The Determinants of Non-Performing Loans in the Polish Banking
Sector – the Household Loans Portfolio
SŁAWOMIR I. BUKOWSKI
Department of Business and International Finance,
Kazimierz Pulaski University of Technology and Humanities in Radom
B. Chrobrego Street, 31, Radom 26-600
POLAND
ANETA M. KOSZTOWNIAK
Department of Monetary Economics,
SGH Warsaw School of Economics
al. Niepodległości 162, Warsaw 02-554
POLAND
Abstract: - The study aims to identify changes in non-performing household loans (NPLs) and their main
determinants in the Polish banking sector for the period 2009-2021. Specifically, we look at the main
determinants of creditworthiness of households which determine the possibility of repayment of principal
installments and interest within the prescribed period. The results of the VECM model confirm the considerable
significance of GDP per capita, gross salaries and lending rates to NPL loans of households. The results of the
response function show a positive impact of GDP per capita and lending rates on NPLs and a negative impact
of real salaries on NPLs. The decomposition of variance in the forecast period confirms an increased level of
explanation of NPL by GDP per capita, gross salaries, and the lending rates.
Key words: non-performing loans (NPLs), loan portfolio quality, debt, households, banking crises, VECM,
Poland
Received: May 20, 2021. Revised: December 21, 2021. Accepted: January 12, 2022. Published: January 14, 2022.
1 Introduction
Monitoring the quality of the portfolio, including
household loans, in the banking sector results from
prudential regulations, because NPLs can cause
monetary crises that can turn into financial crises
(Ghosh, 2015). According to Handley (2010) and
Ivanovic (2016), NPLs can be used as an indicator
of banking crises as they affect a country's economic
growth by reducing credit development. Low NPLs
indicate a strong monetary system in a country,
while high NPLs indicate a weak financial situation.
A growing level of non-performing loans in the
longer term will affect commercial banks first and
then the entire economy of a country (Souza and
Feijó[1]). In addition, rising NPLs are impacting
banking efficiency and causing banking crises
(Vouldis and Louzis [2]). NPLs will block interest
income, limit investment openings, and cause
liquidity crises in the financial system, resulting in
bankruptcy problems and economic slumps. For
these reasons, it is imperative to identify the factors
that influence NPLs to reduce NPL levels for
financial stability and economic goals (Stijepović,
[3]). An increasing liquidity risk at a bank may also
be an effect of NPLs (Quadahet. el. [4]).
The problem of non-performing loans for
economies is noticed by the European Commission
(EC) (Kasingeret al. [5], which has announced
strategies to combat non-performing loans. The first
plan was announced by the ECOFIN Council in July
2017. This plan was then extended with a new
package of measures in March 2018 and a capital
markets recovery package in July 2020. The
outbreak of the COVID-19 pandemic, like the
global financial crisis (2007), may negatively affect
household incomes and thus stimulate the growth of
non-performing loans. Therefore, it is important to
identify the main determinants of credit
deterioration. Bank outstanding loans to total gross
loans according to the World Bank (2021) show
significant differences in the banking sectors of EU
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
247
Volume 19, 2022
countries (including 36.4% Greece, 17.1% Cyprus,
6.6% Bulgaria, 6.2% Portugal, 3.8% Poland, and
1.1% Germany in 2019).
In a financial market such as in Poland (banking-
oriented), banks play a key role in the sustainability
of the banking system (Moradi et al., 2016).
Considering that the share of loans to households in
Poland accounts for 43% of total loans (NBP, 2021)
and their value is approx. 35% of GDP in 2021 (BIS
[6]), it is important to monitor them.
The aim of the study is to indicate changes in the
level of non-performing loans (NPLs) for
households and their main determinants in the
Polish banking sector in 2009-2021. The following
questions are formulated: 1) How have NPL rates
changed in the household loan portfolio in the
analyzed period. Did these changes indicate an
improvement or deterioration in the quality of the
loan portfolio in the Polish banking sector? 2)
Which of the explanatory variables had the strongest
impact on the changes in the NPL ratio and what
was their trend over time?
2 Problem Formulation and
Literature Review
2.1 The main problem
As the growth of NPLs has a negative impact not
only on the banking sector but on the entire
economy, a constant verification of the factors of
NPL change is necessary. In the case of large
economies, problems in a banking system may also
spread to neighboring countries as part of the
spillover effects. The main problem is to identify the
main factors influencing changes in NPLs and to
identify the strength and type (positive/negative)of
this influence. Considering that most authors
conduct research for a whole portfolio, they fail to
the structure of loan portfolios broken down into
loans to households and enterprises. This means that
when the structure of a loan portfolio changes, the
factors explaining NPLs may be different, or their
impact may change to a significant extent.
Therefore, these authors focus on a specific
portfolio of loans to households.
2.2 Literature Review
The National Bank of Poland (NBP) and other
institutions, e.g., the International Monetary Fund
(IMF), state that loans would be considered NPLs if
they do not produce interest and principal for a
minimum of 90 days. The NPL rate is calculated as
the ratio of non-performing loans (impaired loans)
and advances to the gross value of total loans and
advances (NBP [6]). The main reasons for high
NPLs are weak credit procedures, incompetent
credit specialists, high markup spreads, devefective
credit principles and lack of a borrower monitoring
policy . NPLs are a major indicator of credit risk
that affects the banking system of a country.
The rise in NPLs in the last decade has caught
the attention of many scholars around the world,
who try to explain this phenomenon. The
explanatory variables are mainly macroeconomic
and bank-related. Studies regarding the factors that
determine the level of NPLs show that a real
increase in GDP usually translates into higher levels
of income, improving the financial capacity of
borrowers (Marcinkowska et.al., [8], Kosztowniak
[9]). However, on the other hand, when the
economy is below normal conditions or in
a recession, the level of NPLs may increase due to a
resulting rise in unemployment, with borrowers
facing major difficulties to repay their debt (Salas
and Suarina, [10]; Fofack, [11]; Hada et. al. [12]).
Klein [13] investigates determining factors and
their impact on NPLs and on the macroeconomic
performance of Central, Eastern and South-Eastern
European (CESEE) countries for the period between
1998 and 2011, using the time series analysis. He
finds that NPLs respond to macroeconomic
conditions, such as unemployment, GDP growth,
and inflation and highlights that the high NPLs in
these countries affect the economic recovery
negatively. Moreover, NPLs study results for 10
transition countries (Central and Eastern Europe)
over the period 2006 and 2016 and use dynamic
panel estimations to show that GDP growth and
inflation are both negatively and significantly
correlated with the level of NPLs, while
unemployment is positively related to NPLs.
The export growth shows largely non-significant
results, indicating that NPLs within our sample were
mainly affected by domestic conditions rather than
external economic shocks (Mazreku et. al.[14]).
Exchange rates, interest rates and inflation are
other macroeconomic factors that impact the quality
of a bank’s activities. Exchange rate fluctuations
may have a negative impact on the quality of assets,
especially in countries with a large amount of
foreign currency loans. The same is true of interest
rate increases, particularly in the case of loans with
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
248
Volume 19, 2022
flexible interest ratea (Louzis et al., [15]; Zaman
and Meunier [16]). On the one hand, higher inflation
may ease debt compensation by affecting the real
value of unpaid credit, while, on the other hand, it
may also reduce the real income of unprotected
borrowers. In countries where credit rates are
flexible, higher inflation may lead to higher rates
resulting from monetary policy actions to fight
inflation (Nkusu, [17]). According to Salas [10],
banks in the EU have created permissive lending
conditions to attract customers. Low interest rates,
rising house prices and a stable economic
environment characterized the “precrisis” period.
Regarding the Greek banking market, (Louzis,
Vouldis and Metaxas) [2]) find that the real GDP
growth rate, unemployment rate and lending rates
have a strong negative effect on the NPL level,
interpreting them as a sign of poor banking
management.
The size of NPLs plays a key role in the stability
of the banking sector of a country. The factors that
explain the NPLs contain very important
information for banks. Among the banking variables
of NPLs, researches focus on return on assets
(ROA), efficiency of a bank and bank capital.
Godlewski [18] investigates the association between
NPLs and return on assets (ROA) and states that the
lower the rate of ROA, the higher the NPLs and vice
versa. Boudriga et al. [19] confirm from their study
that there is a negative association between ROA
and NPLs. Dimitrios et al. [20] investigate various
determinants of NPLs in the euro banking system
and conclude that ROA has a significant impact
upon NPLs. Rachman et al. [21] conclude that
highly profitable banks have lower NPLs due to
their better advancing activity and effective credit
supervision systems. Berger and DeYoung [22]
conclude that a decrease in the cost efficiency of
commercial banks in the United States would
contribute to increasing loan defaults in future.
Insufficient control of a loan portfolio increases risk
and NPLs. Fiordelisi et al. [23] examine the various
factors that increase the risk level in the EU banks
and conclude that a decreasing efficiency increases
the risk level of banks in future. Furthermore, the
efficiency and performance factors have an
influence on NPLs in the Greek banking sector
(Louzis et al., [2]). Rachman et al.[21] state that
operating efficiency does not influence NPLs.
The effect of bank capital on NPLs works in the
opposite direction. For one part, the incentivised
managers of low capitalized banks tend to get
involved in high-risk investments and give loans
that are issued without a proper credit rating and
monitoring (Keeton, 1999). On the other hand,
banks with a high level of capital tend to give loans
easily as they know that due to these loans banks are
not going to be bankrupt and fail; therefore, banks
are highly engaged with these kinds of risky credit
activities suggesting a positive association between
capital and NPLs (Rajan [24]). Moreover, the
capital adequacy ratio (CAR) shows the ability of an
organization to face abnormal losses and to survive
that situation. Makri et al. [25] also state that there
is a negative association between CAR and NPLs.
Constant and Ngomsi [26] claim that NPLs and
CAR are positively correlated.
Bank profitability and sustainability can only be
assured with a proper flow of interest income
generated through the lending function of banks.
However, a bank’s capital decreases together with
its health, which is becoming fragile, enhancing the
trend of NPLs. Therefore, banks are required to take
proactive action to deal with a bad choice of
borrowers by identifying and understanding the
macroeconomic factors that contribute to the rise of
classified credit in the banking system (Anjom and
Karim [27] and Dimitrios et. al. [28]). According to
Khuhair and Mardini [29]), NPLs also depend on
the effectiveness of asset management by bank staff
who can use the bank's IT infrastructure to assess
creditworthiness (Meshref [30]).
2.3 The changes of household loans and
NPLs
The value of household loans in the banking sector
in Poland showed a general upward trend in
Q1.2009-Q2.2021 (from PLN 398.2 million to PLN
768.4 million), especially in the period Q3.2012-
Q2.2021 (from 7.5% to 6.0%). This means that
NPLs for households were lower compared to the
NPLs for enterprises (7.0-8.0%) by around 2.0 p.p.
During the COVID-19 pandemic, the upward trend
in loans was maintained, except for temporary
declines in Q2.2020 and Q1.2021 (q/q). Maintaining
the demand for loans was a result of programs
implemented by commercial banks, i.e., solutions
aimed at facilitating the situation of borrowers in a
difficult financial situation caused by the COVID-19
epidemic (Fig. 1).
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
249
Volume 19, 2022
Fig. 1. Loans impaired and without impairment
of households and NPL ratio in Poland
For nearly a decade, NPLs have stabilized at the
level of 6.0%. Despite the pandemic, the rate of
NPLs is not rising. One of the reasons is a decline in
the average interest rate on commercial banks’ loans
to households from 5.1% in Q1.2020 to 3.6% in
Q2.2021 amid a lowering of the basic NBP rates
(Fig. 2).
Fig. 2 NPLs and average interest rate on loans to
households and non-profit institutions serving
households (%)
4. Results and Discussion
4.1. Data and Methodology
In the methodological approach used by the NBP
[6], household loans are available to: private
persons, individual entrepreneurs, individual
farmers, and non-commercial institutions operating
for the benefit of households. The article attempts to
assess the quality of the portfolio of loans granted to
households, therefore, respectively, impaired loans
and total loans granted to these households
(included in the so-called phase III, portfolio B) are
considered. The research is based on statistics from
the NBP, Central Statistics Office (CSO), and
Organisation for Economic Co-operation and
Development (OECD Internet databases) (Fig. 3).
Fig. 3. Time series of the model variables
The specificity of the base equation is developed
as a formula:
(1)
The explained variable: – The non-
performing loan ratio.
Explanatory variables:
– Gross domestic product per capita
– Average monthly real gross salary
(analogous period of the previous year = 100)
– Average interest rate on loans to
households and nonprofit institutions serving
households
– natural logarithm
– first differences
– random factor
– period
The model is constructed using the backward
rolling regression method. The methodology of
changes in the quality of the loan portfolio
corresponds to the methodologies used by central
banks, e.g., by the NBP and the IMF ([31],
Matthewes, Guo and Zhang [32], Maggi and Guida
[33], Mazreku et.al.[14]). The study period includes
quarterly data for the period Q1.2009–Q2.2021. The
methods used are known from literature on
international economics and international finance
and econometric methods like the VECM model
(Vector Error Correction Method) including the
impulse response functions. The expected influence
0
1
2
3
4
5
6
7
8
0
100 000
200 000
300 000
400 000
500 000
600 000
700 000
800 000
900 000
Q1-2009
Q1-2010
Q1-2011
Q1-2012
Q1-2013
Q1-2014
Q1-2015
Q1-2016
Q1-2017
Q1-2018
Q1-2019
Q1-2020
Q1-2021
%
PLN million
Loan impaired (PLN million)
Loan without impairment (PLN million)
Gross loans (PLN million)
NPL (%)
0
2
4
6
8
10
12
Q1-2009
Q1-2010
Q1-2011
Q1-2012
Q1-2013
Q1-2014
Q1-2015
Q1-2016
Q1-2017
Q1-2018
Q1-2019
Q1-2020
Q1-2021
Ratio of impaired loans to gross loans, NPL (%) (Rigt axis)
Average interest rate on loans to households and non-profit institutions
serving households (%) (Left axis)
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
250
Volume 19, 2022
of the explanatory variables on the explained
variable (NPLs) is presented in Table 1 (Annex).
The model data is verified on the basis of tests
for unit roots, e.g., Augmented Dickey-Fuller (ADF)
test, and cointegration is tested using the Engle
Granger test. The test results confirm the
applicability of the VECM model. The ADF test
confirms the unit-root null hypothesis: a=1, by
testing without constant the estimated value of (a-1):
-0.317433, test statistic: (4) = -4.38122 with a
critical value equal to -2.93 (with the significance
level of 5%), -3.58 (the significance level 1%) ,and
asymptotic p-value 0.05665, 1st-order
autocorrelation coefficient for e: 0.026.
The lag order for the VAR/VECM model is
based on an estimation of the following information
criteria: the Aikake information criterion (AIC),
Schwartz-Bayesian information criterion (BIC), and
Hannan-Quinn information criterion (HQC).
According to these criteria, the best, that is, minimal
values of the respective information criteria are:
AIC = 2, BIC = 1 and HQC = 1, with the maximum
lag order 4. Ultimately, the lag order 2 is accepted.
To analyze stability of the VAR model, a unit root
test is applied. The test indicates that in the analyzed
model equation roots in respect of the module are
lower than one, which means that the model is
stable and may be used for further analyses (Fig. 4).
Fig. 4. VAR inverse roots in relation to the unit
circle
Due to the occurrence of the unit element in all the
time series and the existence of cointegration among
the model variables, it is possible to extend and
transform the model into vector error correction
models (VECM).
4.2. VECM model and results
Co-integration is verified, justifying the use of the
VECM model for the lag order 2 and co-integration
of order 1. In accordance with the Granger
representation theorem, if variables and are
integrated to the order of I (1) and are co-integrated,
the relationship between them can be represented as
a vector error correction model (VECM)
(Piłatowska [34]).
The general form of the VECM can be written
as:
(2)
where:
(3)
and is a unit matrix.
The results of the VECM model confirm for
NPLs (the explained variable) the validity of
changes in the previous deviations of NPLs, GDP
per capita and relative salaries determining the
creditworthiness potential and interest rates on
loans. The EC1 index (containing the evaluation of
the error correction index) confirms that the
strongest correction of deviations from the long-
term equilibrium occurs in the case of the NPL
equation. Here, 38.66% of the imbalance from the
long-term growth path is corrected by the short-term
adjustment process. The significance of the other
variables is weaker, i.e., GDP per capita (0.15%),
real gross salaries (0.07%), and interest rate on loans
(0.04%). The sign of EC1 indicates that the increase
in GDPpc and real salaries contributes to a reduction
of NPLs (negative relation). The impact of changes
in NPLs from previous periods on changes in NPLs
in the new periods is also negative. There is
a positive relationship between the interest rates on
loans and NPLs. However, it should be kept in mind
that these rates during the system study period were
falling together with the NPLs . The results of the
determination coefficient (R2) indicate a moderately
good adjustment of the VECM model equations to
the empirical data. The results of the Durbin-Watson
(DW) test do not confirm the existence of
a significant residual autocorrelation (Table 2,
Annex).
The research results presented in the article are
consistent with the results of such authors as: Salas
and Suarina [10], Fofack [11], Mazreku et. al.[14].
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
251
Volume 19, 2022
4.3. Impulse response functions
The analysis of the NPL response to impulses from
the explanatory variables confirms the strength of
the influence of these impulses’ changes over time.
About 5-10 quarters of the forecast, the impact of
the explanatory variables on NPLs showed a
stabilization (constant). The results of the impulse
response function indicated that NPL showed
positive but declining responses to own NPL
impulses and negative responses to gross salaries.
This proves a weaker transfer of previous problems
with non-performing loans to subsequent loans and
indicates a positive impact of salaries growth on the
quality (service) of loans.
NPLs reacted with an increased force to impulses
from GDPpc and interest rate on loans to
households (AIRLH). The effect of GDPpc on NPLs
varied over time. At the beginning, as expected,
GDPpc contributed to a decline in NPLs. However,
the increase in GDPpc was conducive to an increase
in demand for loans (including mortgage for
the purchase of real estate). The NPLs showed
increasing trends in response to changes of AIRLH,
i.e., a stronger impact over 0-5th quarter and
stabilization after the 10th quarter (Fig. 5).
Fig. 5. Impulse response functions
4.4. The decomposition of variance
The decomposition of the NPL variance considering
the explanatory variables for 20 periods (quarters)
confirms the forecast of changes in the degree of
NPL explanation. The results of NPLs
decomposition indicate that in the 1st period, these
changes are fully accounted for with their own
forecast errors. In the 4th period (one year), their
own changes diminish (79.1%) and grow in
significance like GDPpc (0.5%), ARGSp (18.5%),
and AIRLH (1.8%). In the following periods, NPLs’
own changes fall in the 20th period (5 years) by
16.1%, while the degree of explanation on the part
of GDPpc rises to 4.6%, AIRLH to 13.3%, and
ARGSp to 65.9%.
The significant degree of explanation of NPLs by
changes in real salaries (to 66%) indicates the
importance of salaries (current incomes) for
servicing loans. Thus, the economic development
(salary growth) and the decline in the
unemployment rate are of key importance for the
quality of household loan portfolios . These results
are confirmed by the results of Mazrek [14], Salas
and Surina [10]. Interest rates on loans are also
gaining importance for future NPLs, which implies
the importance of interest rate policies. Baboučak
and Jančar [35] report similar results.
In the case of GDP per capita, the prevailing
degree of explanation remains on the part of own
(earlier) changes from the 1st quarter (96.1%) to the
20th quarter (83.8%). Nevertheless, in terms of the
explanation of GDP per capita, the share of non-
performing loans is gaining importance. Their share
has been growing since the first quarter (3.9% to
7.1%). Moreover, the degree of explanation of GDP
per capita by NPLs is higher than, for example, by
real salaries (from 0.0% to 2.9%) or interest rates on
loans (from 0.0% to 6.1%). These results confirm
the importance of lowering the NPL in the banking
system for maintaining GDP per capita. The results
are supported, among others, by Louzis, Vouldis
and Matexes [15] (Fig. 6).
Fig. 6 The decomposition of variance for NPL
and GDPpc
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
0.02
0.022
0.024
0 5 10 15 20
l_NPL -> l_NPL
-0.002
-0.001
0
0.001
0.002
0.003
0.004
0.005
0.006
0 5 10 15 20
l_GDPpc -> l_NPL
-0.02
-0.018
-0.016
-0.014
-0.012
-0.01
-0.008
-0.006
-0.004
-0.002
0
0 5 10 15 20
l_ARGSp -> l_NPL
0
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008
0.009
0 5 10 15 20
l_AIRLH -> l_NPL
-0.0034
-0.0033
-0.0032
-0.0031
-0.003
-0.0029
-0.0028
-0.0027
-0.0026
-0.0025
-0.0024
0 5 10 15 20
l_NPL -> l_GDPpc
0.008
0.009
0.01
0.011
0.012
0.013
0.014
0.015
0.016
0.017
0 5 10 15 20
l_GDPpc -> l_GDPpc
0
0.0002
0.0004
0.0006
0.0008
0.001
0.0012
0.0014
0.0016
0.0018
0.002
0 5 10 15 20
l_ARGSp -> l_GDPpc
-0.003
-0.0025
-0.002
-0.0015
-0.001
-0.0005
0
0 5 10 15 20
l_AIRLH -> l_GDPpc
-0.00212
-0.0021
-0.00208
-0.00206
-0.00204
-0.00202
-0.002
-0.00198
-0.00196
-0.00194
-0.00192
0 5 10 15 20
l_NPL -> l_ARGSp
0
0.0005
0.001
0.0015
0.002
0.0025
0.003
0 5 10 15 20
l_GDPpc -> l_ARGSp
0.007
0.0075
0.008
0.0085
0.009
0.0095
0.01
0 5 10 15 20
l_ARGSp -> l_ARGSp
-0.0014
-0.0012
-0.001
-0.0008
-0.0006
-0.0004
-0.0002
0
0 5 10 15 20
l_AIRLH -> l_ARGSp
0.0036
0.0037
0.0038
0.0039
0.004
0.0041
0.0042
0.0043
0 5 10 15 20
l_NPL -> l_AIRLH
0.03
0.035
0.04
0.045
0.05
0.055
0.06
0 5 10 15 20
l_GDPpc -> l_AIRLH
-0.015
-0.014
-0.013
-0.012
-0.011
-0.01
-0.009
-0.008
-0.007
-0.006
-0.005
0 5 10 15 20
l_ARGSp -> l_AIRLH
0.024
0.026
0.028
0.03
0.032
0.034
0.036
0.038
0.04
0 5 10 15 20
l_AIRLH -> l_AIRLH
0
10
20
30
40
50
60
70
80
90
100
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
l_NPL l_GDPpc l_ARGSp l_AIRLH
I_NPL
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
252
Volume 19, 2022
In summary, the results of the impulse response
function and the variance decomposition analysis
more clearly indicate that the changes in NPLs are
influenced by the current creditworthiness and its
changes throughout the entire loan period. This is
mainly evidenced by the growing importance of
relative salaries (ARGSp). Changes in interest rates
on loans are of a significant stabilizing importance
for NPLs. The low interest rate policy pursued by
the NBP for several years has had a positive effect
on the stabilization of interest rates on commercial
banks' loans.
5. Conclusion
Although most authors examine the determinants of
NPL for an entire loan portfolio, these authors have
focused on a specific loan portfolio, i.e., for
households, to get precise results. The reason for the
analysis of NPLs for householdd is their over 40%
share in the total loan portfolio in Poland.
According to the authors, each change in the
structure of the loan portfolio requires knowledge of
the NPL determinants, separately for household and
corporate loans. The NPL analysis for households
focuses on the variables determining the financial
situation and creditworthiness, e.g., the amount of
GDP per capita, real salaries and loan servicing
costs.
In line with the aim of the article, the results of
the analysis confirm a reduction in NPL rates for
household loans in Poland in 2009-2021. These
rates were on average by 2 pp. lower than for
corporate loans. The decline in NPLs, i.e., an
improvement in portfolio quality, was explicated
with the decline in interest rates on loans to
households. The results of the VECM and the level
of the EC1 ratio indicate that in the long run, NPLs
are most affected by deviations in earlier periods.
These results confirm a real continued deterioration
of creditworthiness in the subsequent periods. If
there is a loss of creditworthiness, then a longer time
is required to recover it. The sign of the EC1 factor
indicates a negative relation of GDPpc and real
income from NPLs. There is a positive relationship
between interest rates on loans and non-performing
loans.
The results of the response function indicate
negative NPL responses to impulses (earlier) NPL
fluctuations and real salaries. NPLs react positively
to the per capita GDP and loan interest rate impulse.
With an increase in GDP per capita, the demand for
loans may increase and thus the share of non-
performing loans expands. Increased loan
concentration rates, due to the higher costs of loan
servicing, contribute to an increase in NPL. The
results of the variance decomposition suggest a
significant degree of explanation of NPL by real
salaries. Their share grows to 66% in the forecast
period of 20 quarters. This result confirms the
validity of the current salary (income) for the
capability of servicing the loan. The NPL is also
explained by previous changes to the NPL, i.e., the
condition of non-performing loans (16.1%), the
interest rate on loans (13.3%) and, to a lesser extent,
GDP per capita (4.6%). This means the necessity to
constantly monitor non-performing loans and the
interest rate policy pursued by the NBP and
commercial banks. The decomposition of GDP per
capita also shows the importance of controlling
NPLs (up to 7.1%) and loan concentration rates
(6.2%). Thus, the impact of the quality of the loan
portfolio on the conditions of economic
development, measured with GDP per capita,
is confirmed.
To conclude, the income situation of borrowers,
depending on the pace of salary growth and the
decline in the unemployment rate in Poland, is of
general importance to lowering the NPLs (i.e.,
limiting the negative consequences for the
economy). The authors' contribution to the NPL
analysis is to draw attention to the importance of
changes in the structure of the loan portfolio and the
different determinants of NPLs for household loans
and, for example, for enterprises in individual
countries. The NPL reduction for household loans in
Poland relies on increasing relative salaries and the
ongoing restructuring of non-performing loans. The
importance of loan interest rates has been confirmed
0
10
20
30
40
50
60
70
80
90
100
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
l_NPL l_GDPpc l_ARGSp l_AIRLH
I_GDPpc
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
253
Volume 19, 2022
for NPLs, but their degree of explanation for NPL
changes is nearly 5 times weaker than for real
salries. Thus, our research results broaden the
knowledge about the quality of the loan portfolio
and have significant implications for bank
efficiency, market stability and economic growth in
the economy, and the role of macroeconomic policy.
References:
[1] G.J.D.G Souza, C.AFeijó, Credit risk and
macroeconomic interactions: empirical
evidence from the Brazilian banking system,
Modern Economy, Vol.2, No. 5, 2011, pp. 910-
929, doi: 10.4236/me.2011.25102.
[2] A.T. Vouldis, D. P. Louzis, Leading indicators
of non-performing loans in Greece: the
information content of macro-, micro- and
bank-specific variables,Empirical
Economics,Springer Berlin Heidelberg,
Vol.54,No.3, 2018, pp.1187-1214,
doi:10.1007/s00181-017-1247-0.
[3] R. Stijepović, Recovery and reduction of non-
performing loans – podgorica approach,
Journal of Central Banking Theory and
Practice, Vol. 3, No. 3, 2014, pp.101-118, doi:
10.2478/jcbtp-2014-0017.
[4] H. A. Qudah, K. K. Abdo, L.A. Al_Qudah, O.
Kilani, M. Almanaseh, M. Alqudah, Liquidity
Risk Measurement Study Case (Jordan Islamic
Banks), International Journal of Economics
and Management Systems, No. 6, 2021, pp.
495-503.
[5] J. Kasinger, J. P. Krahnen, S. Ongena, L.
Pelizzon, M. Schmeling, M. Wahrenburg, Non-
Performing Loans - New Risks and Policies?
NPL Resolution After COVID-19: Main
Differences to Previous Crises, European
Parliament, Economic Governance Support
Unit (EGOV) Directorate-General for Internal
Policies PE 651.387, March, 2021, Retrieved
from: https://www.europarl. europa.eu/ Reg
Data/etudes/STUD/2021/ 651387/IPOL_STU
(2021)651387_EN.pdf.
[6] NBP, Assets and Liabilities of Monetary
Financial Institutions, 2021, Retrieved from:
https://www.nbp.pl/homen.aspx?f=/en/statysty
ka/zobowiazania.html.
[7] BIS 2021. Statistics Explorer. Retrieved from:
http://stats.bis.org:8089/statx/srs/table/
f3.1?f=xlsx.
[8] M. Marcinkowska, P. Wdowiński, S. Flejterski,
S. Bukowski, M. Zygierewicz, Wpływ
Regulacji Sektora Bankowego Na Wzrost
Gospodarczy – Wnioski Dla Polski (The
impact of banking sectorregulation on
economicgrowth–conclusions for Poland),
Materiały i Studia, NBP, No. 35, Warsaw,
2014, Retrived from: https://www.nbp.pl/
publikacje/ materialy_i_studia/ms305.pdf.
[9] A. Kosztowniak, Changes in The Quality of
The Loan Portfolio in The Polish Banking
System, Journal of Economics, Business &
Organization Research, CEBOR Academy
Ltd., Italy, 2020, Retrieved from:
https://dergipark.org.tr/en/download/article-
file/1470997.
[10] V. Salas, J. Saurina, Credit Risk in Two
Institutional Regimes: Spanish Commercial and
Savings Banks. Journal of Financial Services
Research, No. 22, 2002, pp. 203–224.
Available online:
https://link.springer.com/article/10.1023/A:101
9781109676 (accessed on 17 May 2020).
[11] H. Fofack, Non-performing Loans in Sub-
Saharan Africa: Causal Analysis and
Macroeconomic Implications. World Bank
Policy Research Working Paper, No. 3769,
2005.
[12] T. Hada, N. Bărbuță-Mișu, I.C. Iuga, D.
Wainberg, Macroeconomic Determinants of
Nonperforming Loans of Romanian Banks,
Sustainability, 2020, No.12, 7533.
https://doi.org/10.3390/su12187533
[13] N. Klein, Non-Performing Loans in CESEE:
Determinants and Impact on Macroeconomic
Performance, IMF Working Paper European
Department, No. 13/72, 2013.
[14] I. Mazreku, F. Morina, V. Misiri, J. V. Spiteri,
S. Grima, Determinants of the Level of Non-
Performing Loans in Commercial Banks of
Transition Countries, European Research
Studies Journal, Vol. XXI, Issue 3, 2018, pp. 3-
13.
[15] D. P. Louzis, A. T. Vouldis, V. L. Metaxas,
V.L., Macroeconomic and Bank-Specific
Determinants of Non-performing Loans in
Greece: A Comparative Study of Mortgage,
Business and Consumer Loan Portfolios.
Elsevier, Journal of Banking and Finance, No.
36(4), 2012, pp. 1012-1027.
[16] C. Zaman, B. Meunier, A Decade of EU
Membership: Evolution of Competitiveness in
Romania. European Research Studies Journal,
No. 20(2A), 2017, pp. 224-236.
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
254
Volume 19, 2022
[17] M. Nkusu, Nonperforming Loans and Macro
Financial Vulnerabilities in Advanced
Economies. IMF Working Papers, 161, 2011.
[18] C. J. Godlewski, Bank Capital and Credit Risk
Taking in Emerging Market Economies,
Journal of Banking Regulation, Vol. 6, No. 2,
2008, pp. 128-145, doi:
10.1057/palgrave.jbr.2340187.
[19] A. Boudriga, N.B. Taktak, S. Jellouli, Bank
Specific, Business and Institutional
Environment Determinants of Banks
Nonperforming Loans: Evidence from MENA
Countries, Economic Research Forum,
Working Paper, 2010, pp. 1-28.
[20] A. Dimitrios, L. Helen, T. Mike,Determinants
of Non-Performing Loans: Evidence from
Euro-Area Countries, Finance Research
Letters, Elsevier, Vol. 18, 2016, pp. 116-119,
doi: 10.1016/j.frl.2016.04.008.
[21] R. A. Rachman, Y.B.Kadarusman,
K.Anggriono, R.Setiadi,Bank-Specific Factors
Affecting Non-Performing Loans in Developing
Countries: Case Study of Indonesia, The
Journal of Asian Finance, Economics and
Business (JAFEB), Vol. 5, No. 2, 2018, pp. 35-
42.
[22] A.N. Berger, R. de DeYoung,Problem Loans
and Cost Efficiency, Journal of Banking and
Finance, Vol. 21, 1997, pp.1-28.
[23] F. Fiordelisi, D. Marques-Ibanez, P. Molyneux,
Efficiency and Risk in European Banking,
Journal of Banking and Finance, Elsevier, Vol.
35, No. 5, 2011, pp. 1315-1326.
[24] R. Rajan, Why bank credit policies fluctuate,
The Quarterly Journal of Economics, Vol. 2,
No. 109, 1994, pp. 399-441.
[25] V. Makri, A. Tsagkanos, A. Bellas,
Determinants of Non-Performing Loans: The
Case of Eurozone, Panoeconomicus, Vol. 61,
No. 2, 2014, pp. 193-206.
[26] F.D. Constant, A. Ngomsi, Determinants of
Bank Long-Term Lending Behavior in The
Central African Economic and Monetary
Community (CEMAC), Review of Economics
and Finance, Better Advances Press, Canada,
Vol. 2, 2012, pp. 107-114.
[27] W. Anjom, A.M. Karim, Relationship Between
Non-Performing Loans and Macroeconomic
Factors (With Specific Factors: A Case Study
on Loan Portfolios- SAARC Countries
Prospective. Asia Pacific Journals of Finance
and Risk Management. No.15(3), 2016, pp. 84-
103.
[28] A. Dimitrios, L. Helen, T. Mike,Determinants
of Non-Performing Loans: Evidence from
Euro-Area Countries, Finance Research
Letters, Elsevier, Vol. 18, 2016, pp. 116-119,
doi: 10.1016/j.frl.2016.04.008.
[29] H. Y. Khudhair, A. Mardani, The Major Issues
Facing Staff in Islamic Banking Industry,
International Journal of Economics and
Management Systems, No. 6, 2021, pp. 200-
211.
[30] H. Meshref, Predicting Loan Approval of Bank
Direct Marketing Data Using Ensemble
Machine Learning Algorithms, International
Journal of Circuits, Systems and Signal
Processing, Vol. 14, 2020, pp. 914-922.
[31] IMF, Financial Soundness Indicators –
Background Paper, Prepared by the Staff of
the Monetary and Financial Systems and
Statistics Departments. Approved by C. S.
Carson, S. Ingves, May 13, 2003, p. 49.
[32] K. Matthewes, J. Guo, N. Zhang, Rational
inefficiency and non-performing loans in
Chinese banking: a nonparametric
bootstrapping approach, Cardiff Economic
Working Papers, E2007/5, Cardiff University,
U.K, 2007.
[33] B. Maggi, M. Guida, Modelling non-
performing loans probability in the commercial
banking system: efficiency and effectiveness
related to credit risk in Italy, Empirical
Economics, No. 41, 2011, pp. 269-291.
[34] M. Piłatowska, Modelowanie Niestacjonarnych
Procesów Ekonomicznych.
StudiumMetodologiczne, (Modelling of Non-
Stationary Economic Processes. A
Methodological Study), Uniwersytet M.
Kopernika, Toruń, 2003.
[35] I. Babouček, M. Jančar, Effects of
Macroeconomic Shocks to the Quality of the
Aggregate Loan Portfolio; Working Paper
Series, No. 1; Czech National Bank: Praha,
Czech, 2005.
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
255
Volume 19, 2022
Contribution of individual authors to
the creation of a scientific article
(ghostwriting policy)
Sławomir I. Bukowski was responsible for the
econometric analysis.
Aneta M. Kosztowniak carried out the concepts,
review of literature and research, preparation of
statistical data, interpretation of research results and
figures, tables and full paper preparation.
Sources of funding for research
presented in a scientific article or
scientific article itself
The funding for the article was sourced from the
Kazimierz Pulaski University of Technology and
Humanities in Radom.
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/de
ed.en_US
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
256
Volume 19, 2022
Annex
Table 1. Model variables and expected impact on the NPLs
No.
Variables
Data source
Impact
1
NPB
“˗ “
2
OECD
“˗ “
3
CSO
“˗ “
4
NBP
“+ “
Table 2.The VECM model
VECM system, lag order 2, observations 2009:4-2021:2 (T = 47)
Cointegration rank = 1, Case 3: Unrestricted constant
(cointegrating vectors, standard errors in parenthes)
(adjustment vectors)
1_NPL
1
0
1_NPL
-
0.38660
1_GDPpc
0.59541
0.09616
1_GDPpc
-0.001559
l_ARGSp
2.26960
0.63079
l_ARGSp
-0.000708
d_AIRLH
-9.91650
3.08760
d_AIRLH
0.0043201
Equation 1: d_l_NPL
Coefficient
Std. Error
t-ratio
p-value
const
5.364580
0.741312
7.237
<0.0001**
*
d_l_NPL_1
−0.211794
0.115883
−1.828
0.0753*
d_l_GDPpc_1
0.116180
0.260513
0.446
0.6581
d_l_ARGSp_1
0.671921
0.357612
1.879
0.0677*
d_d_AIRLH_1
−3.582050
1.631590
−2.195
0.0341**
EC1
−0.386595
0.053413
−7.238
<0.0001**
*
Mean dependent var
0.002563
S.D. dependent var
0.037236
Sum squared resid
0.024165
S.E. of regression
0.024892
R2
0.603911
Adjusted R2
0.553131
rho
−0.098803
DW
2.196533
Equation 2: d_l_GDPpc
Coefficient
Std. Error
t-ratio
p-value
const
0.032466
0.569975
0.0569
0.9549
d_l_NPL_1
−0.015615
0.089099
−0.1753
0.8618
d_l_GDPpc_1
−0.420778
0.200302
−2.1010
0.0422**
d_l_ARGSp_1
0.128208
0.274959
0.4663
0.6436
d_d_AIRLH_1
−0.178242
1.254490
−0.1421
0.8877
EC1
−0.0015586
0.041068
−0.0379
0.9699
Mean dependent var
0.007534
S.D. dependent var
0.019791
Sum squared resid
0.014285
S.E. of regression
0.019139
R2
0.171145
Adjusted R2
0.064881
rho
−0.022234
DW
2.013226
Equation 3: d_l_ARGSp
Coefficient
Std. Error
t-ratio
p-value
const
0.009756
0.342660
0.0285
0.9774
d_l_NPL_1
−0.004598
0.053565
−0.086
0.9320
d_l_GDPpc_1
0.021279
0.120418
0.177
0.8607
d_l_ARGSp_1
−0.279071
0.165301
−1.688
0.0993*
d_d_AIRLH_1
−0.589286
0.754179
−0.781
0.4393
EC1
−0.000708
0.024689
−0.029
0.9773
Mean dependent var
0.000044
S.D. dependent var
0.011424
Sum squared resid
0.005163
S.E. of regression
0.011506
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
257
Volume 19, 2022
R2
0.100925
Adjusted R2
-0.014341
rho
−0.037395
DW
2.070443
Equation 4: d_d_AIRLH
Coefficient
Std. Error
t-ratio
p-value
const
−0.0600174
0.0923611
−0.6498
0.5196
d_l_NPL_1
0.0002609
0.0144381
0.0181
0.9857
d_l_GDPpc_1
0.0106305
0.0324577
0.3275
0.7450
d_l_ARGSp_1
−0.0183039
0.0445554
−0.4108
0.6835
d_d_AIRLH_1
−0.1975700
0.2032820
−0.9719
0.3371
EC1
0.0043201
0.00665476
0.6492
0.5200
Mean dependent var
−0.000011
S.D. dependent var
0.002994
Sum squared resid
0.000375
S.E. of regression
0.003101
R2
0.048738
Adjusted R2
-0.073219
rho
−0.031491
DW
2.019588
Nonte: α = 0.01 (***), where α = 0.05 (**), α = 0.10 (*).
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2022.19.23
Sławomir I. Bukowski, Aneta M. Kosztowniak
E-ISSN: 2224-2899
258
Volume 19, 2022
