Macroeconomic and Bank-Specific Determinants of Non-Performing
Loans in Nigeria
MICHAEL OSUNKOYA, OCHEI IKPEFAN, FELICIA OLOKOYO
Department of Banking and Finance,
Covenant University, Canaanland Ota,
Km 10, Idiroko Road, Ota, Ogun State,
NIGERIA
Abstract: - Proper management of banks’ assets is the most sought panacea by bank managers in Nigeria. Lower
quality of loans in the banking system can lead to higher loan loss provisions and affect banks’ capital adequacy
ratios. This study examines how Non-Performing Loans (NPLs) of Nigeria’s Deposit Money Banks (DMBs)
respond to selected macroeconomic and bank-specific determinants. Using the data within the sample period of
1981-2019 with Autoregressive Distributed Lag (ARDL) model, the study establishes that in the short-run, the
level of NPLs in Nigeria is affected by these macroeconomic determinants namely, the unemployment rate
(UNEMP), gross domestic products growth rate (GDPG) and exchange rate (EXRT) as well as bank-specific
determinant (loan-to-deposit) ratio (LDR). However, in the long-run, GDPG and EXRT have a positive and
significant influence on NPLs. The variables respond in line with our apriori expectation, however, unemployment,
inflation and loans to deposits ratio are insignificant and appear not to affect NPLs in Nigeria in the long-run. The
study recommends that government should ensure that the naira is properly managed as deterioration in its value
portends a grave impact on the rate of non-performing loans. Also, improved infrastructures like good roads, water
and power would enable the borrowers to fulfil repayment plans on time. A robust economy is important for
borrowers to redeem their loan obligations in due time. This can be achieved by ensuring that loans assessed are
channeled to more productive sectors of the economy.
Key-Words: - Bank-Specific Determinants, Deposit Money Banks, Economy, Inflation, Macroeconomic
Determinants and Non-Performing Loans
Received: November 28, 2022. Revised: May 2, 2023. Accepted: May 15, 2023. Published: May 24, 2023.
1 Introduction
The macroeconomic determinants of non-performing
loans have been generating substantial interest from
scholars from developing and developed economies.
Following the outburst of the 2007-2009 global
financial crisis (GFC), the prevalence of increasing
non-performing loans (NPLs) ratio by deposit money
banks (DMBs) despite banks’ recapitalization and
subsequent consolidation of banks in Nigeria are still
a pointer to the fact that the Nigerian banking sector
is still largely vulnerable to macroeconomic shocks.
Meanwhile, proper management of banks’ assets,
which, largely are loans and other credit facilities, is
the most sought panacea by bank owners and
managers in Nigeria. However, lower quality of loans
in the banking system or increasing tendency in
NPLs can lead to higher loan loss provisions, which
can negatively affect the profitability and capital
adequacy ratios of banks. In view of these, the focus
of this study is, how does NPLs of Nigerian DMBs
respond to macroeconomic determinants in Nigeria.
This study attempts to answer this question by
examining empirically, the effect of macroeconomic
variables on NPLs of the DMBs in Nigeria.
In the same vein, the effect of NPLs is enormous
for the banking and financial services industry and
the economy at large. According to [1], banks’
choice for investing a sizeable portion of their assets
in loans and advances (risk assets), despite their
higher risk profile, is because it generates higher
returns. The going concern ability of banks can be
threatened as their ability to create credit is hampered
due to the toxicity of their risk assets. This
deterioration in the risk assets extends to the entire
banking industry resulting from the contagion effects
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1153
Volume 20, 2023
on healthy banks, as the resulting illiquidity of the
affected banks tends to spread from one bank to the
other. As a result, the economy at large bears the
brunt since the instruments for transmitting monetary
policy is effected by the Central Bank through
deposit money banks. The need to deal with a fragile
economy that is faced with macroeconomic
challenges like low economic growth, high
unemployment and volatile exchange rate,
necessitates the need for stern banking reform.
Therefore, our findings in this study should have
significant policy direction for designing stabilization
and adjustment reforms for Nigeria and, as a template
for other emerging economies in Africa. To achieve
this task, this study is divided into five different
parts. After the first part, which is the introductory
and Nigeria’s NPL profile, the second part is a brief
literature review of the subject matter. Section three
focuses on the theoretical framework and method of
analysis, and section four deals with the result and
discussion of findings. Finally, section five centered
on recommendations and conclusions.
Graphical Illustration of Nigeria’s NPL Profile
from 1981 to 2019
The World Development Index (WDI), [2], has the
NPL ratio for the period 1981 to 2019, being 39
years. NPL ratio is measured as ratio of total non-
performing loans to total loans and is shown in Table
1.
Table 1. Nigeria’s NPL Profile from 1981 to 2019
Year 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990
NPLs to Total
Gross Loans (%)
10.30 11.20 12.30 13.20 14.30 15.20 14.60 15.70 17.60 16.60
Year 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
NPLs to Total
Gross Loans (%)
15.60 16.70 15.40 16.70 17.34 19.30 18.50 19.40 25.60 22.60
Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
NPLs to Total
Gross Loans (%)
19.70 21.40 20.50 21.60 18.10 8.80 9.51 7.20 37.26 20.15
Year 2011 2012 2013 2014 2015 2016 2017 2018 2019
NPLs to Total
Gross Loans (%)
5.78 3.71 3.40 2.96 4.87 12.82 14.81 11.68 6.04
Source: World Development Index (WDI) 2021
Fig. 1: Nigeria’s NPL ratios for 39 years from 1981 to 2019
Source: World Development Index (WDI) 2020
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1154
Volume 20, 2023
Figure 1 depicts the trend of non-performing loans
(NPLs) in the Nigerian banking industry for the
period 1981 to 2019. The NPL ratio which stood at
10.3% in 1981 witnessed a mild increase (though
with little decline in some years) till 1998 when it
stood at 19.4%. The NPL ratio increased further by
32% in 1999 and stood at 25.6%. It declined with
effect from the year 2000 up to 2008. The ratio
witnessed successive declines especially effective
2005 to 2008 following the CBN’s recapitalization
policy which compelled Nigerian banks to beef up
their capital from N2billion to N25billion. Higher
capital funds of the emerged consolidated and
stronger /bigger DMBs improved the NPL ratio as
more funds were deployed for lending purposes
following a shift in supply of loanable funds. Credit
standards were lowered following the supply shift in
loanable funds. With keen competition to lend, big-
ticket loans were made and loan loss surfaced a few
years after. This is in line with [3] proposition. This
pushed up the NPL ratio, which peaked at 37.25% in
2009, following the huge loan loss from margin
loans, drastic fall in crude oil price, the global
meltdown and the collapse of the Nigerian Capital
Market.
The post-2009 average of the NPLs ratio is 8.62%
(that is, from 2010 to 2019) - the period after the
Assets Management Corporation of Nigeria
(AMCON) was created. Comparatively, the average
NPL ratio of Nigerian DMB’s pre-consolidation
period (that is, from 1991 to 2005) stood at 10.6%.
This portends that the NPL ratio was very high prior
to the take-over of the toxic loans by AMCON. An
all-year reduction of 17.1% was recorded as the NPL
ratio fell from 37.3% in 2009 to 20.1% in 2010 as a
result of AMCON’s acquisition of the eligible toxic
assets of DMBs. This same reason accounts for a
further decline of 14.4% in 2011 when the ratio stood
at 5.8%. The NPL ratio has maintained a steady rise
from about 3.0% in 2014, to about 4.86% in 2015, to
11.7% in 2016 and 6.0% in 2017. In the past two
years of 2018 and 2019, the ratio has been declining.
The NPL ratio of 6.03% in 2019 is higher than the
prescribed CBN tolerance limit of 5%. Also, when
we consider the total loan portfolio in excess of
N4.7trillion in the books of AMCON as of December
2019, the NPL is in a grave position.
2 Brief Literature Review
In the literature, several previous studies (see [4], [5],
[6], [7], [8], [9], [10] and [11]) had used different
econometric methodologies to assess the impact of
non-performing loans in both developing and
developed economies, with evidence that quality of
bank loans are impacted by factors that are both
internal and external to banks.
The internal factors that determine the level of NPLs
are referred to as bank-specific variables, which
include corporate governance, the size of banks’
assets, loan growth pattern, loan-to-deposit ratio,
capital adequacy ratio and some other indicators.
Factors that are external to the banks are
macroeconomic variables. These include inflation,
gross domestic product (GDP), exchange rate and so
on. Meanwhile, the last GFC of 2007-2009 has left so
much to be investigated in both developing and
developed economies, due to the negative effect of
NPLs, which has characterized past banking
distresses and economic crises across the world.
Similarly, the on-going COVID-19 global pandemic
is a pointer to the fact that economies around the
world could be more susceptible to economic shocks
that are both domestic and foreign.
The authors in their study in [12] used panel data
analysis for a sample taken from 20-Jordanian banks
for the years 2006-2014. It explores the macro and
micro level variables that predict deterioration in the
loan quality at the early stage. The study
recommends proactive corrective action from the
regulatory authority towards the problematic
financial institutions. The study uses a risk-based
computation of deposit insurance premium, as well
as an enhanced supervisory framework, to control the
banking system, thereby securing the stability of the
economy. The result of their findings shows that the
selected indicators display a significant relationship
with the level of NPL ratio. A negative relationship
exists between NPL and GDP growth on one hand
and profitability and risk on the other hand.
The authors in their study in [13] posit a mixed
effect of exchange rate on NPL which is either a
balance sheet effect or an income effect. A balance
sheet effect postulates that depreciation of domestic
currency makes NPL and debt burden to increase.
Also, when the nominal exchange rate appreciates,
foreign denominated loans increase the debt burden
of unhedged foreign loans through increased loan
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1155
Volume 20, 2023
repayment. Thus, there is a positive relationship
between nominal exchange rate and NPL rate.
The authors in their study in [14] which covers the
period between 1981 and 2017 on the impact of
external reserves shocks on selected macroeconomic
indicators find that one standard deviation shock to
external reserves negatively impacts the exchange
rate throughout the quarters. Also, when there is a
strong positive shock to external reserves, the
currency tends to appreciate, that is, the exchange
rate falls.
The authors in [1] formulate four (4) hypotheses
namely the bad luck hypothesis, bad management
hypothesis, skimping hypothesis, and moral hazard
hypothesis. In their views, in the long-run, banks are
liable to experience higher loan delinquencies if they
spend less on their loan underwriting and monitoring.
Also, banks that devote little or no effort to ensuring
that high loan quality is booked, will engage in more
cost-efficient debt recovery.
The study in [15] examines the relationship
between Return on Assets (ROA) and the Tunisian
banking system stability and economic activity
indicators -NPL, Credit to Loan Ratio, Solvency
Ratio. The study considers 10 banks operating in
Tunisia, using quarterly data for the period Q4, 2010
to Q4, 2019. OLS (GMM) method was used to
estimate the model. It finds that the profitability of
Tunisian banks is significantly affected by the level
of NPLs. Thus, banks with lower non-performing
loan tend to have higher profitability and there exists
an NPL ratio threshold of about 27%, the level at
which, the banks’ profits may be fully extinguished.
The authors in their study in [16] investigate the
rising incidence of NPLs and economic performance
in Nigeria. Time series data were used for the study
which covers the period of 1984 – 2012 and
considers NPLs, inflation rate, gross domestic
product, lending rate. An Ordinary Least Square
technique with the aid of E-view as the statistical tool
was used for its data analysis. The findings reveal
that causality between NPLs, lending rate, and the
inflation rate is insignificant and the study concludes
that non-performing loans have a significant
influence on gross domestic product. It is
recommended that the government should invest in
sectors with growth potentials and pay their loan as at
when due as well as ensure early payment of
contractors and other suppliers.
The authors in their study in [7] investigate the
determinants of non-performing loans in Nigeria.
They found that bank size has a positive and
significant relationship with NPL. ROA has a
negative insignificant relationship with the NPL,
loan-to-deposits ratio has a positive and significant
relationship with NPL. Inflation was found to have a
positive but insignificant relationship with NPL. The
lending rate increases the NPL ratio while the
exchange rate increases the NPL ratio. Also, an
increase in the real GDP was found to reduce NPLs
both in the short and long-run with the impact being
significant only in the long-run. Finally, their study
finds that a reduction in the unemployment rate
improves the NPL ratio.
This study is a departure from most of the existing
studies by proposing a more robust methodology,
which is Autoregressive Distributed Lag (ARDL) to
investigate the effect of macroeconomic and bank-
specific variables on Nigerian DMBs’ NPL ratio.
3 Theoretical Framework
The theory argument of non-performing loans as it
links to bank stability is traceable to three important
pillars, which are information asymmetry, adverse
selection and moral hazard theories. According to
[17], information on the basic issues around loan
default further degenerates into banking system
instability. In the literature, the author in [18]
pioneered the information asymmetry theory. This
theory posits that the task of selecting good
borrowers from bad ones is onerous and may lead to
the problems of adverse selection and moral hazard.
Adverse selection theory as initiated by [18] was
later extended by [19]. The theory emanates from the
view or, on the assumption that banks (lenders) are
not always, well equipped with adequate information
to identify credit-deserving borrowers from or
amongst the numerous loan seekers, who have
various credit risk exposures at the point of
requesting a loan. Also, in keeping with the ‘lemons’
outcome, banks are incentivized to offer only their
worst loan assets for sale, rather than selling better
quality assets at prices that would undervalue them
[20]. If the lending banks can recoup the loans, why
then would they want to sell the loan to a bad loan
bank at a higher discount? This bothers on moral
hazard.
Financial intermediaries, especially bankers, are
more likely prone to give out loans to high-risk
borrowers in their pursuit of profitability objectives
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1156
Volume 20, 2023
thereby taking an excessive risk, [1]. Low-
quality/high-risk borrowers tend to default on loans
and may be unconcerned about harsh lending
conditions, [21]. In furtherance to this, the need for
lenders to share and reveal important information on
loan seekers, cannot be overemphasized, thereby
reducing adverse selection problems. The Credit
Bureau Management System (CBMS) that is in place
in Nigeria today whereby banks upload the quality
status, exposures and collateral details of their
borrowing customers in a repository system hosted
by the CBN and private domains increases
information sharing amongst banks.
3.1 Method of Analysis
This study examines the effect of macroeconomic
determinants of NPLs in Nigerian Deposit Money
Banks. In addressing the issues arising from this
objective, this study uses a multivariate econometric
technique which is Autoregressive Distributed Lag
(ARDL) to analyse the objective with annual data
covering thirty-nine years for the period 1981-2019.
3.2 Model Specification
The basic model for the study is presented in its
linear regression form, which links NPLs to the
macroeconomic and bank-specific determinants. This
is presented in its implicit function form:
NPLt = f(UNEMPt, GDPGt, INFt, EXRTt, LDRt,)
(1)
Where:
NPLt which is non-performing loans is the dependent
variable over a period t;
UNEMPt represents the unemployment rate of the
Nigerian economy over a period t;
GDPGt represents the gross domestic products
growth rate over a period t;
INFLt represents the inflation over a period t;
EXRTt represents the exchange rate over a period t;
and
LDRt represents the loans to deposits ratio over a
period t;
The variables in equation (1) are described in the
underlisted explanations:
i) Non-Performing Loans (NPLs): These are part of
the total loan portfolio that borrowers are unlikely
and unable to repay. The higher the non-performing
loans, the higher the loan losses and the lower the
income accruable from the loan portfolio. Bank
management has to ensure that they maintain a
healthy loan portfolio thereby keeping loan losses at
the minimum.
ii) Unemployment Rate (UNEMP): The
unemployment rate is the proportion of people that
are out of work due to the unavailability of jobs
relative to the total labour force. During a period of
high unemployment, the loan default rate increases as
households are unable to service their loan
obligations. The types of unemployment include
cyclical or Keynesian, structural, frictional and
classical unemployment. This study adopts
unemployment as one of the determinants of
macroeconomic variables because of its negative
effect on individuals’ incomes which may likely
result in loan defaults.
iii) Gross Domestic Product Growth (GDPG):
This is the total value of goods and services produced
in a nation in a particular year, adjusted for inflation.
The GDP value is calculated at factor-cost, that is,
excluding taxes imposed and subsidies on goods and
services. Empirical studies show that there is a
relationship between a bank loan and GDPG, and it is
one of the determinants of non-performing loans,
[22], [23] and [24].
iv) Inflation (INFL): This is a persistent rise in the
general level of prices of goods and services in an
economy. Inflation makes borrowers to be unable to
service their loans following a fall in the value of
money (thus, the value of money is measured in
terms of the quantity of goods and services that it can
buy). So, borrowers would have to ration their
limited income amongst competing bills. Inflation in
Nigeria has remained predominantly in a double digit
in the past three (3) decades, [11]. From the
literature, INFL is one of the determinants of non-
performing loans, [9] and [4].
v) Exchange Rate (EXRT): This is the value for
which a nation’s currency exchanges with another
currency. The rise or fall of the real exchange rate
indicates the weaknesses or strength of a nation’s
currency relative to international currencies. It also
serves as the standard for explaining the
attractiveness of local industries in the global market.
The exchange rate is a significant macroeconomic
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1157
Volume 20, 2023
variable used in formulating economic policies and
reforms. These policies assist in accelerating the
achievement of set macroeconomic objectives.
Generally, the exchange rate of a nation depends on
the demand for and supply of a nation’s currency.
The rate appreciates with increased demand while
increase in the supply makes the currency to
depreciate, ceteris paribus. Similarly, changes in
consumers’ tastes and preferences for foreign
products, changes in relative income, inflation rate,
interest rate, and speculative activities on a nation’s
currency may change the demand for that nation’s
currency, and the exchange rate is thus modified.
vi) Loans to Deposits Ratio (LDR): This is the ratio
that relates the total loan portfolio to total deposit
liabilities. It is one of the liquidity indicators and it
indicates if the bank is not over-lending thereby
putting liquidity at risk. Banks as financial
intermediaries are expected to keep some funds in
liquid form so that they will not suffer a lack of
liquidity should the depositors and creditors suddenly
opt for the repayment or liquidity of their matured
investments in the bank. LDR is one of the
determinants of non-performing loans, [25].
3.3 Technique of Estimation
Autoregressive Distributed Lag (ARDL)
To achieve the specific objective, which is to
examine the effect of macroeconomic determinants
on NPLs in Nigerian DMBs, the study adopts
Autoregressive Distributed Lag (ARDL) model
approach. This estimation technique of analysis is
also regarded as the Bounds test, as advanced by
[26]. Unlike many other previous cointegration
techniques; including [27], the ARDL cointegration
approach is more reliable, because it takes into
cognizance, a sample size of around 25 years. It also
allows for the variables with different orders of
integration to be analysed together. The combination
of the series of I(0) and I(1) in ARDL presents a
better approach to the cointegration test in the
literature. The analysis procedure includes the
following:
i) Test for Stationarity
This is a test for unit root among variables of interest
in a model. Stationary series provide plausible
regression as compared to series that are not. A non-
stationary series tends to give spurious results.
Therefore, this study applied the Augmented Dickey-
Fuller (ADF) unit root test to ascertain the level of
stationarity of the variables in the model.
ii) Unit Root Test
The application of Autoregressive Distributed
Lagged (ARDL) and Vector Autoregression (VAR)
approaches requires the absence of unit roots in
variables, as demonstrated by [28]. This assumption
explains a position in standard regression analysis
that all the variables being tested must not have a unit
root. However, many macroeconomic time series
variables are often not found stationary, these
variables trend up and down over time. Hence, before
any meaningful regression analysis can be carried out
on time series variables, it is expected that the test for
stationarity should be done to avoid biased estimates
and spurious results. From the literature, stationery
series has a finite variance, transitory shocks from the
mean, and a tendency for the series to return to their
mean value. This explains that a stationary series has
a mean and variance; and is consistent over time.
iii) Cointegration Test
There are various estimation techniques identified in
the macroeconomic literature that can be adopted to
estimate the co-integration relationship among
variables. For instance, Johansen’s approach can be
used for multivariate cointegration analysis [27].
However, this study adopts the ARDL -Bounds test
approach. This approach was made popular by [26].
This is because ARDL gives a more plausible output
when the series is a combination of I(0) and I(1). The
study uses this methodology to estimate the model
and empirically analysed the long-run relationship, so
as to be able to explain the dynamic interactions
amongst the variables of interest. This approach is
built on the error correction model (ECM) technique.
The ECM involves estimating the ARDL model by
Ordinary Least Square (OLS). This is with a view to
testing if a long-run relationship exists amongst the
variables or not. [29] posit that ECM estimation tests
if the variables are statistically significant at their
lagged levels or not.
This will further explain, whether the null
hypothesis of the existence of no long-run
relationship will not be rejected or not. To achieve
this, a Wald test (which is related to F-statistics for
Bounds-testing) for the joint significance of the
lagged levels of the variables will be performed,
where the null hypothesis is tested against the
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1158
Volume 20, 2023
alternative. If the F-statistics is above the upper
critical value, the null hypothesis of no long-run can
be rejected, irrespective of the integration order of
the variables. Otherwise, if the statistics fall below
the lower critical values, then the null hypothesis can
be accepted. However, if the F-statistics falls
between the lower and upper critical values, the
result is inconclusive. As a result of this, the
asymptotic distribution of the F-statistics is non-
standard under the null hypothesis of no co-
integration, whether the series are I(0) or I(1), [28].
After the long-run relationship was established
amongst the variables, we estimated the long-run
elasticities using the appropriate lag length. There are
three most commonly used information criteria in the
literature, these are the Akaike information criterion
(AIC), the Bayesian information criterion (BIC) and
the Hannan-Quin information criterion (HQIC). This
approach entails successively increasing the lags
from the smallest to the largest lag selected by the
information criteria. Thereafter, use the lag-length
that eliminates serial correlation in the residual, [30].
This assists in deriving the related error correction
for calculating the adjustment coefficients of the
ECM. Hence, the short-run effects of the ECM are
captured by the coefficients of the first differenced
variables in the ECM model.
Moreover, the ARDL approach help in the
estimation of the long-run, short-run and the
adjustment process in the model. Hence, Equation (2)
is the ARDL representation of how macroeconomic
variables impact non-performing loans in Nigeria
within the sample period 1981-2019, and this is
constructed as follows:
Where:
NPL is the Non-performing Loan;
UNEMP is the Unemployment Rate of the Nigerian
economy;
GDPG is the Gross Domestic Products Growth Rate;
INFL is the Inflation Rate;
EXRT is the Exchange Rate;
LDR is the Loans to Deposit Ratio.
Also, where:
λ0 is the intercept term
λ1 is the coefficient of the Non-Performing Loan
λ2 is the coefficient of the Gross Domestic Products
Growth rate
λ3 is the coefficient of the Inflation Rate
λ4 is the coefficient of the Exchange Rate
λ5 is the Loans to Deposits Ratio
is the stochastic error term which signifies all the
variables that affect the dependent variable but are
not considered in the model.
From equation (2),
is the first-difference
operator. The the long-run
coefficients, while are the
short-run coefficients. µt represents the short-run
coefficients.
iv) Wald (Bound) F-Test
The Wald test was applied to establish if amongst the
variables, there exists a long-run relationship in
equation (2), where:
Ho:
0
Ho:
0
The null hypothesis (Ho) shows the non-existence
of a long-run relationship for the model series while
the alternative (H1) connotes the presence of a valid
long-term influence of the explanatory variable on
the explained. In this approach, the F-test computed
is examined against its F-ratio. An F-test above the
upper bound level leads to the rejection of the null
hypothesis, otherwise, it is accepted. However, the F-
bounds test between the lower and upper bound is
considered indeterminate.
v) Error Correction Term
After establishing the long-run estimation of the
model, it is equally important to estimate the short-
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1159
Volume 20, 2023
run coefficients of the model along the path of the
error short-run dynamics. The short-run result
ascertains the difference between true values and the
values of the estimated parameter. Further, the rate of
correction of the short-rum movement towards the
long-term steady state could be evaluated. The
assumption here is that the error estimate is negative
and significant statistically and confined within the
values of 1 and 0.
Table 2. Unit root test
Variables
ADF @ levels
5% critical
value
ADF @ 1st
diff
5% critical
value
Remarks
NPL
-3.320896
-2.941145
N/A
I(0)
UNEMP
0.335008
-2.941145
-4.760507
-2.943427
I(1)
GDPG
-4.158314
-2.941145
N/A
N/A
I(0)
INFL
-2.913973
0.0531
-5.679381
-2.943427
I(1)
EXRT
1.400113
-2.941145
-4.257516
-2.943427
I(1)
LDR
-4.675762
-2.948404
N/A
N/A
I(0)
Source: Authors’ Computation with EViews (2020). Note: N/A = Not Applicable
The study model measures how macroeconomic
variables impact non-performing loans in Nigeria
within the sample period 1981-2019 is specified as
follow:
where Δ connotes differencing operation, ECMt−1 is
the lagged error correction term indicating the
reaction of the system in adjusting to the equilibrium
state in equation (2). In equation (3), η is the value of
the speed of ECM during the transition to the long-
run and it is expected to exhibit a negative and
significant sign for a co-integrating relation to exist
in the long-run.
4 Results and Discussion
4.1 Test for Stationarity
Test for stationarity investigates the evidence of unit
root in the series. A stationary or integrated series do
not vary with time in its functional distribution.
Stationary series helps to improve the extent to which
the results of the parameter estimate could be relied
upon. The Augmented Dickey-Fuller (ADF) test was
employed in measuring if the variables are stationary
or not.
4.2 Unit Root Test
The ADF test in Table 2 suggests that only three of
the variables - NPL, GDPG and LDR are stationary
at level, while the remaining variables -UNEMP, INF
and EXRT are differenced once to achieve the
stationary state. The study result necessitates the
utilization of the ARDL as advanced by [26].
4.3 ARDL Bounds Cointegration Test
The ARDL Bound result portrayed in Table 3
describes the existence of co-integrating relations in
the study model which establishes that the variables
tend not to diverge over the long run. As indicated by
the result of the F-statistic (5.402341) being greater
than the lower (2.62) and upper (3.79) critical bounds
at a 5 percent level of significance.
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1160
Volume 20, 2023
Table 3. Bound test result
F-bounds test
Null hypothesis: No
levels relationship
Test
statistic
Value
Significance
level
I(0)
I(1)
F-statistic
5.402341
10%
2.26
3.35
K
5
5%
2.62
3.79
2.5%
2.96
4.18
1%
3.41
4.68
Source: Authors’ computation using E-Views 10
4.4 Long-run Result
As shown in Table 4, it is obvious that GDPG and
EXRT have positive statistically significant
relationships with the dependent variable, that is, the
NPL ratio. This explains that a one percentage
increase in GDPG will bring about a 0.8395
percentage increase in NPL and a one unit increase in
EXRT will bring about 0.2376 percentage increase in
NPL. Similarly, from Table 4, GDPG and EXRT
significantly contributed to the incidence of NPLs at
5 percent and 1 percent levels of significance
respectively. This result of EXRT confirms the
earlier study by [7], [31] and [32] whose findings
show exchange rate is one of the determinants of
non-performing loans. However, this is in contrast
with the study by [8] that finds EXRT and economic
growth are insignificantly related to non-performing
loans. From this point of view, even though
consistent economic growth should improve the
financial stability of borrowers, and subsequently
increase their propensity to repay loans, this study
finds the relationship between economic growth and
NPL to be otherwise, hence indicating a positive and
significant. This implies the existence of non-
inclusive growth as an increase in the GDPG did not
bring down the level of NPL and exchange rate
deterioration makes the non-performing loan ratio
worsen in the Nigerian banking sector.
The coefficients of UNEMP and INFL are
insignificant and appear not to influence the NPL
level of Nigerian DMBs. This position aligns with [7]
whose study finds that UNEMP and INFL are
insignificant in the determination of NPL. Our result
also finds INFL to be insignificant. This is consonant
with the study by [8]. In the study by [33], UNEMP
and INFL are found to be insignificant in the
determination of NPL in the Sri Lanka banking
sector. However, the study by [8] contradicts this
study in terms of UNEMP which finds it significant.
Meanwhile, contrary to expectation, the coefficient
of inflation is insignificant and as a result, does not
affect NPLs. This position is not different from the
earlier argument of [8], who established in their study
that the increase in non-performing loans ratio in
Nigerian banks is attributed to deficiencies in credit
administration including poor credit assessment,
undue meddling in loans processing, deficient
collateral security, amongst others. This is evident in
the case of Nigeria, where, despite the high rate of
inflation in the last few decades, the economy is still
growing. Therefore, we can conclude that inflation
may not be a critical determinant of NPLs in Nigeria.
The loan-to-deposit ratio (LDR) which measures
banks` total loan portfolio to total deposit liability is
significant though positively related to NPLs at a 10
percent level of significance. From Table 4, a
percentage increase in the LDR of banks will bring
about 23.17 percentage increase in the NPL ratio.
This means that the more the loans that are generated
from the available total deposit liabilities, the higher
the NPL of Nigerian DMBs since interest paid on
deposits are usually higher than on equity funds.
Conclusively, the R-squared (0.826848) and
Adjusted R-squared (0.579488) show that the
explanatory variables jointly explain about 82.68
percent and 57.94 percent, respectively, of variations
in non-performing loans. The Prob. (F-statistic) of
0.012408 indicates a high significance level. The
Durbin-Watson statistic at 2.225091 suggests that the
model has no serial auto-correlation errors.
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1161
Volume 20, 2023
Table 4. Long-run Coefficients results
Variable
Coefficient
Std. error
t-statistic
Prob.*
UNEMP
-1.969847
3.102077
-0.635009
0.5357
GDPG
0.839515
0.302462
2.775610
0.0149**
INFL
0.112164
0.075325
1.489069
0.1586
EXRT
0.237621
0.070713
3.360366
0.0047***
LDR
0.231761
0.111049
2.087009
0.0557*
C
-42.97765
12.40838
-3.463598
0.0038***
R-squared
0.826848
Adjusted R-squared
0.579488
F-statistic
3.342691
Durbin-Watson stat
2.225091
Prob(F-statistic)
0.012408
Source: Authors’ computation with E-Views (2020). Note: *, **, ***, 1%, 5%, 10% significance level.
4.5 Error Correction Model (ECM) Results
Table 5 presents the short-run estimates and Error
Correction Term (ECT) of the model. In the short-
run result, the GDPG and EXRT have a significant
and positive relationship with NPLs. However,
UNEMP and INFL have an insignificant relationship
with NPLs. Looking at the coefficient, a percentage
increase in GDPG will increase NPLs by 0.84
percent and a percentage increase in EXRT will
increase NPL by 0.24 percent. The result further
presents the adjustment or error correction
mechanism, with a coefficient of -0.9639, which
indicates that about 96.39 percent of past errors are
corrected in the current period. Thus, there exists a
high speed of convergence from short-run to long-
run equilibrium conditions. Unemployment had an
insignificant negative relationship with NPLs while
inflation though positively related with NPLs is not
significant. Further insight from the short-run result
shows that lending rate accounted for a 0.23
percentage increase in NPLs at a 10 percent
significant level.
This points to the nature of the high interest rate
by deposit money banks in Nigeria. Despite the fact
that the high cost of capital has the potential of
discouraging investors from borrowing from banks, it
can also increase the tendency for default among
bank clients. It is thus, pertinent, that adequate
consideration is given to cost of capital and servicing
of the deposit money banks’ loans in a bid to make
bank loans more attractive which will invariably help
to reduce the rate of default. A high cost of capital
could also increase total overhead cost and the cost of
production of goods and services thereby introducing
inflationary pressure in the economy, and leading to
reduced demand for goods and services.
The direct relationship between economic growth,
exchange rate and NPLs in the short-run estimates
supports the long-run estimates. These results suggest
growth that is not inclusive enough and high
exchange that reflects over-dependence on foreign
goods and services. It is therefore important that
more emphasis should be placed on local content
development and encouragement of local and
international investors. This will veritably help to
transform the productive base of the economy for a
more inclusive growth experience and favourable
exchange as veritable macroeconomic tools for
controlling NPLs.
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1162
Volume 20, 2023
Table 5. Short-run estimates and Error Correction Term of the ARDL model
Variable
Coefficient
Std. error
t-statistic
Prob.
D(UNEMP)
-1.969847
3.102077
-0.635009
0.5357
D(UNEMP(-1))
-8.075697
2.641465
-3.057280
0.0085***
D(UNEMP(-2))
-5.722537
3.306237
-1.730831
0.1055
D(UNEMP(-3))
-7.268704
3.245119
-2.239888
0.0418**
D(GDPG)
0.839515
0.302462
2.775610
0.0149**
D(GDPG(-1))
0.824837
0.341609
2.414567
0.0300**
D(GDPG(-2))
0.707057
0.384533
1.838742
0.0873*
D(GDPG(-3))
0.625814
0.354092
1.767376
0.0989*
D(INFL)
0.112164
0.075325
1.489069
0.1586
D(EXRT)
0.237621
0.070713
3.360366
0.0047***
D(EXRT(-1))
0.172138
0.092428
1.862409
0.0837*
D(LDR)
0.231761
0.111049
2.087009
0.0557*
D(LDR(-1))
-0.233854
0.109688
-2.131993
0.0512*
D(LDR(-2))
-0.238117
0.094027
-2.532434
0.0239**
CointEq(-1)*
-0.963969
R-squared
0.824131
Adjusted R-squared
0.685287
Durbin-Watson stat
2.225091
Source: Authors’ computation with E-Views (2020). Note: *, **, *** 1%, 5%, 10% significance level.
This paper in essence has been able to investigate
the determinants of NPLs from two integrated
perspectives of macroeconomic and bank-specific
variables employing more recent observations under
a dynamic model estimation algorithm. Previous
alternative studies have been carried out with a major
focus on either of the two aspects while some others
have examined the effect of NPLs on the stability of
the financial institutions, [7], [17], [34]. In the
present study, a robust estimation procedure that
accounts for dynamic relationships within the system
has been incorporated through the autoregressive
distributed lag (ARDL) approach. In this process, the
long-run and short-run estimates and the systemic
mechanism in the adjustment process from the short-
run to the long-run stable state, have been
ascertained. The evidence from this robust analysis
showed a high level of convergence of the system
considering the significant contributory positive
effect of integrated macroeconomic and bank-
specific determinants factors of the NPLs
phenomenon.
5 Recommendations and Conclusion
This study, using ARDL model, investigates the
effect of macroeconomic and bank-specific variables
on non-performing loans in Nigerian Deposit Money
Banks using data set from 1991 to 2019. The method
enables us to assess the significant relationships
between the dependent variable (non-performing
loans) and the independent variables (unemployment
rate, GDP growth, inflation, exchange rate and loans
to deposits ratio) in the long and short-run. The study
establishes that the level of non-performing loans in
Nigeria is affected by the macroeconomic indicators,
which are the independent variables. As expected, the
variables are in line with our apriori expectation,
however, inflation is insignificant and appears not to
affect non-performing loans in Nigeria. Therefore,
the outcome of this study would help stakeholders to
mitigate the challenging effects of non-performing
loans in a developing economy like Nigeria. The
study recommends that the bank regulatory body
should ensure that the exchange rate of the Nigerian
naira is strengthened through the formulation of
policies that limit imports and boost exports. This is
because of EXRT’s positive and significant influence
on NPL. Also, the Deposit Money Banks should be
mandated to improve their drive on liability products
which will boost total deposits (especially the
demand deposits) while the total loan portfolio is
monitored thereby lowering the LDR since it
positively correlates with non-performing loans.
Similarly, a robust economy is important for
borrowers to be able to redeem their loan obligations.
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1163
Volume 20, 2023
Thus, the economic growth impact on non-
performing loans is positive and significant. This
indicates the existence of non-inclusive growth as an
increase in the GDPGR does not bring down the level
of NPL in the Nigerian banking sector. This can be
achieved by ensuring that loans are channeled to
more productive sectors of the economy.
In conclusion, this study suggests that future
studies can include more independent variables such
as corporate governance, credit growth and so on, in
order to see how non-performing loans respond to
other macroeconomic factors and performance
indicators in more developing and emerging markets.
Acknowledgement:
Covenant University, Ota, Nigeria sponsors the
publication of this research paper. It will contribute
to the body of existing knowledge. The authors are
deeply appreciative of this gesture and hereby
express our profound gratitude.
References:
[1] Berger, A. N., & DeYoung, R. (1997). Problem
loans and cost efficiency in commercial banks.
Journal of Banking and Finance, 21(6), 849–
870. Retrieved from https://doi.org/10.1016
/S0378-4266(97)00003-4.
[2] World Bank. (2021). World Development Index
Data Catalog.
[3] Keeton, W. R. (1999). Does faster loan growth
lead to higher loan losses? Economic Review,
Second Quarter, 57–75. Retrieved from
www.kc.frb.org.
[4] Hasanov, F., Bayramli, N., & Al-Musehel, N.
(2018). Bank-specific and macroeconomic
determinants of bank profitability: Evidence
from an oil-dependent economy. International
Journal of Financial Studies, 6(78), 1–21.
Retrieved from
https://doi.org/10.3390/ijfs6030078.
[5] Isaev, M., & Masih, M. (2017). Macroeconomic
and bank-specific determinants of different
categories of non-performing financing in
Islamic banks: Evidence from Malaysia. Munich
Personal RePEc Archive, MPRA Paper No.
79719, 1–24. Retrieved from
https://mpra.ub.uni-muenchen.de/79719.
[6] Kumar, V., & Kishore, P. (2019).
Macroeconomic and bank-specific determinants
of non-performing loans in UAE conventional
banks. Journal of Banking and Finance
Management, 2(1), 1–12.
[7] Akinlo, O., & Emmanuel, M. (2014).
Determinants of non-performing loans in
Nigeria. Accounting and Taxation Institute for
Business and Finance Research, 6(2), 21–28.
Retrieved from www.theIBFR.org.
[8] Adeola, O., & Ikpesu, F. (2017).
Macroeconomic determinants of non-
performing loans in Nigeria: An empirical
analysis. The Journal of Developing Areas,
51(2), 481-490. Retrieved from
https://doi.org/10.1353/jda.2017.0029.
[9] Staehr, K., & Uuskula, L. (2017). Forecasting
models for non-performing loans in the EU
countries. Eurosystem Working Paper Series 9,
1–32. Retrieved from
https://doi.org/10.23656/25045520/102017/
0149.
[10] Memdani, L. (2017). Macroeconomic and bank-
specific determinants of non-performing loans
(NPLs) in the Indian banking sector. Studies in
Business and Economics, 12(2), 125–135.
Retrieved from https://doi.org/10.1515/sbe-
2017-0026.
[11] Fendi, U. A., Sawalha, I. H. S., & Shamieh, J.
(2017). Early warning indicators for monitoring
non-performing loans in the Jordanian banking
system. International Journal of Business and
Social Science, 8(6), 104–114.
[12] Mehmood, A., Hidthiir, M. H. B., & Nor, A. M.
(2019). A conceptual paper for macroeconomic
determinants of non-performing loans. Asian
Journal of Multidisciplinary Studies, 7(3), 6–15.
[13] Macháček, M., Melecký, A., & Šulganová, M.
(2018). Macroeconomic drivers of non-
performing loans: A meta-regression analysis.
Prague Economic Papers, 27(3), 351–374.
Retrieved from
https://doi.org/10.18267/j.pep.656.
[14] Olokoyo, F. O., Fapetu, O., Ucheaga, E. G.,
Adu, O., & Oluwole, F. (2021). Impact of
external reserves shocks on selected
macroeconomic indicators in Nigeria:
Implication for sustainable economic growth.
African Journal of Business and Economic
Research, 2021. Retrieved from
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1164
Volume 20, 2023
https://doi.org/10.31920/1750-
4562/2021/SIn1a2.
[15] Merhbene, D. E., & Eugène-Rigot, C. (2021).
The relationship between non-performing loans,
banking system stability and economic activity:
The case of Tunisia. Graduate Institute of
International and Development Studies,
International Economics Department Working
Paper Series No. HEIDWP03-2021, 1-19.
[16] Clementina, K., & Isu, H. O. (2014). The rising
incidence of non-performing loans and the
nexus of economic performance in Nigeria: An
investigation. European Journal of Accounting
Auditing and Finance Research, 2(5), 89–76.
Retrieved from www.ea-journals.org.
[17] Atoi, N. v. (2019). Non-performing loan and its
effects on banking stability: Evidence from
national and international licensed banks in
Nigeria. Central Bank of Nigeria Journal of
Applied Statistics, 9(2), 43–74. Retrieved from
https://doi.org/10.33429/cjas.09218.3/6.
[18] Akerlof, G. A. (1970). The market for
“lemons”: Quality uncertainty and the market
mechanism. The Quarterly Journal of
Economics, 84(3), 488–500. Retrieved from
https://doi.org/10.2307/1879431.
[19] Rothschild, M., & Stiglitz, J. (1976).
Equilibrium in competitive insurance markets:
An essay on the economics of imperfect
information. Quarterly Journal of Economics
90(4), 90(4), 629–649. Retrieved from
https://doi.org/10.2307/1885326.
[20] Fell, J., Grodzicki, M., Krušec, D., Martin, R.,
& O’brien, E. (2017). Overcoming non-
performing loan market failures with transaction
platforms. Financial Stability Review, 2, 130–
144.
[21] Cincinelli, P., & Piatti, D. (2017). Non-
performing loans, moral hazard & supervisory
authority: The Italian banking system. Journal
of Financial Management, Markets and
Institutions, 5(1), 5–34. Retrieved from
https://doi.org/10.12831/87058.
[22] Bhattarai, B. P. (2018). Assessing banks internal
and macroeconomic factors as determinants of
non-performing loans: Evidence from Nepalese
commercial banks. International Journal of
Accounting & Finance Review, 3(1), 13–32.
Retrieved from https://doi.org/10.46281
/ijafr.v3i1.28.
[23] Das, J. K., & Dey, S. (2019). Factors
contributing to non-performing assets in India:
An empirical study. Review of Professional
Management- A Journal of New Delhi Institute
of Management, 16(2), 62–70. Retrieved from
https://doi.org/10.20968/rpm/2018/vl6/i2/14102
5
[24] Farooq, M. O., Elseoud, M. S. A., Turen, S., &
Abdulla, M. (2019). Causes of non-performing
loans: The experience of gulf cooperation
council countries. Entrepreneurship and
Sustainability Issues, 6(4), 1955–1974.
Retrieved from https://doi.org/10.9770/jesi
.2019.6.4(29).
[25] Adegboye, A., Ojeka, S., & Adegboye, K.
(2020). Corporate governance structure, bank
externalities and sensitivity of non-performing
loans in Nigeria. Cogent Economics and
Finance, 8(1), 1–21. Retrieved from
https://doi.org/10.1080/23322039.2020.1816611
.
[26] Pesaran, M. H., Shin, Y., & Smith, R. J. (2001).
Bounds testing approaches to the analysis of
level relationships. Journal of Applied
Econometrics, 16(3), 289–326. Retrieved from
https://doi.org/ 10.1002/jae.616.
[27] Johansen, S. (1995). Likelihood-based inference
in cointegrated vector autoregressive models.
OUP Oxford.
[28] Chandran, V. G. R., & Krishnan, G. (2009).
Foreign direct investment and manufacturing
growth: The Malaysian experience.
International Business Research, 1(3), 83–90.
Retrieved from
https://doi.org/10.5539/ibr.v1n3p83.
[29] Adediran, O., & Oduntan, E. (2017). Financial
development and inclusive growth in Nigeria: A
multivariate approach. Journal of Internet
Banking and Commerce, 22(8), 1-14. Retrieved
from http://www.icommerce central .com.
[30] Adediran, O. S., & Alege, P. O. (2020).
Autoregressive distributed lag approach to
external credit and economic growth in Nigeria.
Retrieved from https://doi.org/10.4018/978-1-
7998-1093-3.ch003
[31] Adusei, C. (2018). Determinants of non-
performing loans in the banking sector of Ghana
between 1998 and 2013. Asian Development
Policy Review, 6(3), 142-154. Retrieved from
https://doi.org/10.18488/journal.107.2018.63.14
2.154.
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1165
Volume 20, 2023
[32] Zheng, C., Bhowmik, P. K., & Sarker, N.
(2020). Industry-specific and macroeconomic
determinants of non-performing loans: A
comparative analysis of ARDL and VECM.
Sustainability (Switzerland), 12(1). Retrieved
from https://doi.org/10.3390/su12010325.
[33] Kumarasinghe, P. J. (2017). Determinants of
non-performing loans: evidence from Sri Lanka.
International Journal of Management
Excellence, 9(2), 1113-1121. Retrieved from
https://doi.org/10.17722/ijme.v9i2.367.
[34] EL-Maude, J.G, Abdul-Rahman A and Ibrahim,
M (2017). Determinants of non-performing
loans in Nigeria’s deposit money banks.
Archives of Business Research, 5(1) pgs. 74-88.
Retrieved from https://www.researchgate.
net/publication/313623561.
Contribution of Individual Authors to the
Creation of a Scientific Article (Ghostwriting
Policy)
The authors equally contributed in the present
research, at all stages from the formulation of the
problem to the final findings and solution.
Sources of Funding for Research Presented in a
Scientific Article or Scientific Article Itself
Covenant University, Ota, Nigeria sponsors
the publication of this research paper.
Conflict of Interest
The authors have no conflict of interest to declare.
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 BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.103
Michael Osunkoya, Ochei Ikpefan, Felicia Olokoyo
E-ISSN: 2224-2899
1166
Volume 20, 2023
