foreign banks which do not carry banking business in Malaysia. The intention of these representatives’ offices of foreign banks is information exchange and liaison services.

Further, as complement of banking institutions in meeting all the financial needs of economy, non-bank financial institution such as insurance companies, takaful operators and development financial institutions have been established. Currently there are more than 800 insurance and takaful operators’ offices and 100,000 resisted agents in this country (According to Bank Negara Malayisa as of February 2012).

## Total

Commercial Banks

8

17

25

Islamic Banks

10

6

16

International Islamic Banks

0

5

5

Investment Banks

15

0

15

Insurers

19

17

36

Takaful Operators (Islamic Insurers)

9

3

12

International Takaful Operators

0

1

1

Reinsurers

3

4

7

Retakaful Operators (Islamic Reinsurers)

1

3

4

Development Financial Institutions

6

0

6

## Source: Bank Negara Malaysia as of February 2012

There are currently 8 local licensed commercial banks in Malaysia which are Maybank, CIMB Bank, Public Bank, Ambank, Hong Leong Bank, Affin Bank, Alliance bank and RHB bank. Maybank holds a total of RM64163 millions of market capital in financial market and becomes largest market capital possessor in Malaysia. CIMB and Public Bank are also in the list of top 10 firms with largest market capitalization in Malaysia ( Maybank , 2012).

## 1.2) Background of Studies

Many studies have been done to investigate the determinant of bank profitability. For instance, Demirgüc-Kunt and Huizinga (1999), Guru, Staunton and Balashanmugan (2002), Naceur (2003), Mamatzakis and Remoundos(2003), Kosmidou, Tanna and Pasiouras( 2005) and are some of the earlier studies which have been reviewed in chapter 2, literature review.

Most of the findings provided that liquidity risk did bring a certain degree of impact to the bank profitability. For instance, Shen, Chen, Kao and Yeh, (2009), Bordeleau and Graham (2010).Such studies inspire the fundamental idea of this study.

Even though there are many studies considered liquidity risk as a determinant of bank profitability, the lack of studies in bank liquidity and its impact to banks performance is still exist. Earlier studies mostly concentrate on the impact of liquidity risk and seldom look though the liquidity management. Further, there are no studies examined the correlation of liquidity with the performance of local owned commercial banks in Malaysia. Thus, this study fills in the gap by investigating the relationship of liquidity with bank performance of commercial banks in Malaysia.

## 1.3) Problem Statement

There are currently 8 local owned commercial banks in Malaysia. Each bank currently has impressive financial status based on their annual report. However, there was a bank run in history of Malaysia banking sector. In year 1999,120 branches of MBf Financial Berhad faced bank run due to a total RM17 billion withdrawal. In order to resolve this crisis, Bank Negara Malaysia took into control MBf Financial Berhad.

Indeed, the MBf bank run incident has given a warning to banking sector whereby bank should maintain its liquidity to prevent another bank run in Malaysia. Liquid asset such as money market securities and government bonds, have been considered as low income assets. However, these assets increase the stability of bank and therefore creditworthy. Depositors like to deposit their money in a safer bank due to risk adverse behavior. Bank need more deposit to fund their financing and investing activities. Therefore, it is essential to investigate the relationship between liquidity of bank and bank performance.

## 1.4) Research Questions

Given the problem statement as explained in the previous section, this study seeks to answer the following research question:

What is the relationship between liquidity and bank performance?

2. What are the reasons for effect of liquidity on bank performance?

## 1.5) Research Objectives

1. To find out the relationship between liquidity and bank performance.

2. To examine the reason for effect of liquidity on bank performance.

## 1.6) ORGANIZATION OF RESEARH

The research project is divided into five chapters as below,

## 1.6.1) Chapter 1: Introduction

This chapter consists of six sections which are introduction, background of studies, problem statement, research questions research objectives and organization of research. These sections provide the understanding whereby the contribution of the study.

## 1.6.2) Chapter 2: Literature Review

In this chapter, earlier studies will be reviewed in order to provide foundation and guide for this research paper. Further, limitation and recommendation from earlier studies are noted in order to minimize mistakes.

## 1.6.3) Chapter 3: Data and Methodology

This chapter provides the detail of method and data would be used in this study. Selection of methodology and data are based on adopted knowledge from previous chapter.

## 1.6.4) Chapter 4: Analysis of Data

Results that gained from regression and test are discussed in this chapter. All data will be analysis thoroughly in order to receive an accurate conclusion in the end.

## 1.6.5) Chapter 5: Conclusion

A conclusion as overall to this study is included in this chapter. The limitation and recommendation to the research are listed out as well in order to get improvement in future studies.

## 2.0 Introduction

This chapter will provide a through literature review of previous studies on the relationship between liquidity and bank financial performance. The intention of this chapter is to report relevant studies in order to provide a deep understanding to the topic and a basis of empirical framework can be conducted.

## 2.1 Liquidity as determinants of profitability

Bourke (1989) examined the internal and external determinants of bank profitability in countries or territories in Europe. The sample was 90 banks in California, New York, Ireland, Belgium, Denmark, Spain, Australia, Massachusetts, Canada, England and Wales, Holland and Norway. The time period was from 1972 to 1981. Bourke (1989) used linear function and multiple regression which had concluded as function that can produces good result by Short (1979) in earlier research to test the relationship between financial ratios and bank financial performance. The results showed that the liquidity ratio, liquid assets to total assets and loans to deposit ratio are positively related to bank profitability ratio, return on total assets.

Molyneux and Thornton (1992) replicated the earlier research by Bourke (1989) to investigate the determinants of bank profitability of European banks. The sample was banks from 18 countries in European. The time period was from 1986 to 1989. Molyneux and Thornton (1992) replicated the methodology from earlier research by Bourke (1989). The linear equation was determined by pooled time series regression on internal (capital ratios, liquidity ratios, staff expenses) and external (government ownership, interest rates, concentration ratios, inflation and government ownership) determinants of bank performance. The finding provided that liquid assets to total assets ratio had weak negative relationship with bank profitability. Molyneux and Thornton (1992) suggested that the liquidity holding especially imposed by government or authorities could be considered as a cost to the bank.

Chaudhry, Chatrath and Kamath (1995) investigated the determinants of profitability of United States commercial banks. The time period was from year 1977 to 1985 and the sample was divided into large and small bank category. There were 172 banks in large bank categories and 12809 banks in small bank category. The dependent variable was net income to total assets ratio and independent variables were 15 financial ratios. Chaudhry et al. used multivariate regressions method in this research. The finding provided that the provisions for loans losses to total assets ratio had inverse relationship with bank profitability in both categories of banks. Chaudhry ex al. gave the explanation that an increase of loan losses provision would decrease the liquidity of banks and ability of bank in generating revenue.

Demirgüc-Kunt and Huizinga (1999) examined the determinants of worldwide commercial banks profitability and interest margins. The time period was from 1988 to 1995 and the sample of the research was 7900 individual commercial banks from 80 countries. The multiple regression of this research was estimated by weighted least square pooling bank data of the sample. The empirical result showed that the loans to total assets ratio has positive relationship with net interest margin, however, a negative relationship with return on total assets ratio.

Guru, Staunton and Balashanmugan (2002) were attempted to identify the determinant of Malaysia commercial bank profitability. The sample was Malaysia commercial banks and the time period was from 1985 to 1998. Ordinary least squares regression techniques were used in this research to test the relationship of variables. Furthermore, several dummy variables were included in linear model to represent cross-sectional differences and effect of temporal towards bank profitability. The independent variables were internal determinants and external independents. Liquidity was considered as internal determinants and measured by loans to deposits ratio. The finding provided that the loans to deposits ratio had an inverse relationship with net income to total assets. Guru et al. (2002) suggested the reasons which banks might suffer from higher loans cost through high interest interbank borrowing and higher nonperforming loans. Nevertheless, the capital based profitability model regression result showed that there was a positive relationship between loans to deposits ratio and net income to shareholder capital and reserve ratio.

Barth, Nolle, Phumiwasana and Yago (2003) examined the relationship between independence of bank supervision, scope and structure toward bank performance. The sample was 2300 individual banks across 55 countries and the time period was from 1996 to 1999. The ordinary least squares regression analysis was used to test the relationship. The selecting of variables was followed earlier research by Demirgüc-Kunt and Huizinga (1999) and loan to total assets ratio is considered as one of the determinants. Furthermore, the empirical result showed that there was a negative relationship between liquidity assets to total assets ratio and return on assets. The result corresponded with previous finding. (Demirgüc-Kunt and Huizinga, 1999)

Naceur (2003) was inspired by earlier finding (Demirgüc-Kunt and Huizinga, 1999) and attempted to identify the determinants of Tunisian commercial banks for the purpose of better bank policies formulation. The sample was 10 main commercial banks in Tunisia and the time period was from 1980 to 2000. Return on assets ratio and net interest margin were measure of bank performance. The empirical result shows that the bank’s loan to total assets ratio had a positive relationship with both bank performance measures. Naceur (2003) gave the explanation that banks loans can generate more interest income thereby increasing net interest margin.

Demirgüc-Kunt, Laeven and Levine (2003) assessed the impact of banking sector concentration, institutional development and bank regulations on bank performance. The sample was 1430 banks across 72 countries and the time period was from 1995 to 1999. Specific factors, such as bank size, bank equity, bank liquidity, market share of each bank, differences in cross countries financial conditions and macroeconomic are controlled in the research. Empirical showed that the liquidity ratio, liquid assets to total assets ratio had a negative relationship with bank profitability, net interest margins. The result supported the earlier finding of Molyneux and Thornton (1992). The finding provided that liquid assets such as cash and government securities generated a lower interest income and banks that hold high fiction of these assets had lower profitability, which was also net interest margin.

Mamatzakis and Remoundos (2003) examined the determinant of performance of commercial banks in Greek. The time period was from 1989 to 2000 and the sample was 17 commercial banks in Greek. Mamatzakis and Remoundos (2003) followed the earlier work of Molyneux and Thornton (1992) in their estimation. The dependent variable in the study was return on equity and return on total asset. Liquidity ratio, loan to assets ratio was considered as independent variable in the model. The empirical result provided that the ratio of loans to assets had a significantly positive relationship with both profitability ratios. Mamatzakis and Remoundos (2003) suggested that banking system in Greek had a change towards more aggressive form of expansion based on significant increases from interest income.

Goddard, Molyneux and Wilson (2004) examined the interactions between firm growth and banks performance. The sample was 583 banks which located in France, Germany, Italy, Spain and United Kingdom. Firm growth and profit equation were estimated through dynamic panel regressions. Result from cross sectional regression showed that the liquidity ratio is negatively related with bank profitability. Goddard et al. (2004) gave the explanation that banks tend to record modest profitability on average if they maintain a high liquidity ratio. Bank profitability was affected by liquidity and riskier banks tend to generate more profit which corresponds with portfolio theory. This finding supported earlier finding of Molyneux and Thornton (1992) and Demirgüç-Kunt et al. (2003).

Kosmidou, Tanna and Pasiouras (2005) were attempted to investigate the determinants that could affect probability of UK domestic commercial banks. The time period was from 1995-2002 and the sample is 32 domestic UK commercial banks. Kosmidou et al. (2005) estimated all the models by using fixed effect regression where mean deviation data was used to eliminate firm level heterogeneity. The dependent variables were return on total assets ratio and net interest margins. The liquidity ratio used as independent variable in this research was liquid assets to customer and short term funding. The empirical showed that the liquidity ratio had a positive effect on Return on total assets ratio which consistent with earlier finding (Bourke, 1989). Nevertheless, the liquidity ratio had expected negative sign on net interest margins which supported earlier study of Demirgüç-Kunt et al. (2003). Kosmidou et al. gave the conclusion that the impact of liquidity on UK commercial bank profit was not clear cut.

Athanasoglou, Delis and Staikouras (2006) examined the determinants of bank performance in South Eastern European Region (SEE) over the time period 1998 to 2002. The sample of research was banks from seven SEE countries (Bosnia-Herzegovina, Croatia, Romania, Albania, Bulgaria, FYROM, and Serbia Montenegro). Athanasoglou et al. (2006) used generalized least squares to estimate the linear function. The result provides that the liquidity risk variable, which was also loans to total assets ratio had a weak positive impact to the return on assets and return on equity ratio. The result was consistent with earlier studies of Naceur (2003) and Mamatzakis and Remoundos (2003). Athanasoglou et al. suggested the reason that banks tend to maintain an illiquidity position in order to prevent failure since the SEE banking system was still far away from meeting the liquidity standards of developed banking system.

Vong and Chan (2006) assessed the impact of bank characteristic financial structure variables on bank performance in Macao. The time period was from 1993 to 2007 and the sample is 5 commercial banks in Macao. Linear model and panel regression technique were used in this study to produce result. Empirical result provided that the loans to total assets ratio had an inverse relationship with bank performance, which was return on assets ratio in this study. The result corresponded with earlier studies of Barth et al. (2003) and Demirgüc-Kunt and Huizinga (1999). Vong et al. (2006) gave his explanation in earlier finding (Vong et al., 2005) which was interbank placement of idle funds abroad and completion in credit market have diminished the probability of banks in Macao.

Carbó Valverde and Rodríguez Fernández (2007) attempted to investigate the determinants of bank margins in European banking. The time period was from 1994 to 2001. The sample was 19322 banks from seven countries in European countries (Spain, France, Italy, Sweden, United Kingdom, the Netherlands and Germany). The Bank margin as bank profitability in this study is defined as loan to deposit rate spread which computed by the difference between price of loans (interest income to loans ratio) and the price of deposits (interest expense to deposits ratio). The finding provided that the liquid assets to deposits ratio had a positive relationship with bank profitability. Nonetheless, the loans to total assets ratio had a negative relationship with bank margin. Carbó Valverde and Rodríguez Fernández (2007) gave the explanation which banks specialized in lending try to reduce the intermediation costs by exploiting information advantages and offer lower bank margin. The result corresponded with earlier studies of Barth et al. (2003), Demirgüc-Kunt and Huizinga (1999) and Vong et al. (2006).

Pasiouras and Kosmidou (2007) examined the specified characteristics of commercial bank which could affect the profitability among domestic and foreign commercial banks in fifteen European countries. The time period was from 1995 to 2001 and the sample was 584 commercial banks in European Union. The linear equation was estimated by panel cross-sectional regression. The result provided that liquidity risk, net loans to customer and short term funds had a positive relationship with profitability of domestic bank. A higher liquidity would reduce bank profitability. The result supported earlier studies of Molyneux and Thornton (1992), Guru et al. (1999) and Athanasoglou et al. (2006). Nonetheless, the liquidity risk showed a negative relationship with profitability of foreign banks. The result was in accordance with earlier finding. (Bourke, 1989)

Fungacova and Poghosyan (2009) examined the determinants of bank profitability in Russia. The profitability in this study was specified to bank interest margin. The time period is start 1999-2007. Fungacova and Poghosyan (2009) was considered one of a determinant would be liquidity and used the ratio of liquid assets to demand liabilities. In order to evaluate the impact of determinants, fixed effects estimator had been used in regression model. The empirical result showed that liquidity ratio consistent with expected impact which was a negative relationship with net interest margin. The finding was in accordance with earlier studies of Bourke (1989) and Kosmidou et.al (2005).

Ben Naceur and Kandil (2009) assessed the impact of capital adequacy regulations on the profitability and cost of intermediation of banks in Egypt. The sample was 28 banks in Egypt and the time period was 1989 to 2004. Ben Naceur and Kandil (2009) defined the measure of intermediation cost as net interest margin and profitability as return on assets and return on equity ratios. The empirical result showed that the liquidity risk ratio, net loans to customer and short term fund had a positive relationship with net interest margin whereby an inverse relationship between liquidity with bank profitability. The result was consistent with earlier studies of Molyneux and Thornton (1992), Guru et al. (1999), and Pasiouras and Kosmidou (2007). Nevertheless, there was no significant relationship between liquidity risk ratio with return on assets and return on equity ratios. Ben Naceur and Kandil (2009) suggested that excess liquidity could force banks to lower the net interest margin since they try to reduce non-earning assets.

Shen, Chen and Kao (2009) examined the relationship between bank liquidity risk and performance. The sample was banks from 12 countries which are United States, United Kingdom, Taiwan, Switzerland, Netherlands, Australia, Luxembourg, Japan, Canada, France, Germany and Italy. The time period was from 1994 to 2006. The dependent variables of the studies were net interest margin, returns on average assets and return to average equity ratios. The panel variables regression model has been used to estimate the linear function. The empirical results provide that both risky and less risky liquid assets to total assets ratio had negative relationship with liquidity risk. Further, the finding also revealed that liquidity risk was negatively related to the return on average assets and return on average equity ratio. Shen et al. (2009) gave the explanation that in order to maintain stability, banks might need to hold more liquid assets or gain much external funding to meet the demand of fund. Since borrowings raised, it would brought down the creditworthiness of bank, and therefore increased its cost of funding. Nevertheless, the finding also provided that the liquidity risk had a positive relationship with net interest margin. Shen et al. (2009) suggested the reason that a high level of illiquid assets in loans can produced higher interest income.

Bordeleau and Graham (2010) attempted to investigate the impact of liquidity on bank performance. The studies focused on the impact of liquidity assets holding to bank profitability. The time period was from 1997Q1 to 2004Q4 and the sample was 55 banks from U.S and 10 banks from Canada. A nonlinear regression was used in this study to estimate the equation. The dependent variables in the studies were return on assets and return on equity ratios. The finding showed that the relationship between liquid ratio and bank probability took a form of downward-concave parabola, which provided the meaning that profitability of banks was increased by certain amount of liquid assets holding, but a further liquid asset holding would diminished a bank profitability. Bordeleau and Graham (2010) suggested that banks should hold a higher level of liquidity when they maintain a less traditional business model and in the time of weak economic growth.

Olson and Zoubi (2011) examined the profitability of banks in Middle East and North Africa (MENA) accounting based and economic based measurement. The sample was 10 banks in MENA and the time period was from 2000 to 2008. The dependent variables in the study are return to assets and return to equity ratio. The liquidity risk ratio which considered as determinant of profitability was net loans to total assets ratio. Generalized least squares panel estimator has been used in this study to estimate the return on assets and return on equity equations. The empirical result provides that the liquidity risk ratio, net loans to total assets ratio had a positive relationship with bank profitability. The finding was consistent with earlier finding of Naceur (2003), Mamatzakis and Remoundos (2003) and Athanasoglou et al. (2006). Olson and Zoubi (2011) gave the explanation that the returns from loan was much higher other assets.

Staikouras and Wood (2011) attempted to investigate the determinant of profitability of European banks. The sample included 685 banks in European and the time period is from 1994 to 1998. In order to estimate the linear equation in the studies, cross sectional time series regression had been adopted. The dependent variable of the study was return on total assets. Loans to assets ratio was used as one of the independent variable in the model. The empirical results provided that a loan to assets ratio had a positive relationship with return on return on asset ratio. The finding was consistent with earlier studies of Naceur (2003), Mamatzakis and Remoundos (2003) and Athanasoglou et al. (2006).

Lin, Chung, Hsieh and Wu (2012) examined the determinants that could affect interest margin of bank in Asian and their impact on bank diversification. The sample was 262 banks in China, Indonesia, Japan, India, South Korea, Taiwan, Singapore, Thailand and Philippines. The time period was from 1997 to 2005. The dependent variable in this study was net interest margin and the liquidity ratio which considered as independent variable was liquid assets to total liabilities ratio. The cross-sectional model regression results provided that the liquidity ratio had a positive relationship with net interest margin. Further, the coefficient of liquidity ratio was significant and higher in low diversification regime compared to high diversification regime. This indicated that the shock of liquidity risk could be reduced by diversification.

## 2.2) Conclusion

From the literature review done related to this topic, it is obvious that liquidity has affected the profitability of bank in many countries. The knowledge gained form earlier findings will be applied into this study in order to minimize the problems might be faced in subsequent chapters.

## 3.0) Introduction

In this chapter, method used in order to find out the relationship between liquidity and performance of local owned commercial bank will be discussed. This chapter is separated into few sections. Section 3.1 describes the source of data and hypothesis development in detail. Section 3.2 displays the layout of relationship between liquidity and bank performance. Section 3.3 describes the ratios used as dependent and independent of variables in this research study. Section 3.4 provides the estimation techniques used in this study. Section 3.5 explains the estimation test will be used in next chapter. Last but not least, section 3.6 as the conclusion for this chapter.

## 3.1) Source of data

In order to carry out this research study, total 8 local owned commercial banks in Malaysia were selected. The 8 local owned commercial banks are Affin Bank, Alliance Bank, AM Bank, CIMB Bank, Hong Leong Bank, Maybank, Public Bank and RHB Bank. The data used in this research study is secondary data, where it can be obtained from statement of financial position and profit and loss statement in annual report. The time period for this research study is 10 years, start from 2002 to 2011. The data taken out from annual reports is used to calculate the dependent and independent variables.

## 3.1.1) Hypothesis development

Hypothesis is defined by Sekaran and Bougie (2010) as a tentative, yet testable statement, which predicts what expected to be found in the data. The hypothesis is normally what researcher believes to be the outcome of the study. Both dependent and independent variables will be included in hypothesis statement. They are 2 types of theories in a hypothesis that can be tested, null and alternative hypothesis. A null hypothesis (H0) is believed to be true since it is used as fundamental for statement, but has not been proved. The null hypothesis is always test being tested indirectly where the conclusion is either reject H0 or do not reject H0. Alternative hypothesis will only be accepted if H0 is rejected.

## Hypothesis 1

H0: There is no relationship between ratio of loans as percentage of deposit (LOD) and ratio of return on assets (ROA).

H1: There is a relationship between ratio of loans as percentage of deposit (LOD) and ratio of return on assets (ROA).

## Hypothesis 2

H0: There is no relationship between ratio of liquid assets to total deposit (LAD) and ratio of return on assets (ROA).

H1: There is a relationship between ratio of liquid assets to total deposit and ratio (LAD) of return on assets (ROA).

## Hypothesis 3

H0: There is no relationship between ratio of total loans to asset (LOA) and ratio of return on assets (ROA).

H1: There is a relationship between ratio of loans to asset (LOA) and ratio of return on assets (ROA).

## Hypothesis 4

H0: There is no relationship between ratio of total loans to equity (LOE) and ratio of return on assets (ROA).

H1: There is a relationship between ratio of loans to equity and ratio (LOE) of return on assets (ROA).

## Hypothesis 5

H0: There is no relationship between ratio of cash on equity (COE) and ratio of return on assets (ROA).

H1: There is a relationship between ratio of cash on equity (COE) and ratio of return on assets (ROA).

## Hypothesis 6

H0: There is no relationship between ratio of loans as percentage of deposit (LOD) and ratio of return on equity (ROE).

H1: There is a relationship between ratio of loans as percentage of deposit (LOD) and ratio of return on equity (ROE).

## Hypothesis 7

H0: There is no relationship between ratio of liquid assets to total deposit (LAD) and ratio of return on equity (ROE).

H1: There is a relationship between ratio of liquid assets to total deposit (LAD) and ratio of return on equity (ROE).

## Hypothesis 8

H0: There is no relationship between ratio of total loans to asset and ratio (LOA) of return on equity (ROE).

H1: There is a relationship between ratio of loans to asset and ratio (LOA) of return on equity (ROE).

## Hypothesis 9

H0: There is no relationship between ratio of total loans to equity and ratio of return on equity (ROE).

H1: There is a relationship between ratio of loans to equity and ratio of return on equity (ROE).

## Hypothesis 10

H0: There is no relationship between ratio of cash on equity (COE) and ratio of return on equity (ROE).

H1: There is a relationship between ratio of cash on equity (COE) and ratio of return on equity (ROE).

## Liquidity

Bank performance Ratio of loans as percentage of deposit (LOD)

Return on Assets (ROA) Ratio of liquid assets to total deposit (LAD)

Ratio of total loans to assets (LOA)

Ratio of total loans to equity (LOE)

Ratio of cash on equity (COE)

## Liquidity

Bank performance Ratio of loans as percentage of deposit (LOD)

Return on Equity (ROE) Ratio of liquid assets to total deposit (LAD)

Ratio of total loans to assets (LOA)

Ratio of total loans to equity (LOE)

Ratio of cash on equity (COE)

## Models

1) ROA = f (LOD, LAD, LOA, LOE, COE)

2) ROE = f (LOD, LAD, LOA, LOE, COE)

## 3.3.1) Dependent Variables

1) Return on assets (ROA)

Return on assets (ROA) indicates the profitability of bank based on the earning revenue of financial year by its total assets. Return on assets is calculated by dividing net incomes of the bank by its total assets. Earlier studies such as Bourke (1989) and Demirgüc-Kunt and Huizinga (1999) had used ROA as dependent variable in their studies.

2) Return on equity (ROE)

Return on equity (ROE) indicates the profitability of a firm based upon the stockholders’ investment. Return on equity could be calculated by dividing net income for the financial year by common stockholder equity. ROE has been considered as dependent variable to represent profitability in some earlier studies of Athanasoglou et al. (2006) and Shen et al. (2009).

## 3.3.2) Independent Variables

1) Ratio of loans as percentage of deposit (LOD)

Ratio of Loans as percentage of deposit (LOD) indicates the proportion of bank loans are funded by deposits. The higher the ratio, the difficulties that bank might suffer from unexpected deposit withdrawals will be greater since loans are not liquid (Gonzalez-Hermosillo, 1999).

2) Ratio of liquid assets to total deposit (LAD)

Ratio of Liquid assets to total deposit (LAD) indicates the ability of bank to meet unexpected deposit withdrawals with liquid assets (Calomiris and Mason, 1997). Bank with higher LAD has better ability to meet the unexpected deposit withdrawals with liquid assets. Cash is not included as liquid assets in this research paper as it would be measured through ratio of cash on equity (COE).

3) Ratio of total loans to assets (LOA)

Ratio of total loans to asset (LOA) is a measure of illiquidity of asset portfolio (Arena, 2005).The higher this ratio indicates the bank is loaned up and has a low liquidity. Earlier studies such as Carbó Valverde and Rodríguez Fernández (2007) and Olson and Zoubi (2011) used LOA as determinant to measure liquidity risk of bank.

4) Ratio of total loans to equity (LOE)

Ratio of total loans to equity (LOE) is a measure of illiquidity of stockholders’ investment. Total loans to equity could be calculated by dividing total loans for the financial year by common stockholder equity. The higher the LOE, the more risky a bank is considered.

5) Cash on equity (COE)

As opposite to LOE, Cash on equity (COE) is a measure of liquidity of shareholder’s investment. COE could be calculated by dividing cash for the fiscal year by common stockholder equity. Since cash is the most liquid asset in banks, a high COE indicate a more liquid of bank is considered. Further, cash could be distributed to shareholder by lightly taxed method such as cash finance acquisition or share repurchase program (Bagwell and Shoven, 1989).

## 3.4) Estimation of Technique used

This research study is descriptive and historical since it seeks to describe the moves of liquidity of local owned commercial bank from year 2002 to 2011. Method used for sampling is non-probability method whereby banks are selected based on few criteria:

i. The commercial bank is listed under Bursa Malaysia.

ii. The commercial bank is Malaysia local owned basis.

iii. The commercial bank has complete set of data or annual report over time period of 2002 to 2010.

There are total 8 commercial banks fulfill criteria and selected as sample in this research study. By following Bourke (1989) and Molyneux and Thornton (1992), a linear function model will be used in order to investigate the relationship between liquidity and bank performance. Ratio of loans as percentage of deposit (LOD), ratio of liquid assets to total deposit (LAD), ratio of total loans to assets (LOA), ratio of total loans to equity (LOE), and ratio of cash on equity (COE) are considered as indicator of liquidity in this study. Bank performance is indicated by return on assets (ROA) and return on equity (ROE).

## Expected Sign

Return on asset(ROA)

Return on equity(ROE)

Loans as percentage of deposit (LOD)

## Negative (-)

Liquid assets to total deposit (LAD)

## Negative (-)

Total loans to assets (LOA)

## Positive (+)

Total loans to equity(LOE)

## Positive (+)

Cash on equity (COE)

## Negative (+)

The cross sectional time series data was pooled for the purpose of analysis. Data was analyzed through correlation and multiple regression models. Linear equations of this study are estimated by using Ordinary Least Square (OLS) method.

The Linear equations for bank performance are,

ROA= a0 + a1LOD + a2LAD + a3LOA + a4LOE + a5COE + e and,

ROE= a0 + a1LOD + a2LAD + a3LOA + a4LOE + a5COE + e

Where,

ROA = Return on asset

ROE = Return on equity

LOD = Loans as percentage of deposit

LAD = Liquid assets to total deposit

LOE = Total loans to equity

COE = Cash on equity

a 0-5 = Coefficients

e = Error term

Regression model has been used in this study in order to estimate the relationship between variables and the impact of explanatory variables to the response variables.

## 3.5) Result Specifications

Multiple Linear Regression (MLR) which contains several explanatory variables has been used in this study. In order to examine the characteristic and behavior of explanatory variables in this study, several tests have been applied.

## 3.5.1) Mean, Median and Standard Deviation

Mean is the average or expected value of a set of data and is calculated as sum of all data divided by the number of data. Median is the middle score of data. It is free from outlier data which is typically large or small values. Standard deviation measures the dispersion of data from its mean. A wider spread of data has a higher standard deviation. In finance, it indicates the volatility of investment as well (Tan, 2007). Standard deviation can be computed by

http://www.ltcconline.net/greenl/courses/201/descstat/mean.h3.gif

Where n = Number of data,

x̄ = Mean

x = Explanatory variable

## 3.5.2) Skewness and Kurtosis

Skewness is an indicator used to measure the degree of asymmetry of distribution from normal distribution. Based on skewness, the skew of distribution can be separated into 3 types,

i. When skewness is positive, it indicates a right skewed distribution, where data are concentrated on left of mean.

ii. When skewness is negative, it indicates a left skewed distribution, where data are concentrate on right of mean.

iii. When skewness equal to zero, it indicates median, where represent a symmetrical distribution.

Kurtosis is another indicator of distribution. A positive kurtosis indicates a heavy tails and peaknedness distribution while a negative kurtosis indicates a light tail and flatness distribution (DeCarlo, 1997). An "excess kurtosis" is used when interprets kurtosis and the interpretation are,

i. If excess kurtosis is equal to 0, the distribution is mesokurtic which similar to normal distribution.

ii. If excess kurtosis is positive value, the distribution is leptokurtic which indicates the central peak is sharper and tails are fatter compare to normal distribution.

iii. If excess kurtosis is negative value, the distribution is platykurtic which indicates the central peak is broader and the tail is thinner compare to normal distribution.

## 3.5.3) Correlation Coefficient and P-value

Correlation Coefficient is a number between -1 and 1 which measure the strength and direction of linear relationship between 2 variables. Correlation coefficient more than 0 indicates a positive linear correlation between 2 variables and a negative correlation coefficient indicate a negative linear correlation. A 0 correlation coefficient indicate there is no linear correlation between 2 variables.

P-value is probability which measure how likely a null hypothesis can be rejected when the hypothesis is true. The decision rules of P-value approach are,

i) Do not reject H0 if p-value is larger than significance level,α.

ii) Reject H0 if p-value is smaller than significance level,α.

## 3.5.4) Test for Normality of Explanatory Variables (Jacque-Bera test)

Jacque-Bera test is a normality test which used to measure the goodness–of-fit of data set when matching a normal distribution. The calculated test statistic is compared with critical values in chi square distribution with 2 degree of freedom at 0.1, 0.05 and 0.01 significance level.

The hypothesis of this test,

H0: Errors are normally distributed.

H1: Errors are not normally distributed.

It can be computed as

\mathit{JB} = \frac{n}{6} \left( S^2 + \frac14 (K-3)^2 \right)

Where

n = Number of observations

S = Sample skewness

K = Sample kurtosis

## 3.5.5) Test for Autocorrelation (Durbin Watson Test)

In statistics, Durbin Watson tests the presence of serial autocorrelation in the residuals. (Agung, 2009) The presence of autocorrelation in the regression could underestimate the true variance of model. For the purpose of remove autocorrelation, the number of observation could be added and the missing values needed to be identified. The hypothesis which considered in Durbin Watson Test is,

H0 : Residuals are not auto correlated.

H1: Residuals are auto correlated.

If statistic value is less than 2, there is positive serial correlation.

## 3.5.6) T-test

T-test is a statistic that checks if two means are reliably different form each other. T-test also shows if there is significant contribution of explanatory variable to response variable (Agung, 2009).There are 3 types of T-test, independent samples, paired samples, and one sample. The T- value can be computed as the difference between group means divided by variability of groups. The hypothesis statement for T-Test is,

H0: The explanatory variable has no significant contribution on response variable.

H1: The explanatory variable has significant contribution on response variable.

## 3.5.7) F-Test

F-Test is normally done in order to test the overall significant of a model. The result of F-Test shows if there is a linear relationship between all of the explanatory variables considered together with response variable (Agung, 2009).The hypothesis statement for F-Test is,

H0: All the explanatory variables cannot explain variation in response variable.

H1: At least one explanatory variable can be explained variation in response variable.

## 3.5.8) Model Fitness Test (R2 and Adjusted R2)

The coefficient of the determination, R-Squared (R2) provides the proportion of fluctuation that is accounted for by a data set (Agung, 2009). R2 is normally interpreted as goodness of fit of an Ordinary Least Square Regression (OLS). R2 has a range from 0 to 1, which indicates the strength of linear association between explanatory and response variables. An R2 near 1 indicate a high percentage of data that is closest to the line of best fit. In order to select best model from few models, R2 will be compared. However, if the models have different number of explanatory variables, adjusted R2 will be selected as measurement to select best model.

## 3.6) Conclusion

The study in this chapter shows that various statistic and econometric techniques will be adopted in this study in order to investigate the relationship between liquidity and local owned commercial bank in Malaysia.

## 4.0) Introduction

In this chapter, the result of finding and investigation on relationship between liquidity and bank performance will be presented. The chapter is separated into few sections in order to show the results of regression and tests.

## TABLE 4.1: Descriptive Statistics.

COE

LOA

LOD

LOE

ROA

ROE

Mean

1.888347

0.255558

0.608129

0.777030

8.212852

0.008972

0.116540

Median

1.708505

0.252687

0.616470

0.786080

7.670017

0.009654

0.115920

Maximum

4.337355

0.416560

0.778660

1.034760

18.57780

0.015940

0.275010

Minimum

0.272680

0.125330

0.448309

0.508631

4.451190

-0.008600

-0.115800

Std. Dev.

0.810375

0.067850

0.071450

0.106651

2.441263

0.004153

0.057597

Skewness

1.135291

0.481619

-0.188587

-0.330451

1.737265

-1.161665

-0.408219

Kurtosis

4.036628

2.996462

2.878787

3.153493

7.610784

5.524485

5.736876

Based on the table above, the mean of COE is 1.888347. It means that the average number of all COE is 1.888347. However, the mean is easily affected by typical outcome such as too big or too small value. Therefore, the comparison with median is essential. The median of COE is 1.708505 which is much lower than mean. Typically high value might be the reason of the large difference. The maximum outcome of COE is 4.337355 and the minimum outcome is 0.272680. The range between maximum and minimum value is huge compare with other ratios. The huge difference is due to high cash holding by Hong Leong Bank and relatively low COE of Alliance Bank in year 2011. The standard deviation of COE is 0.810375. Since standard deviation measures how far the data vary from mean, it is given that most of the outcomes are lie between 1.077972 and 2.698722. The skewness of COE is 1.135291 which represent that COE has right skewed distribution. Further, the kurtosis of more than 3 means COE has leptokurtic distribution.

The mean of LAD is 0.255558. There is only a least difference when compare with its median, 0.252687. This indicates that the mean of LAD is less affected by outlier or typical value. Further, the maximum value of LAD outcome is 0.416560 and the minimum value is 0.125330. The range between maximum and minimum outcome is relatively lower if compare with COE and LOE. The reason of less difference is due relatively low LAD for all banks, which is lower than 0.5. The low LAD indicates that all banks have insufficient liquidity asset to cover their deposit. This might be due to liquid assets bring lower return than other assets (Demirgüç-Kunt et al., 2003). The standard deviation LAD is 0.067850. Based on its standard deviation, most of the outcomes are between 0.187708 and 0.323408. The skewness of LAOD is 0.481619 which indicates it has slightly right skewed distribution and the kurtosis of nearly 3 indicates the distribution of LAD is mesokurtic distribution. A mesokurtic distribution is identical to normal distribution.

LOA has a mean of 0.608129 and median 0.616470. The relatively low difference between mean and median indicate the outcome of LOA has no outlier. Moreover, the maximum value of LOA is 0.778660 and minimum value is 0.448309. Again, the difference between maximum and minimum value is much lower compare to COE and LOE. This lower difference is another evidence that no outlier in LOA. The mean of LOA, 0.608129 indicate that banks lend out more than half of their total to earn interest revenue. This also indicates that bank might suffer from liquidity risk since most of their assets are illiquid assets, which are loans. Since the standard deviation of LOA is 0.071450, therefore most of the outcomes are between 0.536679 and 0.679579. The negative skewness of LOA indicates that it has left skewed distribution and kurtosis of 2.878787 indicates the distribution of LOA is platykurtic. A platykurtic distribution is flatter than a normal distribution.

The mean of LOD is 0.777030 which is nearer to median 0.786080. The less difference between mean and median indicate the outcome of LOD has less or no outlier. The maximum value of LOD outcome is 1.034760 and the minimum value is 0.508631. The range between maximum and minimum value is small since there is no outlier in the data. Based on the table, most banks have ratio of LOD lower than 1. A lower LOD indicates banks have lower liquidity risk. The standard deviation LOD is 0.106651 hence most of the outcomes are between 0.670379 and 0.883681. Since LOD has negative skewness, therefore the distribution of LOD is left skewed. The more than 3 kurtosis indicate LOD has leptokurtic distribution. A leptokutic distribution is sharper than normal distribution.

According to table 4.1, the mean of LOE is 8.212852. However, the median of LOE is 7.670017. It is a pretty large difference between median and mean, therefore the data of LOE might contain typical outcome which has relatively large or small value. Further, the maximum value of LOE outcome is 18.57780 and minimum value is 4.451190. The range between maximum and minimum value is relatively larger when compare with other ratios. This is due to high loans holding by Alliance bank in year 2002. Since the standard deviation of LOE is 2.441263, most of the outcomes are between 5.771589 and 10.654115. The skewness of LOE is 1.737265 and it indicates the distribution of LOE is right skewed distribution. Moreover, a high kurtosis, 7.610784 also indicates distribution of LOE is leptokurtic distribution which means a high probability for extreme values.

There is a large difference between mean and median of ROA. The mean is 0.008972 and median is 0.009654. The maximum value of ROA outcome is 0.015940 and the minimum value is -0.008600. The large difference between maximum value and minimum value is due to negative value are presented in the data. Alliance bank suffers with huge loss of RM201810000 in year 2006 and this cause a negative ROA in the data. Since the standard deviation of ROA is 0.004153, therefore most of the outcomes are between 0.004819 and 0.013125. The negative skewness indicates the distribution of ROA is left skewed distribution and a kurtosis higher than 3 indicate it is leptokurtic distribution.

Last but not least, the mean of ROE is 0.116540 and it is close to median which is 0.115920. Since the difference between mean and median is small, there is few or no outlier in the data. The maximum value and minimum value of ROE is 0.275010 and -0.115800. Although the minimum value of ROE is negative, the overall difference between maximum and minimum value is still considered small. The standard deviation of ROE is 0.057597 and therefore most of the outcomes are between 0.058943 and 0.174137. Since the skewness of ROE is negative, hence the distribution of ROE is left skewed. However, a kurtosis of 5.736876 which more than 3 indicate the distribution is leptokurtic distribution as well.

Correlation

Probability

1.000000

-0.292742

0.0084

1.000000

-0.188267

0.0944

0.468558

0.0000

1.000000

-0.014568

0.8979

0.521121

0.0000

0.895339

0.0000

1.000000

-0.395057

0.0003

-0.142402

0.2076

-0.167466

0.1376

-0.402871

0.0002

1.000000

0.074890

0.5091

0.129720

0.2514

-0.624004

0.0000

-0.551974

0.0000

-0.199565

0.0759

1.000000

## -----

Correlation coefficient is a measure to indicate the correlation between explanatory and response variables. In order to get more accurate answer, p-value approach with 0.05 and 0.10 significant levels would be used to test the significant of the relationship.

Based on the table 4.2, the correlation coefficient between ROA and LOE is -0.292742 and the p-value is 0.0084. The p- value is much lower than both 0.05 and 0.10 significant level, therefore H0 of p-value approach is rejected and there is a significant linear relationship between ROE and LOE. Since the correlation coefficient is a negative value, therefore, that is a negative linear correlation and ROA is reduced when every time LOE is increased.

The correlation coefficient of LOD and ROA is -0.188267. However, the p-value of 0.0944 only passes the p-value test with 0.10 significant level. This indicates there is only weak linear relationship between ROE and LOD. The negative correlation coefficient indicates that is inverse correlation where ROA is reduced every time LOD is increased.

Moreover, the correlation coefficient of LOA to ROA is -0.014568 and p-value is 0.8979. The large p-value of over 0.5 and 0.10 indicates that, H0 of p-value approach should not be rejected. Therefore, there is no linear relationship between ROE and LOA.

The correlation coefficient of -0.395057 indicates there is a negative linear correlation between ROA and LAD. It means that ROA would drop 0.395057 every time LAD is increased by 1. Further, the p-value of 0.0003 indicates the H0 of p-value approach should be rejected at both significant levels. There is a significant linear relationship between LAD and ROA.

Last but not least, the correlation coefficient of ROA to COE is 0.074890. Based on this result, there is a positive linear correlation between ROA and COE. However, the high p-value of 0.5091 indicates that H0 of p-value approach should not be rejected and there is no significant linear relationship between ROA and COE. Besides, the H0 for hypothesis 5 is failed to be rejected.

Correlation

Probability

1.000000

0.083889

0.4594

1.000000

-0.215734

0.0546

0.468558

0.0000

1.000000

-0.028273

0.8034

0.521121

0.0000

0.895339

0.0000

1.000000

-0.399285

0.0002

-0.142402

0.2076

-0.167466

0.1376

-0.402871

0.0002

1.000000

0.355234

0.0012

0.129720

0.2514

-0.624004

0.0000

-0.551974

0.0000

-0.199565

0.0759

1.000000

## -----

According to result above, the correlation coefficient of LOE to ROE is 0.083889 and p-value is 0.4594. The positive correlation coefficient indicates the ROE and LOE are positively correlated. However, the high p-value shows that the H0 of p-value approach should not be rejected and therefor it is concluded that there is no significant linear relationship between LOE and ROE. Thus, H0 for hypothesis 9 is failed to be rejected.

The correlation coefficient of LOD to ROE is -0.215734. The negative correlation coefficient indicates LOD is correlated with ROE inversely. ROE will be reduced every time LOD is increased. Nevertheless, the p-value of 0.0546 indicate that LOD only pass the p-value approach test at 0.10 significant level, which indicates that there is only weak linear relationship between LOD and ROE.

Moreover, the correlation coefficient of LOA to ROE is -0.028273 and p-value is 0.8034. The extremely high p-value indicates that the H0 of p-value test should not be rejected at both significant levels and there is no significant linear relationship between LOA and ROE. Thus, H0 for hypothesis 8 should be rejected.

The correlation coefficient of -0.399285 indicates there is a negative linear correlation between LAD and ROE. It shows that ROE would be reduced by 0.399285 every time LAD is increased by 1. Moreover, the p-value of 0.0002 indicates the H0 of p-value approach should be rejected at both significant levels. Therefore, there is a significant linear relationship between LAD and ROE. The H0 for hypothesis 7 should be rejected as well.

Lastly, the positive correlation coefficient of COE to ROE provides that there is a positive linear correlation between COE and ROE. ROE is increased while COE is increasing. Further, the relatively low p-vale of 0.0012 also indicates the H0 of p-value approach is rejected and there is a significant linear relationship between COE and ROE. The H0 for hypothesis 10 should be rejected.

20.76714

3.092798

0.523176

1.534505

111.1056

39.23629

27.19021

## Probability

0.000031

0.213014

0.769828

0.464287

0.000000

0.000000

0.000001

In order to compare the test statistic with chi-square distribution with 2 degree of freedom, critical values of 5.99 at 0.05 significant level and 4.61 at 0.10 significant level would be used.

Based on the result above, it is failed to reject H0 of Jarque-Bera test at 0.05 significant level for both LAD and LOD, therefore, both explanatory variables have normal distribution. Further, the result also shows that H0 of Jacque-Bera test for LOA at 0.10 significant level should not be rejected as well. It indicates, LOA has normal distribution as well.

Since test statistic of COE, LOE, ROA and ROE is much higher than critical value, therefore the H0 of Jacque-Bera test is rejected and all of them have non-normal distribution. Nonetheless, the result of Jacque-Bera is valid for ‘large samples’, therefore, the result at here can only be considered as ‘approximate’ since the sample size of this study is relatively small compare with other research (Agung, 2009).

## Table 4.4.1: Pooled Ordinary Least Square (OLS) Result of Return on Assets, ROA (Response Variable) and Explanatory Variables.

Coefficient

Standard Error

T-statistic

Probability

Coefficients, C

0.014420

0.011315

1.274418

0.2065

Loans as percentage of deposit , LOD

-0.021910

0.009540

-2.296712

0.0245

Liquid assets to total deposit ,LAD

-0.018009

0.009935

-1.812719

0.0739

Total loans to assets, LOA

0.034809

0.019650

1.771465

0.0806

Total loans to equity, LOE

-0.000662

0.000296

-2.237534

0.0283

Cash on equity, COE

0.000236

0.001097

0.215268

0.8302

R-squared(R2)

0.330787

F-statistic

7.315545

0.285570

Probability(F-statistic)

0.000013

Durbin-Watson statistic

1.374900

According to regression result in table 4.4.1, the model between ROA and explanatory variables is as below

ROA = 0.014420 - 0.021910LOD - 0.018009 LAD + 0.034809 LOA - 0.000662LOE

(0.009540) (0.009935) (0.019650) (0.000296)

+ 0.000236 COE + e

(0.001097)

Further, the coefficient of determination R2 for ROA regression model shows that there is 33.08% of the total variation in ROA is described by the variation in explanation variables.

Moreover, the Durbin Watson result for ROA regression model is 1.3749. Since the value is lower than 2, it is accepted that there is a positive serial correlation in the model where error in a time period is positively correlated with errors in next time period.

Based on the result in table 4.4.1, F-statistic of ROA regression model is 7.315545.Therefore H0 of f-test is rejected and regression models are significant at 0.1%. Therefore, it can be concluded that at least one explanatory variable can be explained variation in ROA.

In order to apply t-test, the critical value must be determined. Since the degree of freedom is 80-5-1=74, therefore the critical value for 0.05 significant level is 1.9925 and 0.10 significant level is 1.6657 for two tailed test. The plus-minus sign of coefficient decides the type of relationship between explanatory variables and explanatory variables. A positive coefficient represents a positive relationship between explanatory and response variable and vice versa.

According to the result in table 4.4.1, the coefficient of LOD is -0.021910 which mean that there is a negative relationship between LOD and ROA. The negative sign is consistent with expected sign in earlier chapter. Further, based on the t-statistic -2.296712, LOD has significant contribution to ROA at significant level of 0.05. Thus, H0 for hypothesis 1 is rejected. The negative relationship between ROA and LOD is consistent with earlier finding of Guru et al. (2002). Guru et al. (2002) suggested the reason that banks might suffer from higher loans cost through high interest interbank borrowing and higher nonperforming loans.

Based on the model above, ratio of loans as percentage of deposit (LAD) has a negative relationship with ROA. The negative relationship between LAD and bank performance indicator is consistent with expected sign. Moreover, the T-statistic and p-value of LAD is -1.812719 and 0.0739 which indicates that there is significant contribution of LAD to ROA at 0.10 significant level. Thus, H0 for hypothesis 2 is rejected. Bank with high liquidity assets holding might not generate high income since most of the liquidity assets generate low return due to riskless characteristic. Nonetheless, the result is contrast with earlier finding of Kosmidou et al. (2005). The reason of inconsistent might be the different in sample size and demographic group.

The coefficients of total loans to assets (LOA) in ROA models is positive. This indicates that LOA bring a positive impact to bank performance. Besides, the T-statistic and p-value of LOA is 1.771465 and 0.0806. Therefore, LOA has significant contirbution to ROA at 0.1 significant level and H0 for hypothesis 3 should be rejected. The positive relationship between LOA and bank performance is in accordance with expected sign and earlier studies of Naceur (2003), Mamatzakis and Remoundos (2003) and Athanasoglou et al. (2006). Olson and Zoubi (2011) suggested the reason that returns from loan is much higher than other assets. The explanation is consistent with market portfolio theory, whereby high risk high return.

The negative coefficient of total loans to equity (LOE) in ROA model shows that there is a negative relation between these 2 variables. Further, based on the T-statistic and p-value in table 4.4.1, there is a significant contribution of LOE to ROA at 0.05 significant level. Thus, H0 in hypothesis 4 is rejected. The negative relationship between LOE and ROA is contrast with expected sign. The high interest interbank borrowing and higher nonperforming loans rate might be the reason that LOE has negative relationship with ROA (Guru et al., 2002).

The coefficients of cash on equity (COE) in ROA regression model show COE has a positive impact to bank performance indicator. The positive relationship is consistent with expected sign. However, the T- statistic and p-value of COE shows that the T-test is failed to apply in both 0.05 and 0.10 significant level.

## Table 4.4.2: Pooled Ordinary Least Square (OLS) Result of Return on Equity, ROE (Response Variable) and Explanatory Variables.

Coefficient

Standard Error

t-Statistic

Probability

Coefficients, C

0.160667

0.161720

0.993490

0.3237

Loans as percentage of deposit , LOD

-0.312468

0.136347

-2.291708

0.0248

Liquid assets to total deposit ,LAD

-0.228808

0.141990

-1.611431

0.1113

Total loans to assets, LOA

0.364285

0.280841

1.297121

0.1986

Total loans to equity, LOE

0.001356

0.004228

0.320636

0.7494

Cash on equity, COE

0.012963

0.015684

0.826515

0.4112

R-squared(R2)

0.289278

F-statistic

6.023896

0.241256

Probability(F-statistic)

0.000099

Durbin-Watson statistic

1.266001

According to regression result above, the model between ROE and explanatory variables is as below

ROE = 0.160667 - 0.312468LOD - 0.228808LAD + 0.364285LOA + 0.001356LOE

(0.136347) (0.141990) (0.280841) (0.004228)

+ 0.012963COE + e

(0.015684)

The coefficient of determination R2 interpret of goodness-of- fit of a regression. The coefficient of determination R2 for ROE model shows that there is 28.92% of the total variation in ROE is described by the variation in explanation variables.

Moreover, according to table 4.4.2, the Durbin Watson result for ROE model is 1.2660. Since the value is lower than 2, it is accepted that there is a positive serial correlation in the model where error in a time period is positively correlated with errors in subsequent period.

The F-statistic of ROE regression model is 6.023896. Based on this result, the H0 of F-test should be rejected and the regression model is significant at 0.1%. Therefore, it can be concluded that at least one explanatory variable can be explained variation in ROE.

Based on the table 4.1.2, the coefficient of LOD is -0.312468. This indicates there is an inverse relationship between LOD and ROE. The inverse relationship is in accordance with expected sign. Further, the T-Statistic and p-value of LOD are -2.291708 and 0.0248. Therefore, the t-test result shows that LOD has significant contribution to ROE at 0.05 significant level. Thus, H0 for hypothesis 6 is rejected. The result of inverse relationship is contrast with earlier finding of Guru et al. (2002) where LOD has positive relationship with ROE. The difference of result might be due to vary sample size and demography background.

The negative coefficient of LOD in ROE model shows that there is a negative relation between these 2 variables. This is consistent with expected sign. However, LOD is failed to apply t-test since the probability is higher than 0.10.

The coefficient of LOA is 0.364285. The positive coefficient indicates that LOA has positive impact to ROE which is consistent to expected sign. Nevertheless, LOA is failed to apply as well due to a p-value which more than 0.05 and 0.10 significant level.

Based on the table 4.1.2, LOE has a positive relationship with ROE due to positive coefficient. The result is in accordance with expected sign. The extremely high p-value indicates that t-test is failed to apply on LOE.

Last but not least, the coefficient of COE shows there is a positive relationship between COE and ROE. The positive relationship is consistent with expected sign. Again, COE is failed to apply t-test due to a high probability value.

For the case of COE in ROA model and LOA, LAD, LOE and COE in ROE regression model, the high p-value indicates that the difference is not significant. However LOA and LAD are still have some confidence since the p-value is still low. The confident level is decreasing when there is an increase in p-value. The high p-value does not give the meaning that relationship between explanatory and response variable. It only indicates that the sample size is not big enough to give a result (Agung, 2009). This would be included as part of limitation and future research in next chapter.

Coefficient

t-statistic

Probability

Coefficient

t-statistic

Probability

-0.021910

-2.296712

0.0245

-0.312468

-2.291708

0.0248

-0.018009

-1.812719

0.0739

-0.228808

-1.611431

0.1113

0.034809

1.771465

0.0806

0.364285

1.297121

0.1986

-0.000662

-2.237534

0.0283

0.001356

0.320636

0.7494

0.000236

0.215268

0.8302

0.012963

0.826515

0.4112

0.330787

0.289278

0.285570

0.241256

7.315545

6.023896

## Prob.

0.000013

0.000099

According to the result of F-test for both ROA and ROE, it is proved that there is at least one explanatory variable can be explained variation in both response variables. Nevertheless, based on the result of t-test for both models, there are 4 explanatory variables of ROA model can apply the t-test and show significant contribution to ROA. However, there is only 1 explanatory variable of ROE model can apply the test.

In order to decide a better model, the adjusted R2 of both models are compared. A R2 is interpreted as goodness of fit of an Ordinary Least Square Regression (OLS). Since ROA has higher adjusted R2 than ROE, hence ROA model is considered better than ROE.

## 5.0 Introduction

In this chapter, the finding and summary of this research would be presented. Further, the implication of this study would be discussed as well. Section 5.1 shows the summary and conclusion for the main finding in this research. Section 5.2 would discuss the recommendations from this study and section 5.3 discusses the limitations of this research and suggestion of research can be further studied in future. Section 5.4 is the conclusion for this study.

## 5.1 Summary of Finding

This study examined the relationship between liquidity and profitability of local owned commercial banks. From the findings it is concluded that liquidity has impact on bank performance. Both liquidity risk and liquidity level can affect the performance of bank.

The liquidity indicators in this study LAD and COE have different relationships with bank performance. The reason is that bank can increase their stability then profitability by increasing liquidity (Wagner, 2007). However, the increases of profitability can only reach to certain level and the profit start to diminish for further liquid asset holding (Bordeleau and Graham, 2010).

Moreover, the liquidity risk indicators LOD, LOA and LOE also provide varies relationships with bank profitability. Indeed, a higher liquidity risk can bring higher interest revenue to banks (Shen et al., 2009 and Olson and Zoubi, 2011). Further, the revenue which generates by liquid assets such as government bond or money market securities is much lower than illiquid assets (Demirgüç-Kunt et al., 2003). However, bank with high lend out rate could receive losses due to non-performing loans and high borrowing cost (Guru et al., 2002).

Last but not least, ROA model is concluded as better bank performance indicator in this study. Based on the adjusted R2, ROA has better goodness-of-fit when Ordinary Least Square Regression (OLS) is applied. This supports earlier studies, for instance Athanasoglou et al. (2006) and Staikouras and Wood (2011) , have chosen ROA as bank performance indicator.

## 5.2 Recommendations

This study recommends that local owned commercial bank should take liquidity seriously in order to maintain stability while generating revenue. Further, banks are suggested to improve their current practice of liquidity management. An effective liquidity management is essential to make sure bank can allocate its asset efficiently to prevent temporary short of funds or bank run during crisis.

Moreover, banks should assess their liquidity thoroughly in order to create best fit liquidity policy. Banks can maximize its revenue generation through bearing certain level of liquidity risk. Further, profitability and stability of bank could be increased through holding certain level of liquid assets. Banks should determine it and formulate liquidity policy based on such determination.

Moreover, Bank Negara should assess liquidity of bank at regularly. This can be achieved through regulation such as submission of liquidity report at regular interval. Last, the strength of security market should be increased in order to develop a better banking sector for competition. Thus, banks would take liquidity seriously and considered it as extra advantage.

## 5.3 Limitations and Future Research

As sampling which has described in previous chapter, there are only 8 local owned commercial banks were selected for this research study. However, from the result of empirical testing, it shows that there are few limitations of this research.

Firstly, the sample size of this study is too small. The small sample size might be the main reason there are some results from empirical test are not significant. The reason that only 8 commercial banks are selected is due to too narrow requirement of sampling which is only listed local owned commercial bank would be selected. There are \$ commercial in Malaysia. However, there are only 8 banks are local owned and listed under Kuala Lumpur Stock Exchange, Bursa Malaysia.

Moreover, the time period of 10 years is considered short to collect enough data. This is due to difficulty in collecting data over 10 years. Normally, banks only present 5 years annual report on their website and 10years on Bursa Malaysia website. The insufficient long time period could lead insufficient data to get accurate result.

As limitation stated above, there are few suggestion for future research. Firstly, a larger sample size shall be used to get more accurate result or minimize variables which are not significant. In order to get a larger sampling size, the requirement of data can be amended to wider coverage, for instance, Islam banks or foreign owned banks could be included. Secondly, it is advised that the time period shall longer than 10 years. A longer time period indeed give a better result in empirical test.

Further, based on the result of this study, it is significant that return on total asset (ROA) is a better bank performance indicator. The suggestion for future research therefore can take this into account when choosing bank performance indicator.

## 5.4 Conclusion

In conclusion, it can be concluded that liquidity is a very important determinant to banks performance. Liquidity of bank not only affects stability of banks but profitability as well. The growths of loans and deposit have been relatively quick in the first decade of the 21th century. Thus, it is essential that bank could manage their liquidity well to avoid for bank run even during crisis.