The impact of liquidity on bank performance – the case of vietnam

List of tables and figures  Tables: Table 1: Overall studies Table 2: Model choosing process Table 3: Description of the variables used in the regression model Table 4: Summary of data

Dissertation submitted in partial fulfillment of the

Requirement for the MSc in Finance

FINANCE DISSERTATION ON THE IMPACT OF LIQUIDITY ON BANK PERFORMANCE – THE CASE

OF VIETNAM

TRINH TRUONG AN

ID No: 17047360 Intake 6

Supervisor: Dr Pham Thu Thuy

September 2023

Acknowledgement

First and foremost, I want to express my sincere thanks to Dr Pham Thu Thuy,

my supervisor, for guiding me since the first day of doing my dissertation Every of her advice, patience as well as the consultation helped me to understand the problem under many aspects and how to become an independent learner in doing research

Next, I want to send my gratitude to my entire lectures and tutors for encouraging, supporting and providing me many valuable knowledge during the course Moreover, I would like to express my appreciation to the Banking Academy, the University of the West of England and ISBA program for providing me a professional environment, giving me a chance to complete my project Besides, I also want to express my gratefulness to Dr Roberto Ercole, my Economic Research Method lecturer, for providing me some basic, essential skills and information before my dissertation started In the process of education, I hardly avoid making mistakes because limited in both time due and experience, therefore, I welcome all comments from teachers to achieve more knowledge for the future

Last but not least, I want to show my appreciation to my family and my friends for motivations as well as criticism These assistances provide the aids I need in deploying this research

Thank you!

in ROA The current non – performing loans of many banks is rising and many banks can easily put up with liquidity problems Besides, discussion of other variables also points out some views of the Viet Nam banking system operation under 11 years Furthermore, this paper also suggests some policies for both commercial banks themselves and the State Bank of Viet Nam to improve the effectiveness of capital as well as the quality of banking system in long term

Contents

Acknowledgement 1

Abstract 2

I Introduction 6

II Literature review 9

1 Theory of liquidity 9

2 Definition of bank performance 10

3 Impact of liquidity on bank performance 10

4 Research gap 15

III Data and Methodology 18

1 Data sources and research methods 18

2 Correlation matrix 18

3 Estimation technique 19

4 Specification of variables 21

5 Regression model 24

IV Result – Interpretation – Discussion 25

1 Data descriptive 25

2 Correlation matrix 26

3 Interpretation of the regression results 27

4 Discussion 32

i The impact of internal factors 32

ii External factors 39

V Policy suggestion 43

VI Limitation and conclusion 45

VII Appendix 47

VIII Reference 56

List of tables and figures

 Tables:

Table 1: Overall studies

Table 2: Model choosing process

Table 3: Description of the variables used in the regression model Table 4: Summary of data

Table 5: Variable correlation matrix

Table 6: Hausman test for ROE model

Table 7: Hausman test for ROA model

Table 8: Regression result of FEM and FGLS

ROE Results

Table 9: POLS Model

Table 10: Breusch – Pagan Test for heteroscedasticity

Table 11: Test for multicollinearity

Table 12: Fixed Effect Model

Table 13: Random Effect Model

Table 14: Test for heteroskdasticity

Table 15: Feasible Generalized Least Squares Model

ROA Results

Table 16: POLS Model

Table 17: Test for heteroskdasticity

Table 18: Test for multicollinearity

Table 19: Fixed Effect Model

Table 20: Random Effect Model

Table 21: Test for heteroskdasticity

Table 22: Feasible Generalized Least Squares Model

 Chart:

Chart 1: Cash asset of the U.S commercial banks

Chart 2: Viet Nam GDP growth rate

List of abbreviation

No Abbreviation Meaning

2 SVB Silicon Valley Bank

3 SBV State bank of Viet Nam

4 ROA Return on assets

5 ROE Return on equity

6 OLS Ordinary Least Squares

8 FEM Fixed Effects Model

10 FGLS Feasible Generalized Least Squares

I Introduction

According to Bordeleau and Graham (2010), ‘liquidity’ was an instrumental factor in the global financial crisis During the crisis, the raise of uncertainty caused funding sources to evaporate, eventually led to the shortage of cash in the commercial bank in order to fulfill their debt obligation They also mentioned that, in some severe cases, some of banks failed and some were forced into mergers As a result, the ‘contagion’ will spread to many other countries Therefore, although the financial crisis has passed for more than a decade, the question of liquidity problem still exists till now Moreover, there was a debate about whether banks had appreciated the consequence of liquidity management or not after the crisis (Marozva,2015)

Recently, the whole world did face up with the Covid-19 pandemic and the bank business model has many changes pre and post – pandemic Vives et al (2020) said that: ‘The Covid-19 crisis most likely means interest rates will remain low for much longer.” This situation will benefit the banks in short run since they can easily access to the liquidity support from central bank as well as create more loans However, the authors also predict that in few years later, non – performing loans will increase and it could threaten the banks’ solvency like the case of global financial crisis in 2008 The prediction was right According to KPMG (2022), the global supply chain disruption was caused by many factors which then lead to the raising of material cost and eventually increase the inflation rate As stated by Mena (2023), the inflation fight started in March 2022 when the Federal Reserve (Fed) announced the first time of interest rate increase and by the end of July 2023, the interest rate stand at 5.12% after

11th increase

Due to the high interest rate, Silicon Valley Bank (SVB), Signature Bank and Silvergate were collapsed (Polychroniou, 2023) In the case of SVB, the bank at that time hold up to 55% of fixed – income securities such as U.S government bond Although the investment was considered low – risk since the bank planned to hold to maturity, their depositors did not think so As the same time, their depositors chose to withdraw their money In order to cover the cash shortfalls or liquidity risk, SVB decided to sell $21 billion of its securities portfolio at a loss of $1.8 billion SVB after that tried to raise more than $2 billion of new capital due to the drain on equity capital This move of SVB sent a bad signal to their customers, who were already losing confidence in the bank The second wave of withdrawal came and led to the collapse

of SVB (Tennekoon, 2023)

In the case of Viet Nam, at the moment, many banks are facing up with high non – performing loans ratio due to the ‘greediness’ and uncontrollable issue of corporate bonds Since they were guaranteed by commercial banks before went out the market, they earned a lot of trust and confidence from the customer However, the high inflation led to the raise of interest rate, those corporations failed to fulfill their debt obligation So, the debts were transfer from group 1 and 2 to group 3, 4 and 5 The banks are now holding a lot of illiquid asset, mostly real estate and the question of liquidity and banks’ performance is raised once again

For this reason, this research is conducted to extend the understanding of the correlation between liquidity and banks’ performance in Viet Nam by two research questions:

1 What is the impact of liquidity on banks' performance in Viet Nam?

2 What are suggestions for the State bank of Viet Nam and commercial banks themselves to improve and control the liquidity problem in order to enhance the overall performance of banking system?

The structure of this paper will follow six main parts After the introduction part, literature review section will give definition and theory of liquidity as well as banks’ performance will be discussed Moreover, other previous researches will be presented so as to select good variables for Viet Nam banking environment Next, the methodology will illustrate the methods, data collection, variables and model of this paper The empirical results will be shown, interpreted, and discussed focus on the correlation between liquidity and banks’ performance Then, some policy suggestions will be stated for a better upcoming situation And the final part will reveal the limitation and summary of the research

II Literature review

1 Theory of liquidity

According to Bibow (2009), the concept of ‘liquidity’ was first developed by John Maynard Keynes in his book The General Theory of Employment, Interest and Money in 1936 The ‘Liquidity Preference Theory’ was explained under the relationship between interest rate and the amount of money one wishes to hold The theory suggests that for a longer period of time, investor should ask for a higher interest rate or premium due to the increase of risk and cash is the highest form of liquid asset Moreover, there are three motives behind the demand for liquid cash by the public: the transaction motive, the precautionary motive, and the speculative motive (Keynes, 1937) In the past, the banking system was still under – developed, therefore, the public will demand for cash to make a transaction easier Nowadays, since the banking system

is well – developed, billions of transactions are made every day Therefore, the public will demand for cash mostly because of precautionary motive, and speculative motive

as the case of SVB is a great example of these two motives The damaged of financial institution reputation will cause a tremendous effect to the liquidity management and

in a severe situation, it can cause a bank run or a crisis For instance, after facing a lot

of problems like high leverage, poor controls and risk management, high real estate concentration, questionable accounting and poor disclosure, and weak government oversight, Lehman Brothers lost its market confidence; apparently, most banks withdrew their services and credit lines to Lehman Eventually, the bank experienced

g intermittent liquidity problems due to their inability to fulfil its short-term obligation (Mawutor, 2014) In order to calculate the liquidity of a firm, current ratio or cash ratio,

etc are usually being used (Rehayem, 2019) However, for commercial bank, the it is very important to consider money deposit in both short-term and long-term

2 Definition of bank performance

In the words of Bagh et al (2016), profitability or financial performance of a firm including commercial bank entails measuring the results of a company's overall strategies and activities in term of money In the world and Vietnam, return on assets (ROA) and return on equity (ROE) are two popular ratios which are used for measuring bank performance According to Gallo (2016), ROA is calculated by dividing net income by total assets and for ROE is dividing net income by total equity While ROE presents the rate of return on the investment of banks’ shareholders, ROA presents the ability of generating income from utilizing bank assets Clearly, when comparing banks with the same size, a bank with higher ROE and ROA shows a better result in business and investment policies According to U.S Securities and Exchange Commission (2012) the profit before tax of Bear Stearns dropped 3.2% in 2007, compared with 2006 and return on average common equity was only 1.8% for fiscal 2007, compared with 19.1% for fiscal 2006 Soon, in early 2008, Bear Stearns was bought by JPMorgan (Reuter, 2008)

3 Impact of liquidity on bank performance

According to the National Credit Union Administration CAMEL rating system

is a tool to evaluate overall financial performance of banks and financial institutions CAMEL stands for: C: Capital A: Assets M: Management E: Profits L: Liquidity Therefore, it is generally accepted that when the liquidity is well managed, the

performance will be expected to be better However, there are some claims that banks have to sacrifice their liquidity to achieve a better performance

Nishanthini and Meerajancy (2015) used Current ratio and Quick ratio to find the correlation with ROA, ROE and Net Profit margin of both State Banks and Private banks in Sri Lanka from 2008 to 2012 The result showed that liquidity and bank performance have a negative relationship They said that if a bank holds a lot of liquid assets to cover liquidity problems, the profit will decrease since those assets are mainly cash and cash equivalent meaning it provides little or no return It was also noted that the time frame they chose was after the global financial crisis; therefore, the action of the banks might be more cautious than pre- crisis

The research of Adeyanju (2011) also came with the same result The paper used both primary and secondary data and concluded that illiquidity and excess liquidity are ‘financial diseases’ to the profitability of the bank in Nigeria Akhater (2018) examined 30 commercial banks in Bangladesh from 2011 to 2016 Along with some profitability variables like ROA, ROE and NP, he used loan ratio and interbank ratio to check the correlation Based on panel data collected, the Fixed Effect Regression model was used and he concluded that after maintaining minimum liquidity, customer’s deposits and borrowings should be used in high-quality loans to increase the earning for their shareholders or increase ROE

The other research in Bangladesh was done by Paul et al., (2020) with a sample

of forty commercial banks from 2009 to 2018 The authors used only ROE for dependent variable while there were five liquidity representations: Loan to Deposit

Ratio (CDR), Liquid Assets Ratio (LAR) and Current Ratio (CR) The results conclude that LDR, DAR and CDR did have correlation with ROE; however, LAR and CR were not significant The authors suggested that the best solution for Bangladesh banks was

to keep its liquidity and profitability equally

Malik et al., (2016) chose twenty-two private sector banks in Pakistan from

2009 to 2013 and Ordinary Least Squares (OLS) was used to measure the relationship between these two variables The results said that there was a relationship between liquidity and ROA; however, for other two dependent variables ROE and return on investment (ROI), the results were insignificant The suggestion focused on restructuring the liquidity management strategies to improve yields on shareholders’ equity and enhance the use of assets

According to Lukorito et al., (2014), the financial sector in Kenya was dominated by commercial banks which in this case is quite similar to Vietnam The paper selected 43 commercial banks from 2009 to 2013 to seek to find out the relationship between liquidity and ROA The conclusion was for every 1% increase of liquidity, the profitability increased by 86.3% Therefore, liquidity had a very strong positive relationship with the profitability in this country The research suggests that for banks to achieve significant profits, they should make substantial investments in assets It's also important to maintain sufficient liquidity, potentially through short-term marketable securities, to ensure profitability Additionally, there's a recommendation

to actively seek out feasible investment prospects and connect these opportunities with customer deposits

Lartey et al., (2013) selected seven out of the nine listed banks on the Ghana Stock Exchange to find out the relationship between liquidity and profitability The project used time series analysis for trend measurement of liquidity and profitability ratio In conclusion, from 2005 to 2010, both the liquidity and the profitability levels

of the listed banks were dropping within the period 2005-2010 There was a very weak positive relationship between these two variables The authors also said that if the bank holds too much liquid assets or low‐return assets for a long time, the opportunity cost would eventually outweigh the benefit of any increase in the bank’s liquidity resiliency

as perceived by funding markets Along the same lines, Charmler et al (2018) studied the impact of liquidity of performance of commercial banks in Ghana using a sample

of 21 banks with 10-years period from 2007 to 2016 By using the ratio of liquidity assets divided by total assets and liquidity assets divided by total loans A weak effect

of the first ratio on ROA was detected while there is no significant result for ROE with the second ratio The ratio of liquidity assets divided by total loans can be considered

as quick ratio for banks It measures the short-term solvency of commercial banks and

it gives the idea for bank's liquidity

The research of Ibrahim (2017) chose 5 commercial banks in Iraq from 2005 to

2013 and OLS model was conducted Along with ROA, the independent variables were Loan Deposit ratio, Deposit asset ratio and Cash deposit ratio All three variables had positive relationships with ROA, therefore the recommendation mainly focused on increasing customer deposit and holding more cash However, the research had many limitations Firstly, there were more than thirty banks in Iraq at that time which means the sample collected was too small Moreover, many banks did not publish their annual

reports, particularly income statements and balance sheets in their site and in the Iraqi Stock Exchange (ISX) For these reasons, the result of this research could be very biased

Pasiouras and Kosmidou (2007) on the other hand chose European Union area measure factors that impacted on the profitability of both domestic and foreign banks The sample consisted of 584 commercial banks operating in the 15 EU countries over the period 1995–2001 The liquidity factor had a mixed result For domestic banks, the increase of liquid assets held by the bank would create a downfall to profitability while for foreign banks the result was opposite Moreover, two macro variables: Inflation and Gross Domestic Product (GDP) growth were also measured Both inflation and GDP growth had the same result with the liquidity variable which were positive association for domestic banks and negative association for foreign banks The authors indicated that domestic banks could have many opportunities since they can easily adjust the interest rates accordingly and consequently to earn higher profits

The case of Turkey was quite different Alper and Anbar (2011) chose ten commercial banks in Turkey over the period 2002 - 2010 with both internal factors and external factors affecting bank performance The overall result concluded that liquidity variable, inflation and GDP growth rate had no relationship to bank performance The real interest rate on the other hand had a positive relationship with ROA and ROE and the banks should focus on increasing bank size and non - interest income to improve profit

4 Research gap

Overall, there is a relationship between liquidity and bank performance both in positive and negative signs In previous researches, panel data is mainly used and many methods like OLS, Fixed Effect Regression or Random Effect Regression are also utilized, and they achieved different result But different countries have different characteristics and the majority of previous researches were finding out the correlation between liquidity and bank performance A lot of studies said that the relationship is negative; however, none of them ignore the important of liquidity management Most authors agreed that the liquidity should be remain at a minimum level and the rest of the money should be invested in some high liquid asset like treasury bond or securities

to maximize profit

So, this paper will analyze the situation of Viet Nam focus on liquidity and bank performance with the aim liquidity will have a positive sign which is rising of liquidity will enhance the bank performance In addition, the research gap of this paper is the difference in time frame and country choosing; the variables will be selected and made some adjustments to fit with Viet Nam environment Recently, the whole world including Viet Nam has witnessed a lot of bad news such as the trade war between the U.S and China in 2018 – 2019 (Gorman, 2022), the appearance of Covid – 19 at the end of 2019 led to the statement of lockdown in many countries or supply chain disruption (Tan et al., 2022) and the war between Russia and Ukraine in 2022 (Bigg, 2023) It is noticeable that these events happened among the biggest nations in the world; therefore, many policies or financial key factors like inflation and interest rate

of small countries like Viet Nam have been changing to adapt with the unstable

situation Hence, the research gap of this paper will focus on the unstable period recently The time frame is 11 years from 2012 to 2022 because the data of some banks

is not available before 2012 and it is not presented quarterly also

Nishanthini and

Meerajancy (2015)

Sri Lanka (2008 – 2012)

Describe statistics and inferential statistics

(-)

Adeyanju (2011) Nigeria

(Collection of both primary and secondary data)

Pearson correlation data analysis

(-)

Akhater (2018) Bangladesh

(2011 – 2016)

Fixed Effect Regression Paul et al., (2020) Bangladesh

Ordinary Least Squares (OLS) Lukorito et al.,

(2014)

Kenya (2009 – 2013)

Descriptive and inferential statistics

(+)

Lartey et al.,

(2013)

Ghana (2005 – 2010)

(+)

Charmler et al

(2018)

Ghana (2007 – 2016)

Panel regression analysis

(+)

Ibrahim (2017) Iraq

(2005 – 2013)

Ordinary Least Squares (OLS)

(+)

Pasiouras and

Kosmidou (2007)

15 EU countries (1995 – 2001)

Random Effect Regression

Both (+)/(-)

Alper and Anbar

(2011)

Turkey (2002 – 2010)

Fixed Effect Regression

No relationship

Source: Author compilation

Table 1: Overall of studies

III Data and Methodology

1 Data sources and research methods

The research will use the combination of both cross-sectional and time series data (panel data) from 2012 to 2022 For internal bank’s variables like ROA, ROE and some independent variables, the data is taken from Vietstock will then be double checked with the annual audited consolidated financial report of each bank before calculated The list of 30 banks was extracted from State Bank of Viet Nam (SBV) However, due to the incomplete data over several years, the quarterly data cannot be found and the data is collected annually The list of selected banks will be left at the Appendix Next, the independent variables ‘Inflation’ and ‘GDP’ are collected from the World Bank Data The dataset will then be encoded in STATA to choose which is the best model to measure the correlation of liquidity and bank performance

In summary, the research sample comprise 30 commercial banks with 330 observations from 2012 to 2022 Consequently, the panel data used in this paper is strongly balance

2 Correlation matrix

The correlation between dependent variable and independent variables could

be found using Stata The result will be shows in next part of this research The correlation matrix gives the initial look at the relationship of all variables which anticipates the probability of multicollinearity to occur This kind of problem may lead

to a high standard error and thus, bias the sign and the accuracy of the estimated results (Baum, 2006) The coefficients range between -1 and 1 and the correlation between

variable should lie below 0.8 to encounter a significant multicollinearity problem The variables that violating this problem will be eliminated to get better model

3 Estimation technique

There are three model that can use to deal with a balanced panel data: Pooled OLS (POLS), Fixed Effects Model (FEM) and Random Effects Model (REM) Each model has its own pros and cons

According to Stock and Watson (2019), the Pooled OLS model is simply OLS technique where coefficients remain constant across both time and space The major problem of POLS is that it cannot distinguish between various cross sectional units; therefore, it can be blamed for camouflaging or manipulating the uniqueness existing within each cross sectional unit Moreover, it can easily be violated with autocorrelation error So, both FEM and REM are suggested to overcome these problems

Fixed Effect Model assumes that each individual has its own time – invariant characteristic that affects the dependent variable These individual specific characteristics can present unobserved factors that remain constant over time In other word, it can control for unobserved heterogeneity which is the drawback of POLS However, it also exists with disadvantage FEM is not suitable for measuring the effect of a variable that does not change over time

Random Effect Model (REM) on the other hand estimates the effect of individual specific characteristics that are inherently unmeasurable It assumes that the existence effects of independent variables on the dependent variables across individuals are random differences It can address the drawback of POLS but it

sometime can be bias with unobserved independent variable that have relationship with both dependent and independent variables

In this paper, all three model are employed Wald test or F – test will be used to choose between POLS and FEM, Breusch-Pagan Lagrangian Multiplier test for OLS or REM and Hausman test for FEM and REM Moreover, Variance Inflation Factor (VIF) will be used to check multicollinearity among variables After the final model is chosen, Feasible Generalized Least Squares (FGLS) method will be used to fix the error like heteroskdasticity

FE vs POLS

H0 = μ1= μ2 = … = μ

Wald test

RE vs POLS H0 = Var(μi) = 0 Breusch-Pagan Test

ROA Return on assets Net income / Total

assets, 100%

Nishanthini and Meerajancy (2015), Adeyanju (2011), Akhater (2018), Paul

et al., (2020), Malik et al., (2016), Lukorito et al., (2014), Lartey et al., (2013), Charmler

et al (2018), Ibrahim (2017), Pasiouras and Kosmidou (2007), Alper and Anbar (2011)

ROE Return on equity Net income / Total

equity, 100%

Independent variable LIQUID1

(+)

The ratio of cash and cash equivalent to total assets

Cash and cash equivalent/Total assets, 100%

Charmler et al (2018), Paul et al., (2020), Alper and Anbar

LIQUID2

(+)

Customer deposits to total assets

Customer deposits/Total assets,

100%

(2011), Alper and Anbar (2011)

NPL

(-)

Non – performing loans

(Current debt + Special mention debt)/Loans to customer, 100%

Pasiouras and Kosmidou (2007), Paul et al., (2020)

DBFC

(+)

Deposits and borrowings from other credit institutions

Deposits and borrowings from other credit institutions/Total asset, 100%

Akhater (2018)

LTA

(+/-)

Loan to assets ratio (+/-)

Total debt/Total assets

Akhater (2018)

INF

(+/-)

Inflation (+/-)

Inflation, 100% Pasiouras and

Kosmidou (2007), Alper and Anbar (2011)

GDP

(+)

Gross domestic product growth rate

GDP, 100% Pasiouras and

Kosmidou (2007), Alper and Anbar (2011)

Table 3: Description of the variables used in the regression model

In previous researches, liquidity variable usually grouped as liquid asset but in this paper it will contain only cash and cash equivalent According to National Institute for Finance of Viet Nam, 2020 was the final year of proposing cashless payment in Viet Nam Although there were up to 30 million people used cashless payment and the growth of mobile banking increased to more than 200%, Viet Nam along with Thailand and Japan still were three countries that had the highest cash payment in Asia at the end of 2022 (Bao, 2023) This situation proves that cash is still standing at the highest ranking of liquidity in the head of every citizen in both developing and developed nations

Moreover, the ‘SEC’ variable use the trading securities of the banks instead of securities for investment Viet Nam bank usually use the excess cash for trading short term securities in order to gain more profit Since this is short term trading, it can be easily withdrawn to cover the liquidity problems

Instead of taking total deposit, the variables in this paper are broken down to customer deposit and deposit and borrowings from other credit institution It is easily

to understand that if a bank meets some pessimistic news from both internal and external, it may send a bad signal to its customer at the same time which may lead to the situation like the case of SVB On the other hand, deposit and borrowings from

other credit institution is a new variable since this is the deposit of firms into the bank and in some severe cases, it can be withdrawn be these firm Since the amount of this money is usually very huge, it can directly contribute to the liquidity of the banks

+ 𝜷𝟕𝑰𝑵𝑭 + 𝜷𝟖𝑮𝑫𝑷 + 𝜺

In particularly,

 β0: the intercept of the regression model

 β1, β2, β3, β4, β5, β6, β7, β8 are coefficients of explanatory variables

 ε: the error term apprehending the influences of omitted variables and other factors that are not included in the model

IV Result – Interpretation – Discussion

1 Data descriptive

Table 4: Summary of data

The summary statistics of all variables is shown in table 4 Overall, the ROE of thirty banks in Viet Nam has an average value equal to more than 9%, much higher than the average value of ROA which is more than 0.7% The coefficient variable SEC

is very small and there are some reasons for it The observation is just 204 while the mean value is just around 0.009% according to the Ministry of Finance (2006), since the financial system in Viet Nam is not well developed and the security market was just founded in 2000, many banks do not choose the security market to store their excess cash as it is still too risky for them Moreover, in the list of thirty banks, there are many young banks with less than five years’ old as of 2012 Viet Nam commercial banks usually focus on selling overall banking services instead of choosing one particular major like technology of SVB Therefore, young banks in Viet Nam tend to tradeoff their profit to gain market share and to maintain good performance This also

explains why the min value of non – performing loan is zero In order to become

prestigious banks, in the early years, they were very careful when lend out money for

both individuals and corporations

2 Correlation matrix

Table 5: Variable correlation matrix

According to Gujarati and Porter (2009), a correlation matrix is a table

presenting correlation coefficients among variables In order to eliminate

multicollinearity problem, the correlation between two variables needs to be under 0.8

and in table 5 all variables are accepted It can be seen that ROE has negative

correlation with almost other elements except LTA while ROA has positive correlation

with DBFC and SEC and negative correlation with the rest independent variables

Since ROA is not used to explain ROE, it will not be considered invalid

3 Interpretation of the regression results

Among all models, the Hausman test was used to chosen between FEM and

REM The tables below show the result of Hausman test for both ROA and ROE model

which investigate the impact of the ratio liquid assets on ROE and ROA

Table 6: Hausman test for ROE model

Table 7: Hausman test for ROA model

The P value of both model smaller than 0.05, hence, the null hypothesis is

eliminated Thus, Fixed Effect Model is the best model for the explanation of this

dataset in both cases However, in both case, the models meet heteroskedasticity

problem Therefore, Feasible Generalized Least Squares is applied to fix the problems

FEM – ROE FGLS – ROE FEM – ROA FGLS – ROA LIQUID1 309.3896*

(0.016)

443.6495***

(0.000)

21.35619 (0.057)

25.11974*** (0.000) LIQUID2 -0.3858099***

-48.26184*** (0.000) NPL -1.292945*

(0.010)

-0.974389***

(0.000)

-0.0768421 (0.077)

-0.0633432*** (0.000)

-0.0565272*** (0.000)

LTA 0.4502286*

(0.036)

-0.1664424 (0.236)

-0.0789339***

(0.000)

-0.1507157*** (0.000)

It is clear to see that in FGLS (ROE) model, ‘LTA’ is the only variable that is insignificant with the P – value higher than 0.05 while the rest of variables are significant at 0.1 percent levels For FGLS (ROA) model, all variables are significant

at 0.1 percent levels According to Baum (2006), R-squared or call the goodness-of-fit

is the percentage of the response variable variation and the higher the value of squared is, the better the model is The R – squared of FEM (ROE) is 0.1353 which means only 13.53% model explained by LIQUID1, LIQUID 2, SEC, NPL, DBFC, LTA, INF and GDP The R – squared of FEM (ROA) is better when nearly 50% model explained by independent variables

R-As a result, the new estimated model is specified as follow:

𝑅𝑂𝐸 = 0.7177081 + 443.6495LIQUID1 − 0.553124LIQUID2 − 662.0316SEC

− 0.974389NPL − 0.8117134DBFC − 0.1664424LTA+ 1.261251INF − 0.4701459GDP + 𝜀

𝑅𝑂𝐴 = 0.1820931 + 25.11974LIQUID1 − 0.0415577LIQUID2 − 48.261846SEC

− 0.0633432NPL − 0.0565272DBFC − 0.1507157LTA+ 0.0987816INF − 0.0322546GDP + 𝜀

It is noticeable that the independent variables show the same sign in both ROE and ROA models The coefficient of cash and cash equivalent is positive which means for every 1% increase of LIQUID1, ROE will increase more than 443% while ROA will increase 25.12%, ceteris paribus The LIQUID2 has a negative sign with bank performance; therefore, if all other things being equal, 1% increase of customer deposit

to total assets will decrease the result of ROE by 0.55% and 0.042% for ROA

Securities to total assets has the same sign with LIQUID2 For every 1% increase of securities to total assets will make ROE decrease 662% and 48.26% for ROA Moreover, for every 1% increase of non – performing loans, ROE will decrease 0.97% and ROA will decrease 0.063%, ceteris paribus Deposits and borrowings from other credit institutions has the same characteristic with customer deposit; therefore, this variable also has negative sign For every 1% increase of DBFC, ROE of banks will lose 0.811% and ROA will lose 0.0057%, ceteris paribus LTA variable on the other hand only have significant with ROA and has negative sign So, if all other things being equal, for every 1-unit increase of total debt to total assets ratio, ROA will decrease 0.15 unit Normally, the increase of inflation and decrease of GDP may cause a bad performance result for every institutions including banks The result of this dataset shows differently For every 1% increase of inflation, ROE will increase 1.26% and 0.098% for ROA In contrast, for every 1% increase of GP, the banks lose 0.47% in ROE and 0.032% in ROA

Overall, the FGLS result points out that liquidity has positive relationship with bank performance in Viet Nam Although the result of ROE model may sound

‘unreasonable’ mostly in cash and cash equivalent ratio and securities to total asset ratio, it still shows that there is a correlation between liquidity and bank performance For ROE model, the lacking of data from securities trading might be a problem lead to the ‘unreasonable’ coefficient Moreover, the variables used to find out the relationship with ROE may unsuitable in the case of Viet Nam which explained why the R – squared

is only more than 13% Therefore, the result of ROA will be more reliable in this paper Comparing the result of this project to other previous researches in other nations, it has

a similar result with Paul et al., (2020), Lukorito et al., (2014), Lartey et al., (2013), Charmler et al (2018), Ibrahim (2017), Pasiouras and Kosmidou (2007) And it is opposite to Nishanthini and Meerajancy (2015) and Adeyanju (2011)

4 Discussion

i The impact of internal factors

It is obvious that cash is one of the most important factor that affect bank profitability and in this paper, it has a positive relationship However, the huge amount

of unutilized cash might be a big problem In a normal condition where inflation is controlled, the central bank like Fed or SBV usually maintain a low interest rate to stimulate lending and consuming to push up economic growth

Chart 1: Cash asset of the U.S commercial banks

For example, before the global financial crisis, the low interest rate did seduce banks to make loans as much as possible According to Turner (2023), up until early

of August 2007, banks were still able to call for money from interbank wholesale markets to fund for their ever-expanding mortgage businesses Instead of raising

deposit and lending them, many US banks and UK banks like Northern Rock chose to originate and distribute mortgages In other word, this process included selling the cash flows coming from mortgage repayments After securitization, these cash flow were known as mortgage-backed securities (MBS) or collateralized debt obligations (CDO) Before Northern Rock collapsed, only 25% of mortgage held by this bank came from traditional deposits (House of Commons Treasury Committee, 2008) The Financial Services Authority (FSA) (2008) also concluded that although the balance sheet was solvency, it did not maintain enough cash resources to meet its payment obligation Moreover, in the history of the U.S., 2008 also saw a record low of cash asset This is the prove that indicated banks not only could utilize the excess cash but they did it way too much Instead of holding the basic form of cash, they invented many ‘instruments’ like MBS and CDO and considered these ‘instruments’ as liquid assets As stated by MacGregor (2020) and confirmed by SBV (2009), the situation of Viet Nam in 2008 was quite similar but was not that severe Although the house price dropped around 40% and many banks met liquidity risk, there was no bank collapsed It is understandable that cash cannot create profit itself; however, the banks should not use

it freely in all circumstances After the banks meet their reserve requirement, the excess cash can be used in other high and stable liquid assets as treasury bonds More importantly, every bank need to develop its own risk management system to predict, act and improvise under any circumstances of the market For instance, in chart 1, it is obvious that after the appearance of Covid – 19, the demand for cash increased sharply The banks at this point need to prepare many scenarios whether there will be a vaccine