Internal Liquidity Determinants Analysis of Commercial Banking
Industry of Jordan
MOHAMMED IBRAHIM SULTAN OBEIDAT
Accounting Department, Business School
Jadara University
JORDAN
NADEEN “MOHAMMED ADNAN” “MOHAMMED YASIN” DARKAL
Department of Business Administration, Business School
Jasdara University
JORDAN
Abstract: - The study aims to determine the possible internal factors affecting the liquidity position of the
commercial banking industry of Jordan. Several possible determinants are taken into account in the study
including, profitability, credit growth rate, customer deposits, financial leverage, capital adequacy, and bank size.
The secondary data covering the period 2008-2019, of 13 out of 15 listed banks at Amman Stock Exchange, is
gathered and analyzed. In total, 1,092 observations are employed in the analysis to achieve the goals of the study.
All hypotheses are tested under 95 level of confidence, which means 5 percent coefficient of significance.
Descriptive statistics including the mean, standard deviation, and the minimum and maximum values, in addition to
correlations, are employed in the analysis of data. Using correlation and regression methods in hypotheses testing,
the study declares that profitability, capital adequacy, and bank size, each of which, has a significant positive
impact on bank liquidity. In addition, the study finds that financial leverage and customer deposits have a negative
significant effect on bank liquidity. Moreover, the study finds no significant impact of credit growth rate on bank
liquidity.
Key Words: - Commercial Banks, Liquidity Determinants, Credit Growth Rate, Customer Deposits, Capital
Adequacy, and Financial Leverage.
Received: March 13, 2021. Revised: January 20, 2022. Accepted: February 12, 2022. Published: March 16, 2022.
1 Introduction
Liquidity is important as profitability is, because
each of which may lead to collapse and bankruptcy
of business organization. The global financial crisis
of 2008, shed light for analyzing liquidity and its
different determinants. Studying liquidity
determinants can be accomplished using different
methods. Using financial ratios in measuring the
level of liquidity is preferable, and leads to good
conclusions regarding liquidity decision. Despite
that, some interested people believe that liquidity
creation method in measuring liquidity may lead to
better findings, than ratios method [9], but still the
ratios method leads to reasonable conclusions, and it
is followed in the current study for investigating the
liquidity of Jordanian commercial banks.
Within the global crises of 2008, most banks of
different countries showed a decline in liquidity,
especially in USA and other western countries, and
the commercial banking system faced a difficulty in
liquidity creation. Some experts and authors
mentioned that the difficult of liquidity creation faced
by commercial banks of some western countries, is
mainly due to low efficiency of markets, and
financial problems of counterparties [9]. Some found
that the liquidity of commercial banks declined
during the crises by 8 percent [12].
The commercial banking industry is the most
important economic sector affecting the efficiency
and effectiveness of different production sectors.
Commercial banks provide funds to other business
organizations, and can finance the different
investment prospective. The liquidity level of
commercial banks is important, because when high
level of liquidity is available to commercial banks,
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
382
Volume 18, 2022
banks will have more ability to finance other
business organizations of other sectors, and can
satisfy the funding needs of business organizations.
In opposite when a low liquidity level is available to
commercial banks, other business organizations will
face difficulties to finance its investments.
The liquidity of commercial banking is a crucial
issue for the economic growth of different countries
all over the world. Commercial banking liquidity
offers loans and help financing other firms of
different industries. In addition, many new small
businesses emerged when it offered loans and
financing by commercial banks. Commercial banks
can finance new and current business organizations,
only when enough liquidity is available, but
whenever there is a lack in liquidity of commercial
banks, new businesses emerging will be less, and the
investment process of the current firms will be below
the average of normal situations. The emergence of
new businesses and the growing investments create
job opportunities for local and for foreigners of rare
qualifications internally. In Jordan, there is a severe
problem of unemployment, less investments, low
attraction of foreign investments, high inflation, and
low rate of economic growth. Therefore, commercial
banks of Jordan can play a vital role in declining the
negative effect of all of these bad economic
indicators. Again, this can occur only when high
level of liquidity is available to commercial banks.
Therefore, the problem of the study is the possible
internal determinants of liquidity level of Jordanian
banks, and this problem can be better introduced
through the following question. What are the internal
determinants of liquidity of the Jordanian
commercial banks?
The current study is important, where its
importance is due to the importance of banking
industry, and its required funding role to other
productive aspects in different countries. The
findings of the study are important to commercial
bank managements, because commercial banks can
make a balance between its liquidity and
profitability, through deposit and credit policies.
Following some policies, which keep banks liquid,
and less risky, and avoiding some policies will
improve the financial performance of commercial
banks. Despite that, some aspects of its importance
are indirectly mentioned above, but it deserves to be
mentioned again following using a straightforward
method. The importance of the study is stemmed
from the importance of funds to investments.
Whenever, funding is available, more investments
can be initiated, and more production institution can
be created, because funding plays the most important
role in activating investments. More investment leads
to more economic growth, less unemployment, better
economic and living well-being. High liquidity level
of commercial banks means that more funding
support, where commercial banks, in this case, can
provide for productive entities inside the country, but
low levels of liquidity leads to less investments,
higher unemployment and less economic and living
well-being for the different groups of people inside
the country.
The objectives of the study can be summarized as
follows. The first objective is to determine the level
of liquidity of commercial banking industry of
Jordan. The second objective is to determine the
possible internal indicators of liquidity of Jordanian
commercial banks. It takes into consideration some
internal factors, where the managements of banks can
exercise a large degree of control over these assumed
determinants. The last objective of the study is to add
more to the available literature with regard to the
factors affecting the liquidity position of Jordan.
The remaining sections are structured as follows.
Section 2 includes the related literature, and shows
some of previous studies that carried out in the field
of liquidity and the determinants of liquidity. The
hypotheses of the study are presented in section 3,
whereas section 4 shows the methods followed in the
study. Section 5 shows the results and analysis, and
section 6 reveals the conclusions of the study.
2 Review of the Literature
The key role of banks and other depository entities is
the transformation of short-term deposits into long-
term loans [11]. The financial obligations that banks
may face require sufficient level of liquidity to meet
these obligations when they do, with no losses. From
this idea the term of liquidity risk comes, which
refers to failure of banks to meet these obligations
without cost.
The financial crises of 2008 began during the
first half of 2007, as credit crisis and later
transformed to liquidity crisis [6]. The decline in
housing prices in of US caused an increase in
mortgage lending that led to a liquidity crisis in 2007.
The financial crisis led to bankruptcies, quasi-
bankruptcies, in addition to a decline of financial
performance of large banks and other financial
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
383
Volume 18, 2022
institutions. The crises later led to a deterioration of
international stock markets, liquidity shortage of
interbank markets, and later extended in 2010, to a
sovereign debt crisis in some European countries,
such as Greece, Spain, Portugal, Italy, and. Before
the financial crisis began in the first half of 2007,
new regulations for banks were in practice in the
form of Basel II, but during the 2008 financial crisis,
only banks that have enough levels of liquidity could
resist the shortage in liquidity problems and continue
meet its obligations. This causes Basel Committee on
Banking Supervision (BCBS) to issue new banking
regulations called 12 Basel III, where these new
regulations give more attention to the management of
capital, equity, and liquidity, and was introduced as a
regulatory framework for banks, all over the world
[7].
Basel Committee in Banking Supervision [1],
defines liquidity as “the ability of a bank to fund
increases in assets and meet obligations as they come
due, without incurring unacceptable losses” [6].
Liquidity of banks is very important issue, where this
importance comes from its role of transforming
short-term deposits into long-term loans.
Banks face many problems in its operations that
make a threat to its solvency. Examples of risks that
banks may face include interest rate, market rate, off-
balance sheet, foreign exchange, and other risks.
Nevertheless, banks may continue subject to
solvency risk, when a bank is unable to generate
liquidity to pay its deposits [14]. Therefore,
commercial banks’ managements are required to give
more attention to liquidity, in order to be able to
avoid losses that appear regarding payments of
deposits when due. In the context of liquidity risk,
Saunders et al (2006) distinguish between two
sources of liquidity risk, the asset side source, and the
liability side source. The assets side of the balance
sheet may be an actual source of liquidity risk when a
bank exercises what is called, off-balance sheet
obligations. This occurs when a bank agrees for a
contract to grant a loan, where based on this this type
of contracts, customers receive an agreed amount,
and the bank is required to offer the borrowed
amount immediately on demand, where this type of
contracts needs more liquidity. In addition, the
liability side of the balance sheet may be another
source of liquidity risk. This occurs when depositors
withdraw a large amount of their deposits, where this
causes a decline in liquidity. Because of that, banks
can avoid such these difficulties and avoid incurring
additional cost, when banks keep enough levels of
liquidity. When a bank faces such liquidity problems,
whether it is an asset or a liability side cause, banks
may use its reserves if sufficient, because rarely
banks keep high amounts of liquid assets because
these liquid assets generate no interests, or very low
rate of interests. A bank facing a liquidity risk may
borrow additional funds, or liquidate some of its
current illiquid assets.
3 Prior Researches
The issue of bank liquidity is given enough attention
since the appearance of 2008 financial crises, but
before that time, bank liquidity had not been given
the required attention in researches and studies. Still,
the issue of commercial banks liquidity is below the
required attention of authors and practitioners in the
developing and Arab countries.
Laštůvková [9], carried out a study aiming to
identify the factors influencing the liquidity of
Czech, Slovak and Slovenian’s commercial banks.
The objective of the study is to determine the impact
of some possible internal and external factors on
commercial bank’s liquidity. Secondary data
covering the period 2001-2013, of a sample of
Czech, Slovak, and Slovenian banks, had collected
and analyzed. The multiple linear regression had
employed in testing the hypotheses. The results
showed that certain factors have a multiple effect on
the different forms of liquidity, while other factors
only affect specific forms of liquidity. The study also
found that small banks are more sensitive to specific
forms of liquidity, while the opposite, is the correct
for large banks. In addition, the study reveals that the
more flexible regulations, lead to more optimization.
Bansal and Bansal [2], investigated the
determinants of liquidity for determining which
among these affect liquidity of some Indian firms.
The relevant secondary data, covering the period
1999-2008, of a sample consisted of 100 textile and
chemical Indian enterprises. The stepwise regression
had used in the analysis of data, and the results
showed that cash flow, debt ratio, and free cash
flows, are significant determinants of liquidity.
The purpose of Ben Moussa’s Study [3], was to
identify the factors affecting the liquidity of Tunisian
banks. The secondary data covering the period 2000-
2010, of 18 Tunisian banks, had been collected and
analyzed. Two measures of liquidity had used in the
study including liquid assets to total assets, and total
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
384
Volume 18, 2022
loans to total deposits. Using both of the static and
dynamic panel methods, the study demonstrated that
financial performance, capital to total assets, opening
costs to total assets, GDP rate of growth, inflation
rate, and delayed liquidity, each of which has a
significant impact on liquidity. Moreover, the study
finds that the liquidity of Tunisian banks is not
affected by each of bank size, loans to assets ratio,
deposits to total asset, financial costs to total credits
ratio, and bank size.
Lotto and Mwemezi [10], carried out a study
to determine the most important determinants of
bank liquidity. Secondary data covering the period
2006-2013 of a sample consisted of 49 banks of
Tanzania had gathered and analyzed. Using the panel
of regression, the study showed that capital
adequacy, bank size, and interest rate margin, had a
negative significant effect on bank liquidity. The
study also demonstrated that each of the rate of
inflation and the non-performing has a positive
significant effect on the liquidity of commercial
banks. Moreover, the study findings revealed that
each of GDP, and GDP growth rate does not affect
the liquidity of Tanzanian banks.
Bista and Basnet [4], analyzed bank liquidity of
Nepal, as an attempt to determine the factors
affecting liquidity. Secondary data covering the
period 2002- 2018, due to 12 commercial banks in
Nepal, had gathered and analyzed. The multiple
regression method is used in hypotheses testing and
other analysis of data. The results revealed that
similarly, deposits, capital adequacy, remittances,
and bank size are significant determinants of bank
liquidity, and where deposits increase liquidity, the
capital adequacy leads to liquidity reduction of
commercial banks. The study also revealed that
internal factors affect bank liquidity more than
macroeconomic external factors. In addition, it
revealed that at the long run, capital adequacy, bank
size, and capital expenditures lead to an increase in
bank liquidity, whereas deposits decrease liquidity of
banks.
Nguyen and Vo [13], carried out a study related
to bank liquidity and objected for examining the
determinants of liquidity of 17 listed banks at the
Vietnamese Stock Exchange, HOSE, HNX and
UPCOM. The study used quarterly financial
information covering the period 2006-2020. The
required macroeconomic data regarding GDP and the
rate of inflation had collected from International
Monetary Fund and the General Statistics Office of
Vietnamese. The panel data method is used in the
study, and it showed that total assets size, return on
total assets, and credit growth are positively
associated with liquidity of banks, whereas the
interaction between bank size and return on total
assets negatively affect the liquidity of listed banks at
HOSE, HNX, UPCoM. The study recommended the
managements of commercial banks to focus effective
credit growth, developing the scale of total assets,
and ensure high rate of return on total assets, to
improve the position of liquidity.
Mohamad [11], prepared a study to identify the
determinants of liquidity of Turkish commercial
banks. In more details, the goal of the study was to
identify the factors affecting the liquidity of
conventional banks of Turkey. Secondary data
covering the period 2005-2013, of 21 listed Turkish
banks, had collected and used in the analysis. Both of
the liquid to total assets, and liquid assets to customer
deposits and funding, were used as measures of
liquidity in the study. The ordinary least square
method was used in testing the hypotheses of the
study. The results showed that only bank
capitalization has a positive significant impact on
both measures of liquidity, but the loan loss reserves
ratio positively affects liquidity only when it is
measured based on the liquid assets to customer
deposits and funding ratio, while bank size has a
negative significant impact on the first ratio. The
findings of the study also showed that profitability
has a negative significant impact on liquidity only
when bank liquidity is measured based on liquid
assets to total assets ratio.
4 Study Hypotheses
As a result of the review made to the related
literature and the survey that made to the prior
researches findings, several hypotheses are
developed. Using the null form of hypotheses, these
hypotheses are listed bellows as follows.
Ho1. The profitability of the commercial banks of
Jordan does not affect the liquidity position of these
banks.
Ho2. The rate of credit growth of commercial banks
of Jordan does not influence the liquidity of these
banks.
Ho3. The listed commercial banks of Jordan client
deposits does not significantly affect the liquidity
position of this group of banks.
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
385
Volume 18, 2022
Ho4. The Jordanian commercial banks’ financial
leverage does not significantly affect the liquidity
position of these banks.
Ho5. Commercial bank’s capital adequacy has no
significant influence on the liquidity of commercial
banks of Jordan.
Ho6. The liquidity position of the commercial banks
of Jordan is not significantly affected by the bank
size.
Ho7. There is no grouping effect of bank
profitability, credit growth rate, client deposits, debt,
capital adequacy, and bank size, on the liquidity
position of commercial banks of Jordan.
5 Study Methodology
The population of the study includes the different
non-Islamic banking enterprises of Jordan. In total,
and by the end of 2020, there were 15 commercial
banks listed at ASE, of these, 2 are Islamic, and the
rest are 13 non-Islamic banks. The data due to each
of these 13 non-Islamic banks had gathered and used
in the analysis of data testing of hypotheses. The two
working Islamic banks in Jordan were eliminated
because the data of this type of banks is inconsistent
with the data of non-Islamic commercial banks. The
collected data covers a 12-year period (2008-2019),
and it had collected using the annual reports of ASE.
The dependent variable of the study is bank
Liquidity, whereas 6 variables are independent and
thereafter examined with regard to their effect on
liquidity. The independent variables of the study
include profitability, capital adequacy, credit growth
rate, client deposit, financial leverage, and bank size.
The liquidity of banks means the ability of
commercial banks to meet the obligation when they
due. Table (1) shows how each of the dependent and
independent variable, is measured. The table shows
that commercial bank liquidity is the relationship of
liquid assets to total assets (Mohamad, 2016).
Profitability is measured using ROA as a good
indicator for profitability, where ROA is computed
by just dividing net income by total assets. Credit
growth is computed using the equation available in
table 1, as follows (Dao and Nguyen, 2020).
𝐶𝑅𝐺
=𝐿𝑜𝑎𝑛𝑠 𝑜𝑓 𝑡ℎ𝑒 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑦𝑒𝑎𝑟 𝐿𝑜𝑎𝑛𝑠 𝑜𝑓 𝑡ℎ𝑒 𝑝𝑟𝑖𝑜𝑟 𝑦𝑒𝑎𝑟
𝐿𝑜𝑎𝑛𝑠 𝑜𝑓 𝑡ℎ𝑒 𝑝𝑟𝑖𝑜𝑟 𝑦𝑒𝑎𝑟
Based on this equation, the credit growth rate can
be found by deducting the prior year loans from the
current year loans, and the result of this subtraction
process, is divided by the prior year loans. The
financial leverage ratio gives an idea regarding the
borrowed amounts of money, and the credit received
by a bank, and it is computed through debt to equity
ratio, which is simply the relationship of liabilities to
equity. The calculation of capital adequacy ratio
differs from Basel 1 to Basel 2, issued in 2004. In
general, the computation of capital adequacy in this
study is based on Tier 1 capital, so capital adequacy
is used in this study as a ratio of shareholders’ equity
to total assets (Dao and Nguyen, 2020).
Table 1. Study Variables and Measurement
Variable
Variable Name
Label
Dependent
Liquidity
LIQ
Independent
Profitability
ROA
Credit Growth
CRG
Client Deposits
DPS
Financial Leverage
FLR
Capital Adequacy Ratio
CAR
Bank Size
BSZ
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
386
Volume 18, 2022
The regression model is designed to include the
different variables as follows:
LIQ = a + bROA + cCRG + dDPS + eFLR + fCAR +
gBSZ + E (1)
Where, a, is a constant, indicating the level of
liquidity when the value of the each independent
variable equals zero, and each of b, c, d, e, f, and g, is
also a constant referring for the corresponding
variable slope. ROA is the rate of return on total
assets, and CRG, refers for the rate of credit growth
from year to year. Where DPS refers for deposits,
FLR refers for bank financial leverage. Moreover,
CAR denotes for capital adequacy, and BSZ refers
for bank size.
Descriptive statistics including the mean,
standard deviation, minimum value, and maximum
value, are used in the analysis of collected data,
whereas except for the last hypothesis, the simple
linear regression method is used in testing the first
six hypotheses, and the multiple linear regression
method is used in testing the last hypothesis. The
decision base of null hypotheses acceptance or
rejection is based on the comparison between the
computed and the tabulated t-vale with regard to the
first six hypotheses, and the comparison between the
computed and the tabulated f-value regarding the last
hypothesis. Underline this rule, the null hypothesis is
accepted when the computed t or f-value is less than
the corresponding tabulated one, where in opposite,
the null hypothesis is rejected when the computed t
or f-value, is higher than the tabulated. Another rule
is used in the decision of null hypotheses acceptance
or rejection, is the comparison between the computed
and the predetermined coefficients of significance.
Following this rule, the null hypothesis is accepted
and its alternative is rejected, when the coefficient of
significance is higher than the predetermined one,
and in opposite, the null hypothesis is rejected and its
alternative is accepted, when the computed one is
less than the predetermined. In occasion, both
decision criterion lead to the same decision.
6 Results and Analysis
The main objective of the study is to examine some
internal possible factors affecting the liquidity of
listed commercial banks of Jordan. As mentioned
above, the simple linear regression method is used in
testing the first 6 hypotheses, whereas the last
hypothesis is tested using the multiple linear
regression method.
Descriptive Statistics
The mean, as a measure of central tendency, and the
standard deviation, as a measure of variation, in
addition to the minimum and maximum values, are
all descriptive statistics that are used for the different
variables, and the outputs are shown in table (2). The
table shows that the least liquidity ratio is 0.12, and
the highest one is 0.52, at a 0.31 mean, and 0.084
standard deviation. In occasion, liquidity is measured
using the debt ratio, where debt ratio is computed by
dividing total liabilities by total assets. The liquidity
of banks seems acceptable and within the range.
Regarding profitability, zero is the least profitability
ratio, whereas the highest is 0.03, where the highest
value refers for low profitability of the entire
industry. The mean of profitability is 0.012, with a
standard deviation of 0.005. These values indicate
that the commercial banks achieve low profitability.
Asset utilization seems low also, where the least is
0.02, and the highest is 0.07, with a 0.053 mean, and
0.011 standard deviation. The rate of credit growth
refers for a high rate of increase in credit. The least
value of credit growth rate is -0.16, and the highest is
0.85, with a mean of 0.084, and 0.121 standard
deviation. Higher credit than the average leads to
higher bad debts therefore; the commercial banks of
Jordan are required to adopt more conservative credit
policy. The minimum value of client deposits is 0.48,
and the maximum is 0.81, with a mean of 0.656, and
0.069 standard deviation. The client deposit is
acceptable, and can be considered within the range.
The least debt ratio is 3.53 and the highest is 12.32,
with a mean of 6.579, and 1.589 standard deviation.
The logarithms of total assets minimum value is 8,
and the maximum is 10.42, with a mean of 9.238,
and standard deviation of 0.429. Actually, except two
banks, the assets of the remaining Jordanian
commercial banks are low.
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
387
Volume 18, 2022
Table 2. Descriptive Statistics
No. of
Observations
Least
Value
Highest
Value
Mean
Standard
Deviation
Liquidity Ratio
156
0.12
0.52
0.310
0.084
Profitability
156
0.00
0.03
0.012
0.005
Capital Adequacy
156
0.08
0.22
0.1358
0.027
Credit Growth Rate
156
(0.16)
0.85
0.084
0.121
Client Deposits
156
0.48
0.81
0.656
0.069
Debt Ratio
156
3.55
12.32
6.579
1.589
Long. Assets
156
8
10.42
9.328
0.429
To check whether the data can be used for
analysis, the normal distribution, multicollinearity,
and correlation tests are made. The results of these
tests are appearing in table (3). The tolerance and
Variance Inflation Factors (VIF) had computed to
ensure that the variables are not overlapped. The VIF
for all variables is less than 10, suggesting the
absence of overlapping variables (Gujarati, 2003, p.
496).
Table 3. Variables Multicollinearity
Variable
multicollinearity
Tolerance
VIF
Profitability
0.856
1.168
Asset Utilization Efficiency
0.802
1.247
Credit Growth Rate
0.948
1.055
Client Deposits
0.839
1.191
Debt Ratio
0.759
1.318
Log. Assets
0.948
1.246
Hypotheses Testing
The simple linear regression method had employed in
testing the first sixth hypotheses, whereas, the
multiple linear regression method had used in the last
hypothesis test. The different hypotheses are tested
under 95 percent level of confidence, where this is
equivalent to 5 percent coefficient of significance.
First Hypothesis
This hypothesis is initiated to enable testing whether
bank profitability affects liquidity. Profitability is
measured using ROA, where ROA is the ratio of net
income to total assets. Using its null form, the
hypothesis is listed again as follows.
Ho1. The profitability of the commercial banks of
Jordan does not affect the liquidity position of these
banks.
The test of the hypothesis shows that R equals
0.251, and R2 equals 0.063, where the value of R2
means that ROA explains only 6.3 percent of the
change taking place in liquidity. Table (4), shows the
related coefficients and statistical values of the first
hypothesis test.
According to the information included in the
table, the outputs reveal that the computed t-value
equals 3.224, and 0.002 coefficient of significance.
Comparing the computed t-value with its
corresponding tabulated one, which equals 1.96, the
computed one seems higher than the tabulated.
Moreover, comparing between the coefficient of
significance with the predetermined one, that equals
0.05, the computed one seems less than its
corresponding predetermined one. Because the
computed t-value is greater than its corresponding
tabulated, and because the coefficient of significance
is less than its corresponding tabulated one, the null
hypothesis is rejected, while its alternative one is
accepted. This result means that profitability has a
positive significant impact on bank liquidity.
Table 4. 1St Hypothesis Test
Model
B
Std. Error
Beta
T
Sig.
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
388
Volume 18, 2022
Constant
0.257
0.018
14.345
0.000
Profitability
4.387
1.361
0.251
3.224
0.002
R = 0.251
R2 = 0.063
Adj. R2 = 0.057
Second Hypothesis
The second hypothesis of the study was developed to
enable examining whether the rate of credit growth
of commercial banks of Jordan, affects liquidity.
Credit growth is measured by deducting the
attributed credit amount to the preceding year from
the attributed credit amount of the most recent year,
and then dividing the result of subtraction by the
credit amount of the preceding or old year. The null
hypothesis is listed as follows.
Ho2. The rate of credit growth of commercial banks
of Jordan does not influence the liquidity of these
banks.
The test of the hypothesis reveals that R equals
0.020, and R2 equals zero. Because R2 equals , then
the rate of credit growth completely doesn’t
contribute in explaining the change in liquidity.
Table (5), shows the related statistics to the
hypothesis.
The data in the table reveals that t-value equals
0.252, and p-value is 0.802. This means that the
computed t-value is less than the tabulated, which
equals 1.96, and the computed p-value (sig.) is higher
than the predetermined, which is equal to 0.05.
Therefore, the null hypothesis is accepted, and its
alternative is rejected. This result means that liquidity
of banks is not influenced by the rate of credit
growth.
Table 5. 2nd Hypothesis Test
Model
B
Std.
Error
Beta
T
Sig
Constant
0.312
0.008
37.715
0.000
Credit
Growth
-0.014
0.056
-0.020
-0.252
0.802
R = 0.020
R2 = 0.000
Adj. R2 = - 0.006
Third Hypothesis
To enable testing whether client deposits affect
liquidity of commercial banks of Jordan, this
hypotheses is developed. The ratio of client deposits
is the relationship of client deposits to total assets.
The null form of the first hypothesis is again listed as
appearing below.
Ho3. The listed commercial banks of Jordan client
deposits does not significantly affect the liquidity
position of this group of banks.
The test of the hypothesis reveals 0.211 value of
R, and 0.045 value of R2. The value of R2 indicates
that client deposits explains 4.5 percent of the change
in liquidity. Table (6), includes, among different
values, the related coefficients of the third hypothesis
test.
The table reveals that the computed t-value is
2.678, and the computed coefficient of significance is
0.008. Because the absolute computed t-value is
greater than the tabulated, and because the coefficient
of significance is less than its corresponding
predetermined, the null hypothesis is rejected, and
instead the alternative one is accepted. In brief, the
analysis indicates that liquidity level is negatively
influenced clients deposits.
Table 6. 3rd Hypothesis Test
Model
B
Std. Error
Beta
T
Sig.
Constant
0.480
0.064
7.55
0.000
Client
Deposits
- 0.248
0.096
- 0.211
- 2.578
0.008
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
389
Volume 18, 2022
R = - 0.21
R2 = 0.045
Adj. R2 = 0.038
Fourth Hypothesis
Financial leverage is measured through the
relationship of total liabilities to total equity. To test
the impact of leverage on bank liquidity, the fourth
hypothesis is initiated. The null hypothesis is
appearing below.
Ho4. The Jordanian commercial banks’ financial
leverage does not significantly affect the liquidity
position of these banks.
The test of the hypothesis shows that R equals 0.305,
and R2 equals 0.093. Therefore, financial leverage
explains 9.3 percent of the change in liquidity. Table
(7), includes the different related statistics to the
hypothesis test.
The table refers that the computed t-value equals
3.979, and the computed coefficient of significance
(p-value) equals zero. Since the absolute computed t-
value is greater than the tabulated, and because the
coefficient of significance is less than the
predetermined, the null hypothesis is rejected, and its
alternative is accepted. The result refers for that there
is a negative effect of bank leverage on its level of
liquidity.
Table 7. 4th Hypothesis Test
Model
B
Std.
Error
Beta
T
Sig.
Constant
0.417
0.028
15.088
0.000
Fin. Leverage
- 0.016
0.004
- 0.305
- 3.979
0.000
R = - 0.301
R2 = 0.093
Adj. R2 = 0.087
Fifth Hypothesis Test
Capital adequacy is measured through the
relationship of total equity to total assets. The null
hypothesis of the hypothesis is listed for the second
time as follows:
Ho5. Commercial bank’s capital adequacy has no
significant influence on the liquidity of commercial
banks of Jordan.
The test of the hypothesis shows 0.299
coefficient of correlation (R), and 0.089 coefficient
of determination (R2), where the value of R2 means
that capital adequacy of commercial banks of Jordan
explains 8.9 percent of the change in liquidity. Table
(8), shows the related coefficients and statistical
values of the fifth hypothesis test.
The table indicates that the computed t-value
equals 3.89, and the computed coefficient of
significance equals zero. Since the computed t-value
is higher than the tabulated, and because the
computed coefficient of significance is less than the
predetermined, that equals 0.05, the null hypothesis
is rejected, while its alternative is accepted. In other
words, the test shows that there is a positive
significant influence of capital adequacy on bank
liquidity.
Table 8. 5th Hypothesis Test
Model
B
Std.
Error
Beta
T
Sig
Constant
0.182
0.034
5.394
0.000
Cap.
Adequacy
0.940
0.242
0.299
3.890
0.000
R = - 0.299
R2 = 0.089
Adj. R2 = 0.084
Sixth Hypothesis The base-10 natural logarithms of total assets, is the
most common used measure of size. To determine
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
390
Volume 18, 2022
whether bank size affects its liquidity, the sixth null
hypothesis is listed again as follows.
Ho6. The liquidity position of the commercial banks
of Jordan is not significantly affected by the bank
size.
The test of the hypothesis shows 0.283 coefficient of
correlation (R), and 0.074 coefficient of
determination (R2), where the value of R2 means that
a commercial bank size explains 7.4 percent of the
change in liquidity. Table (9), shows the related
coefficients and statistical values of the fifth
hypothesis test.
The data in the table indicates that the computed t-
value equals 3.658, and the computed coefficient of
significance equals zero. When the computed t-vale
is compared with its corresponding tabulated t-
values, this comparison indicates that the computed t-
value is more than the corresponding one , which
equals 1.96. The comparison between the coefficient
of significance and the predetermined one, which
equals 0.05, shows that the computed p-value (sig.) is
less than the predetermined, which equals 5 percent.
Since the computed t-value is greater than the
tabulated, and because the computed coefficient of
significance is less than the predetermined, the null
hypothesis is rejected, and its alternative is accepted.
This result means that bank size positively affects
liquidity.
Table 9. 6th Hypothesis Test Coefficients
Model
B
Std.
Error
Beta
T
Sig.
Constant
- 0.210
0.142
- 1.475
0.142
Bank
Size
0.056
0.015
0.283
3.658
0.000
R = 0.283
R2 = 0.080
Adj. R2 = 0.074
Seventh Hypothesis
This hypothesis is initiated to enable testing the total
impact of the entire group of the considered
independent variables together, on the liquidity
position. Therefore, the hypothesis examines the
entire grouping effect of all independent variables,
which had examined individually above. Multiple
linear regression is used in testing this hypothesis.
The hypotheses is again listed, as appearing below.
Ho7. There is no grouping effect of bank
profitability, credit growth rate, client deposits, debt,
capital adequacy, and bank size, on the liquidity
position of commercial banks of Jordan.
Table (10) shows the main coefficients and
values of the 7th hypothesis test. The table reveals
that the correlation coefficient (R) equals 0.486, and
the determination coefficient equals 0.236, which
means that the variables that are taken into
consideration in the study explain in total 23.6
percent of the change occurring at liquidity.
Considering the table, it shows that f-value
equals 7.659, and the computed p-value is zero. The
comparison between the computed and the tabulated
f-value reveals that the computed is greater than the
tabulated. Moreover, the comparison of the computed
coefficient of significance with the predetermined
one (p-value), which equals 0.05, reveals that the
computed is less. Because of the results of these two
types of comparison, the null hypothesis is rejected,
and the alternative one is accepted. This result means
that the group of independent variables, including
profitability, rate of credit growth, client deposits,
debt, capital adequacy, and bank size, when taken
together as one unit, they have a significant effect on
bank liquidity.
Table 10. 7th Hypothesis Test
Model
Sum
of
Square
s
Degre
es of
Freed
om
Mean
Square
F
Sig
Regression
0.262
6
0.044
7.659
0.000
Residual
0.849
149
0.006
Total
1.11
155
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
391
Volume 18, 2022
Model
B
Std.
Error
Beta
T
Sig.
Constant
-0.477
0.290
-1.643
0.102
Profitability
2.717
1.359
0.156
2.000
0.047
Credit
Growth Rate
0.032
0.051
0.046
0.619
0.537
Client
Deposits
-2.11
0.095
-0.173
-2.223
0.028
Financial
Leverage
0.012
0.016
0.217
0.717
0.475
Log. Assets
0.067
0.015
0.339
4.542
0.000
Capital
Adequacy
1.380
0.925
0.439
1.491
0.138
Therefore, when the constants of the multiple
regression model are solved, the solved model
appears as follows.
LIQ = - 0.477 + 2.717ROA + 0.032CRG – 2.11DPS
+ 0.012FLR + 1.38CAR + 0.067BSZ – 1.831 (2)
6 Findings and Conclusions
The main objective of the study is to examine some
internal factors that may affect the liquidity position
of commercial banks of Jordan. The related literature
review and the findings of prior researches had
considered carefully. Moreover, the appropriate
secondary data related data was collected through the
website of ASE, and analyzed using descriptive
statistics, and the hypotheses were tested using both
of the simple and multiple linear regression methods.
Several possible indicators of liquidity were analyzed
in the study including, bank profitability, rate of
credit growth, financial leverage, capital adequacy,
customer deposits, and bank size. Using the simple
and multiple linear regression methods, the results
showed several beneficial findings, especially for the
managements of commercial banks of Jordan.
Based on hypotheses testing, the study finds that
profitability, capital adequacy, and bank size, each of
which, has a positive significant impact on liquidity
position of banks. This means that, as the bank
profitability is higher, as its liquidity is better, and as
the bank equity ratio is greater, as its liquidity is
higher or better. Moreover, a bank with higher equity
ratio to total assets, are in better liquidity situation.
Because the study finds that bank size has a positive
impact on liquidity, and because bank size is
measured by natural logarithms of total assets, then
maintaining more total assets leads to better level of
liquidity.
Based on hypotheses testing, the study also finds
that financial leverage and client deposits, each of
which, has a negative impact on liquidity position of
banks. This means that more borrowing to finance
the assets of banks, leads to low liquidity. Despite
that more customer deposits are assumed to improve
liquidity, but the study shows a different finding.
This is because more client deposits leads to more
lending transactions, because banks pay interests for
depositors, therefore, to cover the cost of client
deposits, and to earn more profits, commercial banks
grant more loans to customers, and this is why
deposits leads to less liquidity. More deposits
received by a bank leads to more credit granted to
customers by the same bank, and because granting
more credit needs for easy credit policy, where this
will increase bad debt, because more customers will
find themselves unable to pay the principal, nor the
interests. With regard to rate of credit growth, the
study finds no significant impact of this variable on
commercial banks’ liquidity. The findings of the
study are in consistence with Bansal and Bansal
(2012), Lotto and Mwemezi (2018), Nguyen and Vo
(2021), Mohamad, (2015), et al.
References:
[1] Bank for International Settlements, (2008)
Basel Committee on Banking Supervision:
Principles for Sound Liquidity Risk
Management and Supervision.
[2] Bansal, R. and Bansal, V. (2012), A Research
Paper on Determinants of Corporate Liquidity
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
392
Volume 18, 2022
in India, International Journal of Marketing and
Technology, Vol. 2, No. 4, pp. 103-117.
[3] Ben Moussa, M. A. (2015), The Determinants
of Banks Liquidity: Case of Tunisia,
International Journal of Economics and
Financial Issues, Vol. 5, No. 1, pp. 249-259.
[4] Bista, R. B., and Basnet, P. (2020),
Determinants of Bank Liquidity in Nepal,
Quantitative Economics and Managerial
Studies, Vol. 1, No. 6, pp. 390-398.
[5] Dao, B. T. T. and Nguyen, K. A. (2020), Bank
Capital Adequacy Ratio and Bank Performance
in Vietnam: A Simultaneous Equations
Framework, Journal of Asian Finance,
Economics and Business, Vol. 7, No. 6, pp. 39-
46.
[6] Ferrouhi, E., (2014), Bank Liquidity and
Financial Performance: Evidence from
Moroccan Banking Industry, Business Theory
and Practice, Vol. 15, No. 4, pp. 351-361.
[7] Gideon, F., et al (2012), Bank Liquidity and
the Global Financial Crisis. Journal of Applied
Mathematics, Vol. 2012, No. 1, pp. 1-27.
[8] Gujarati, D.N. Basic Econometrics. 4th
Edition, McGraw-Hill, New York, (2003).
[9] Laštůvková1, J. (2016), Liquidity Determinants
to the Selected Banking Sectors and their Size
Groups, Acta Universitatis Agriculturae Et
Silviculurae Mendelianae Brunenses, Vol. 64,
No. 3, pp. 971-978.
[10] Lotto, J. and Mwemezi, J. (2015), Assesing the
Determinants of Bank Liquidity: Experience
from Tanzanian Banks, African Journal of
Finance and Management, Vol. 24, No. (1-2),
pp. 76-88.
[11] Mohamad, A. (2016), Determinants of Bank
Liquidity: Empirical Evidence on Turkish
Banks, Master Thesis, Institute of graduate
studies and research.
[12] Moore, W. (2010), How do Financial Crises
Affect Commercial Bank Liquidity? Evidence
from Latin America and the Caribbean. MPRA
Paper, No. 21473.
[13] Nguyen, H. T. V. and Vo, D. V. (2021),
Determinants of Liquidity of Commercial
Banks: Empirical Evidence from the
Vietnamese Stock Exchange, Journal of Asian
Finance, Economics and Business, Vol. 8, No.
4, pp. 699-707.
[14] Saunders, A., Cornett, M. M., & McGraw, P.
A., Financial Institutions Management: A Risk
Management Approach,McGraw-Hill/Irwin,
2006.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the Creative
Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en_US
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2022.18.38
Mohammed Ibrahim Sultan Obeidat,
Nadeen Mohammed Adnan Mohammed Yasin Darkal
E-ISSN: 2224-3496
393
Volume 18, 2022