INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2529
www.rsisinternational.org
Unveiling the Relationship: Earnings Quality and Income Protection
Performance through the Hidden Markov Model in Economic
Growth Perspectives
Rosita Husain
1*
, Wan Nurul Fatimah Wan Mohamad Nawi
1
, Hanissah Hamzah
1
, Nur Zafirah A
Khadar
1
, Nik Nor Amalina Nik Mohd Sukrri
2
1
Fakulti of Business and Accountancy, Universiti Selangor, 40000 Shah Alam, Selangor, Malaysia
2
Faculty of Business, Economics and Social Development, Universiti Malaysia Terengganu, 21030,
Terengganu, Malaysia
*Corresponding author
DOI: https://dx.doi.org/10.47772/IJRISS.2025.91100203
Received: 10 November 2025; Accepted: 20 November 2025; Published: 04 December 2025
ABSTRACT
Over the past three decades, especially since the recent financial crisis, the volume of academic research
analysing banks' efficiency and productivity has grown substantially. The risk-taking behaviour of banks used
to be of relatively low interest; however, in recent years, this interest has exploded, especially in regards to its
correlation with monetary policy. Measuring credit risk has been and will continue to be a top priority in the
banking industry, and the recent global economic crisis has had a direct impact on this process. In this research,
the authors of a hidden Markov model-based earnings quality measurement propose and verify its usefulness.
Earnings fidelity is a measure of how closely reported earnings correspond to the company's true financial
health. The percentage represents the level of accuracy achieved. The impact of earnings management on
financial institutions' bottom lines is the focus of this research paper. Within the framework of a Bayesian
hierarchical model that allows for cross-sectional heterogeneity, we employ a Markov chain Monte Carlo
procedure to estimate the variable in question. The forward earnings response coefficient is positively
correlated with earnings fidelity, suggesting a positive relationship between the two.
Keywords: Earnings management, bank performance, risk-taking, corporate governance, Markov model.
INTRODUCTION
Corporate governance research focuses on the factors that influence a company's success. An important focus
of this body of work is the role of financial management in organisational performance. Financial statements
for businesses include information on a company's earnings, also known as the "bottom line" or "net income"
(Akram et al., 2015). It is widely accepted that earnings are a good indicator of a company's financial health as
well as its profitability and market strength. Profitability can be measured in a variety of ways by both external
and internal users. Managers' pay plans, debt covenants, investors, and creditors, to name a few, all rely on
earnings data (Tabassum et al., 2015; Dechow, 1994). Earnings management through manipulation of
accounting earnings has been studied in accounting and finance literature. Based on study of Ljubisavljević
and Jakobsson (2022), the pandemic caused by the COVID-19 virus has resulted in a number of financial
difficulties. In times of financial distress, companies are typically incentivized to use earnings management
techniques so that they can alter the picture that they have of their current financial situation. This is because
market-based pressure encourages companies to do so.
However, studies have shown that institutional and macroeconomic factors may be more influential during
times of crisis. This is due to increased scrutiny and acceptance of abnormal results, both of which may make
such practises less attractive.
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2530
www.rsisinternational.org
Earnings management is the tools or processes used by insiders, within the constraints of Generally Accepted
Accounting Principles (GAAP), to manipulate earnings or change the reported number against the interests of
outsiders (Cimini, 2015; Healy and Wahlen, 1999; Schipper, 1989). Companies, on the other hand, only use
earnings management when it is an option (Teoh et al., 1998). Earnings management's primary goal is to
smooth out the flow of money. In order to achieve this goal, businesses can use a variety of methods. On the
one hand, managers attempt to reduce reported earnings when a company's previous earnings were high. As a
result, managers try to boost reported earnings if previous earnings are low (Tabassum et al., 2015; Zang,
2011). In addition, executives use earnings management to maintain a high level of performance and to achieve
personal contractual goals linked to the reported accounting figures (El Sood, 2012; Goulart, 2007).
It is possible for executives to inflate or underflate the reported income to demonstrate the performance of the
company and obtain higher compensation tied to the firm's expected future performance (eg. firm performance
and earning management; Mostafa and Ibrahim, 2019) . As a result, the way managers implement earnings
management practise can have either a positive or negative impact on the performance of the company
(Bornemann et al., 2012). Earnings reports are impacted by several factors, including the fundamental
performance of a company as well as the measurement process, which is governed by reporting standards,
auditing technology, and managerial discretion (Dechow et al., 2000; Nikolaev, 2017). In previous studies, the
level of earnings quality was evaluated based on statistical properties of earnings (such as smoothness,
earnings kinks, and target beating) as well as regression-based abnormal accrual models (Dechow et al., 2010).
Researchers use models that are based on regression to differentiate "abnormal" accruals from "normal"
accruals based on fundamental performance, and they base the definition of accruals quality on abnormal
accruals (Dechow and Dichev, 2002). Recent studies have shown, however, that using proxies for
unobservable "true" earnings may lead to confusion between performance shocks and reporting discretion
(Guay et al., 1996; Ball, 2013; Dichev et al., 2013; Owens et al., 2017).
An important influence on a company's profitability and performance is earned management practise.
Financial performance is the primary source of information for investors and other stakeholders, so an
examination of the link between earnings management and performance is critical. In order to prevent future
scandals, the demand for research into the causes of accounting figure manipulation and their effects on firm
stability has increased since the last financial crisis (Ardekani et al., 2012). Indeed, since the last financial
crisis, earnings management has received a growing amount of attention from scholars (Filipa and
Raffournierb, 2014; Cimini, 2015; Debnath, 2017).
To mitigate the danger of using unreliable proxies, our approach uses a statistical model called the Markov
Model to decouple the two concepts of accounting quality and the process of actual earnings. Each company is
thought to undergo state transitions according to a Markov process, at least in our model. The company's
underlying performance is represented by the state, which can be either Low (L) or High (H). The status of the
company is hidden from view, but at the end of each period (h), it announces whether its earnings were low (l)
or high (h). Inferring a company's unobservable state from its earnings signals is possible because the
likelihood that it will issue a particular signal depends on the state. As a result, the probabilities that a firm will
stay in its current state or switch to the other state and the probabilities that it will emit a given earnings signal
are uniquely associated with each firm in each period. Based on the design of this HMM, we can make an
educated guess as to how well earnings signals reflect the true health of the business. Earning Fidelity is the
name given to this metric.
A consensus has not yet emerged regarding the optimal earnings quality metric (Dechow et al., 2010). New to
empirical research that proposes accounting quality measures based on distributional properties of earnings
(e.g., smoothness) and reduced-form regression models is the Bayesian hierarchical method (e.g., accruals
quality). Our structurally estimated earnings quality measure is theoretically motivated and provides at least
two advantages over existing quality measures. Initially, the Bayesian hierarchical method considers
firmspecific parameters to be random variables drawn from a distribution. As a result, the method combines
firmspecific data with essential prior information (i.e., the distribution of all firms' parameter values).
Secondly, this method does not restrict connections between companies. Companies in the same industry or
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2531
www.rsisinternational.org
published in the same year, for instance, need not be connected in a particular way. This hypothesis has the
potential to minimise measurement error.
Different conclusions can be drawn from theoretical and empirical studies of accounting quality. In contrast to
the static models of financial reporting that have been the focus of the analytical literature (Christensen and
Feltham, 2003), the empirical literature has proposed quality measures based on accounting data from multiple
periods. The hidden Markov model allows us to make our reporting systems dynamic rather than static. By
providing an estimate of the reporting system's properties in a multiperiod setting, the theory of earnings
quality is brought closer to practise. Accounting structural estimations typically ignore cross-firm
heterogeneity (Bertomeu et al., 2018; Zakolyukina, 2019; Du, 2019), but this is something that can be fixed by
employing Bayesian hierarchical modelling. Using Bayesian hierarchical modelling and Markov chain Monte
Carlo (MCMC), respectively, Bernhardt et al. (2016) and Zhou (2021) estimate the impact of investor learning
on cross-sectional variations in the manager's voluntary disclosure and the stickiness of analyst
recommendations in a cross-section of analysts. For the first time, we use a firm- and time-dependent Bayesian
method to estimate earnings quality in our study.
2. The Method and Forecasting Model
2.1 Data
Extending previous empirical studies, this paper examines the impact of earnings management on bank
profitability. Financial institutions play an important role in maintaining the integrity of the global financial
system (Cheng et al., 2011). Aside from that, banks were seen as an important player during the most recent
financial crisis (Faulkender et al., 2010; Claessens and Horen, 2015). These findings show how important it is
to look into the connection between earnings management and bank performance. Islamic banks that were
active between 2000 and 2021 are included in the study's initial sample. In the end, we have a final sample of
150 bankyear observations representing 15 Malaysian Islamic Banks after restricting all banks with the data
needed to analyse our primary variables of interest. The Orbis database is where we get the numbers and dates
for loan loss provisions. Any information that was missing was gleaned from the annual financial reports. The
data analysis is using the R project for statistical computing software version 4.3.0.
2.2. Earnings Management in Markov Model Forecasting
The measure of the earnings' quality that we conceptualise as Earnings Fidelity is the degree to which a
company's earnings signal reveals the underlying state of the company, where "state" refers to the economic
conditions that have an impact on the company's fundamental performance. The recipients' confidence in their
inferences about the state of the company is boosted when the earnings signal contains more accurate
information. We need a theoretical model in which earnings serve as a noisy indicator of the firm's underlying
condition so that we can formalise earnings fidelity. This will allow us to measure earnings more accurately.
For the purpose of this function, we make use of a highly stylized model.
Numerous empirical studies on earnings management can be found in accounting and financial literature.
Different measurements, such as discretionary accruals, smoothness, timeliness, loss avoidance, and investor
responsiveness, are utilised as proxies for earnings management in these empirical studies (Dechow et al.,
2010). For the banking sector, discretionary loan loss provision (DLLP) is one of the most frequently used
proxies for measuring earnings management practise (Burstahler and Dichev, 1997; Anandarajan et al., 2003;
El Sood, 2012). On the other hand, Hegde and Kozlowski (2021) find that the following year's earnings and
loan growth are significantly higher for banks that report higher provisions during prosperous times, while
non-performing loans continue to rise for the same banks during recessions. The economic downturn of 2008
had no bearing on these numbers. Investors were already skeptical of DLLP prior to the financial crisis.
Regulators need to make sure that bank loss reserves continue to be a reliable source of information even as
new accounting standards call for more subjective judgment. Before calculating DLLP, loan loss provision
must be separated into discretionary and non-discretionary components. To estimate the non-discretionary
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2532
www.rsisinternational.org
components of loan loss provision, we follow prior research (Beaver and Engel, 1996; Kanagaretnam et al.,
2004; Zoubi et al., 2007; Cheng et al., 2011; Ben Othman and Mersni, 2014) and use the following model.
LLP
it
= +
01
NPL
it−1
+
2
NPL
it
+
3
TL
i t,
+ ........................................(1)
Where:
LLP
it
= Total provision for loan losses for bank i at year t, deflated by beginning loans.
NPL
it−1
= The beginning balance of non-performing loans for bank i at the end of year t, adjusted for inflation.
NPL
it
= Variation in the value of non-performing loans for bank i at year t, deflated by the value of beginning
loans.
TL
it
= Change in total loan value for bank i at year t, deflated by initial loan balances.
The discretionary loan loss provision, also known as DLLP
it
it is computed by using the residual from
Model 1. According to studies that were conducted in the past (such as the ones conducted by Kanagaretnam et
al. (2004) and Ben Othman and Mersni (2014), the loan loss provision has a positive correlation with the levels
of non-performing loans, beginning balances of non-performing loans, and total loans. This is because the
probability of a loan going into default increases when a bank extends more credit, which forces the bank to
increase the amount of money it sets aside to cover bad loans. From the standpoint of earnings quality, banks
have incentives to reduce earnings volatility by decreasing earnings in years with an unexpectedly strong
performance and increasing earnings in years with a poor performance. A more consistent flow of earnings
could help reduce the information asymmetry between managers and outside investors (Beatty et al., 2002;
Beatty & Harris, 1999; Liu & Ryan, 2006; Beatty et al., 2002; Beatty & Harris, 1999). There is evidence in
many previous studies that managers smooth earnings through LLP and recognise security gains and losses.
Consequently, these variables must be considered when evaluating the quality of earnings.
There are many different approaches that can be taken to calculate efficiency indicators that consider the risk-
taking behaviour of financial institutions like banks. According to the findings of previous studies, the amount
of nonperforming loans can be factored into the production function of a bank as a form of negative output (or,
in terms of the profit function, as an expense that reduces total profits). If either the interest payments or the
principal payments on a loan are more than ninety days past due, the loan must be categorised as
nonperforming in accordance with Spanish accounting standards. In addition, regardless of the status of the
borrower's payments, all loans that have been granted to defaulting borrowers are regarded as nonperforming
loans.
In accordance with Spanish banking regulations, bank managers estimate LLP based on a stringent set of rules
established by the Bank of Spain, which are heavily dependent on the length of time payments are past due.
However, Bank of Spain regulations determine the minimum losses a bank must recognise once a loan has
been deemed nonperforming, leaving banks with a great deal of leeway. To mitigate the effects of the potential
manipulation of LLP, we use anticipated LLPs as expenditures rather than realised LLPs. This reveals whether
banks' LLP decisions to manage earnings or capital (and circumvent strict accounting rules by over- or
underprovisioning assets or misclassifying them) are effective. According to Pérez et al. (2008), if they were
successful, stringent regulations on LLP might be unnecessary, and "there is merit in having more
principlesbased accounting standards" (Pérez et al., 2008).
After estimating the extent of earnings manipulation in the Spanish banking system, we estimate a nonconvex
model of the short-run profit frontier. This model follows Fare et al. (1994) by classifying the inputs as variable
(xv) and fixed (xf) based on the original variables (in the case of a bad output, only considering the realised
LLPs) Primont, (1993), for a short run cost frontier definition). We will therefore model variable profit
maximisation.
manip
M V O
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2533
www.rsisinternational.org
(r
jm
,P
jv
,P
jo
) = max( ,z u
m
,x npl
v
,
o
) r u
jmm
− p x
jv u
− p npl
jo o
) ...........................
...(2)
m=1 v=1 o=1
Where:
j
z u
j jm
u
m
,m =1,..........,M,
j=1
j
z x
j jv
x v
v
, =1,..........,V,
j=1
j
z x
j jf
x
jf
, f =1,..........,F,
j=1
j
z npl
j jo
npl o
o
, =1,..........,O,
j=1
j
z
j
=1,..........,M,
j=1
z
j
= 0,1 where r
jm
M
+
is the vector of output prices for bank j, r
jm
0, and we also have variable inputs
(netputs) with prices p
jv
V
+
,v =1,........,V . Analogouslyu
j
M
+
is the vector of output quantities for j x,
jv
V
+
are the variable net puts for bank j and x
jf
F
+
are the fixed netputs for the same bank. However, compared
with the contributions of Färe et al. (1994) and Primont (1993), we are considering here the role of risk via
LLPs. Therefore, we have npl
j
O
+
is the amount of NPLs for bank j o, =1,....,O,and p
jo
O
+
will be their
prices.
2.3 Indicators of Income Performance
The goal of this study is to see if bank earnings management has an effect on the bottom line. The return on
assets (ROA) and return on equity (ROE) ratios are therefore used to assess a company's financial performance
in line with previous empirical research (Kothari et al., 2016; Akram et al., 2015; Alhadab and Alsahawneh,
2016). A company's ROA is an important measure of profitability. It measures a company's total assets divided
by total income. These assets can be used to generate income for the company, and this ratio measures that
ability. As a result, this ratio provides insight into the efficiency with which a company's assets are turned into
revenue. The more efficiently a company utilises its own resources, as according to Wen (2010), the higher the
ROA ratio. In the context of financial reporting, ROE is the ratio of a company's net income to its total
shareholder equity. In other words, it's a way of quantifying how much of their money they're getting back in
terms of profit. Internal cash flow is expected to be a strength of a company with a high return on equity
(ROE). As a result, a higher ROE signifies better business performance. The ROE ratio measures a company's
ability to generate a profit for its shareholders. ROE is also a measure of a company's executive's ability to
manage shareholder capital (Ongore and Kusa, 2013).
We estimate a model in which the dependent variable is a proxy for banks' performance (ROA/ROE) and the
main explanatory variable is a proxy for discretionary loan loss provision in order to examine whether earnings
management (via manipulation of loan loss provision) has an impact on banks' current and future performance
(DLLP). Determinants of performance are included in the study, following previous research (Moyer et al.;
Collins et al.; Ahmed et al., 1999; Kanagaretnam et al., 2004; Liu and Ryan, 2006). The (𝐸𝐵𝑇𝑃
𝑖𝑡
), risk-
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2534
www.rsisinternational.org
adjusted Tier 1 capital ratio (𝐶𝐴𝑃_𝑇𝑖𝑒𝑟1
𝑖𝑡
), firm size ( 𝑆𝐼𝑍𝐸
𝑖𝑡
), and leverage ratio (LR) are all adjusted for
(𝐿𝐸𝑉
𝑖𝑡
). There is no way to predict the sign (+/-) of direction for these determinant variables because prior
research Darjezi (2006) showed that the discretionary loan loss provision could be measured with errors. Using
these models, we look at the relationship between DLLP and performance (ROA/ROE).
ROA
t i+
= +
0 1
DLLP
it
+
2
EBTP
it
+
3
CAP Tier_ 1
it
+
4
SIZE
it
+
5
LEV
it
+ ..................................(3)
ROE
t i+
= +
01
DLLP
it
+
2
EBTP
it
+
3
CAP Tier_1
it
+
4
SIZE
it
+
5
LEV
it
+ ...................................(4)
Where:
i = 1
ROA= Return on assets for bank i at the year t.
ROE = Return on equity for bank i at the year t.
DLLP
it
= Discretionary loan loss provision that is calculated as the residual from model (1) for bank i at the
year t.
EBTP
it
= Earrings before tax and loan loss provision for bank i at the year t.
CAP Tier_ 1
it
= Risk-adjusted Tier 1 capital ratio for bank i at the year t.
SIZE
it
= Natural logarithm of market value for bank i at the year t.
LEV
it
= leverage ratio that is calculated as total debt divided by total assets for bank i at the year t.
2.4 Advantages and Drawbacks of The Method
Our approach would be severely lacking without the use of the hidden Markov model and the Bayesian
hierarchical framework. Each prototype is out of sight. Both have the potential to enhance the quality of
earnings measurement. To make use of the hidden Markov model, it is not necessary to monitor surrogate
measures for the true states. Because it does not rely on empirical proxies for the underlying states, our
approach to decoding earnings signals is less vulnerable to misspecification and variable omission. The
likelihood of inaccurate results is reduced as a result. Developing firm-level empirical measures is a common
task in the accounting literature, and there are two common approaches that need to be compared in order to
understand the relative benefits of the Bayesian hierarchical framework. These techniques are employed in the
creation of firm-level empirical measures. To begin, a researcher may assume that the unidentified parameters
are the same across all companies in a given group. Cross-sectional or panel regressions could then be used to
estimate the parameters (Dechow and Dichev, 2002; Hribar et al., 2014).
On the other hands, it’s possible that this approach won't work if the parameters change along a crosssection.
Second, researchers can estimate parameters for each company using firm-specific time-series regressions
(Lipe et al., 1998). Firm-specific analysis relaxes the assumption that model parameters are the same for all
firms, but it requires a long sequence of observations to reliably estimate firm-level parameters, which is
frequently impractical. There is no guarantee that a time series will remain stationary, even if it is very long.
Firm-specific estimates should be "shrunk" to the population mean when there is a dearth of firm-specific data.
The reliability of our measure is based on the assumption that the financial reporting process has a hidden
Markov structure. Nevertheless, there is no reason to believe that this structure is less plausible than those
underlying reduced-form accounting quality measures, which could lead to measurement error if not described
accurately. There is no a priori reason to assume that accounting signals are a linear function of proxy values
for the underlying states, which is why some measures using reduced-form relationships are derived.
Firm-specific parameters are selected from a distribution using a hierarchical Bayesian framework. These
estimates are based on a combination of company data and population distribution. The importance of the first
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2535
www.rsisinternational.org
is inversely proportional to the volume of data collected specifically from the company. Using long-term data
sets is the most important factor in determining firm-specific estimates. Relatively close estimations occur
when time series constraints are applied. Studies show that Bayesian hierarchical estimates have a significant
advantage over conventional estimates in terms of precision (Greenland, 2000). As a result, a Bayesian
hierarchical framework can significantly improve firm-specific estimates for firms with limited time series.
Readers are referred to prior research in Bayesian analysis. Check out the article to learn more about the
advantages of the Bayesian hierarchical framework in terms of technology (Gelman, 2006; Gelman et al.,
2013).
There are three potential costs associated with the Bayesian method. First, even though Bayesian hierarchical
model estimates frequently have a lower variance than firm-specific estimates, they may have a greater bias
(Greenland, 2000). Any misspecification of the prior distributions of unknown parameters can result in biased
parameter estimates. The influence of priors may not be negligible when the ratio of data-based information to
"parameters" is low (Rossi et al., 2005). Lastly, the estimation algorithm utilised in MCMC simulation is
frequently computationally demanding. This is especially true for firm-level parameters for which each
iteration of the Metropolis-Hasting algorithm implementation requires the calculation of the likelihood
function. In spite of these potential costs, we believe the Bayesian hierarchical model can be a valuable tool for
accounting researchers interested in estimating firm-level measures.
3. Analysis of The Results
3.1 Descriptive Statistics
Table 1 below described the descriptive statistics for variables included in the Model 1 and Model 2 and
assuming that the total loan loss provision ranges from RM -1,069,832 to RM 35,120,480 and has a mean
value of RM 1,028,756.8; non-performing loans range from RM 2,933 to RM 42,563,000 and have a mean
value of RM 19,646,287.9; and total loans range from RM 450,650 to RM 3,498,000,000 and have a mean
value of RM 431,400,000. These statistics indicate that the average percentage of nonperforming loans for
Islamic banks is approximately 2%; the lower this percentage, the lower the risk of bank failure. Table 1 also
provides statistics for the aforementioned variables scaled by the initial loan balance.
Table 1. Descriptive Statistics for Variables of Loan Loss Provisions.
Model
Mean
Standard Deviation
Min
LLP
it
1028756.841
4525463.900
-1069832.140
NPL
it
19646287.872
1365478.916
2933.000
TL
it
431,400,000.000
596,200,000.000
450650.000
LPP TLit / it−1
0.005
0.005
-0.002
NPLit−1 /TLit−1
0.085
0.061
0.000
NPL TLit / it−1
0.037
0.017
-0.954
TL TLit / it−1
0.056
0.023
-0.421
N
150
Table 2 provides descriptive statistics for the dependent and independent variables utilised in Models 3 and 4,
which are used to investigate the relationship between earnings management and performance. The main
independent variable of interest, DLLP
it
has a mean value of 0.006 and a range of 0.008 to 0.045. In contrast,
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2536
www.rsisinternational.org
ROA and ROE (our performance proxies) have mean values of 1.056 and 9.084, minimum values of -4.230
and -98.021, and maximum values of 8.415 and 42.071, respectively. Earnings before tax and loan loss
provision EBTP
it
ranges from RM -45,620,000 to RM 35,644,159 with a mean value of RM 1369,5411.2;
adjusted risk Capital Tier 1 ratio CAP Tier_ 1 ranges from 6.150 to 30.109, with a mean of 11.352; banks size
ratio SIZE ranges from 12.329 to 23.754, with a mean of 20.105; and leverage ratio LEV (LEV) ranges from
3.960 to 94.652 with a mean value 34.092.
Table 2. Descriptive Statistics for Variables of Association between Earnings Management and
Performance.
Model
Mean
Standard Deviation
Min
DLLP
it
0.006
0.004
0.008
ROA
1.056
0.104
-4.230
ROE
9.084
15.419
-98.021
EBTP
1369541.213
7048563.281
-4.562e+09
CAP Tier_ 1
11.352
4.514
6.150
SIZE
20.105
2.583
12.329
LEV
34.092
15.265
3.960
N
150
3.2 Regression Coefficient DLLP with ROA
Results of our primary investigation into how earnings management affects the current and future performance
of banks are presented in this section (Models 2). Earnings management has a significant negative impact on
current and future operating performance, as shown in Table 3 (the first performance proxy). Specifically, in
Table 3, column 1 shows that the DLLP
it
coefficient is negative (-42.764) and statistically significant at a 1%
level for the current ROA
it
. At the 1% level, the coefficients of the other explanatory variables EBTP
it
,CAP Tier_
1
it
,SIZE
it
, and LEV
it
show that they are significant predictors of current operating performance ROA
it
.
Explanatory power (Adj.R
2
) for the model is 0.590, which means that explanatory variables account for 59.0%
of the variance in current operating performance ROA
it
, as shown in Column 1 of Table 3.
In addition, the results for future performance ROA
it+1
can be found in column 2 of Table 3. Earnings
management in the current year has a negative effect on operating performance in the following year,
according to the DLLP
it
coefficient (-42.119) which is statistically significant at 1%. Prior studies have shown
that earnings management has a negative impact on a company's bottom line. The explanatory variables, as
shown in Table 6's column 2, are also good indicators of future performance, as shown in the table. Column 3
of Table 3 shows that earnings management has a negative impact on future performance when ROA
it+2
is the
dependent variable. At the 1% level, we can see a negative (- 46.811) correlation coefficient in column 3 of
Table 3 for the measure of DLLP
it
.
According to Table 3, the manipulation of the loan loss provision DLLP
it
has a significant and long-lasting
impact on current ROA
it
and future operating performance. This supports our first hypothesis that banks'
earnings management negatively impacts their current and future operating performance. Earlier studies
(Cohen and Zarowin, 2010, Kothari et al., 2016) have shown that earnings management has a detrimental
effect on future operating results.
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2537
www.rsisinternational.org
Table 3. The Relationship between Earnings Management and Performance by ROA
Variables
ROA
it
Constant
1.287*
(1.811)
DLLP
it
-35.628***
(-2.111)
EBTP
0.000***
(6.548)
CAP Tier_ 1
0.534***
(2.644)
SIZE
-0.048**
(-1.997)
LEV
0.011***
(7.202)
N
150
Adj. R-squared
0.879
Note: ***, **, * Significant at the 1%, 5%, and 10% level.
DISCUSSION AND CONCLUSION
An investigation into the current and future effects of earnings management on bank performance is presented
in this study. Few studies have examined how earnings management practises in Europe's banking industry
affect current and future performance. This study is one of those few. An analysis of 150 bank-year
observations from 2000 to 2021 from 15 Islamic Banks in Malaysia is presented in this article, which adds to
the body of knowledge. The new data demonstrates that Malaysian Islamic Banks that use discretionary loan
loss provisions to manage earnings have lower current and future operating performance (as measured by ROA
or return on assets). This evidence is consistent with previous studies that show a negative correlation between
earnings management and future performance (Cohen and Zarowin, 2010, Kothari et al., 2016)..
Therefore, the findings of this paper have important implications for Malaysian regulators who want to reform
the regulatory environments of the Malaysian banking industry to reduce practises of earnings management
involving the manipulation of provisions for loan losses. These banks' current and future performance has been
adversely affected by the manipulation of their loan loss provision, which has been discovered to be a problem
in Malaysian Islamic Banks. For this reason, it is imperative that all parties involved (such as government
regulators, financial institutions, and independent audit firms) work together to address these reforms.
Based on the method that we discover for Markov Model with earning fidelity; we've created an entirely new
way to assess the quality of an individual's earnings. Estimates of the firm's economic condition are based on a
structural model that separates it from its reporting system. To arrive at the Fidelity estimate, all that is needed
is one measure of earnings and no additional information beyond what is already available (such as
smoothness and accruals quality). Even small businesses can benefit from using this method. There is a
significant conceptual difference between our study and previous research on earnings persistence in that we
did not make a distinction between earnings signals and underlying states. If a company's past earnings are
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2538
www.rsisinternational.org
used to infer its current state, then valuing it must be a filtering problem. We believe that both economic
persistence and earnings consistency are related, but that earnings persistence is distinct from them.
Even though we've focused on earnings, our methodology can be applied to other research contexts. An
advantage of this method is that it can be used when empirical proxies fail to capture latent states accurately
enough. Other types of accounting information (such as recognising a loss contingency) could also be used to
estimate the accuracy of the firm's underlying condition. Because it does not rely on empirical surrogates for
latent states, this method has a significant advantage over other approaches. A less reliable source of earnings
signal is not considered by our metric. Managers' decisions within a given framework of accounting rules and
the specifics of that framework influence the fidelity of earnings in an intuitive way. Future studies could focus
on separating these components and determining how important they are.
Extensive attention is paid to credit risk from both theoretical and empirical perspectives. Nevertheless, despite
the growing number of contributions, most of this research has focused on specific topics, such as how to
evaluate the risks of individual loan operations and or borrowers, and how lenders (banks) respond to these
evaluations (Salas & Saurina, 2002). In this paper, we examine the relationships between bank performance
and risk-taking behaviour, i.e., how various (non-financial) firm characteristics, particularly creditworthiness,
are related to various measures of bank performance. Our study differs from previous contributions that have
emphasised the significance of the relationship between banks and their (non-financial) borrowers in that we
attempt to model explicitly the links between the financial situation of the borrowing firms and the risk banks
assume, as well as how bank performance influences this link.
Research in this area is promising because the analysis can be improved from multiple angles, resulting in
more accurate and precise economic policy recommendations. Even though our study was innovative because
of the proposed efficiency measures, additional measures can be used to further strengthen the analysis. The
potential for LLP manipulation is often overlooked when evaluating these measures, and we believe it is
critical to take a similar approach when defining efficiency. By considering various time lags, we can examine
how bank inefficiency and the risk characteristics of their customers interact over time.
ACKNOWLEDGEMENT
REFERENCES
1. Ahmed, A. S., C. Takeda, and S. Thomas. (1999). ‘‘Bank Loan Loss Provisions: A Reexamination of
Capital Management, Earnings Management and Signaling Effects”. Journal of Accounting and
Economics, 28, 1–25.
2. Alhadab, M., and Alsahawneh, S. (2016). Loan Loss Provision and the Profitability of Commercial
Banks: Evidence from Jordan. International Journal of Business and Management, 11, No. 12.
3. Akram, M. A., Hunjra, A. I., Butt, S., and Ijaz, I. (2015). “Earnings management and organizational
performance: Pakistan VS India”. Basic Research Journal of Business Management and Accounts, 4
(9), 211-220.
4. Anandarajan, A., Hasan, I., and Vivas, A. L. (2003). “The role of loan loss provisions in earnings
management, capital management, and signaling: The Spanish experience”. Advances in International
Accounting, 16, 45-65.
5. Ardekani MA, Nejat Y, Hashemijo M (2012). “Acquisition, Earnings Management and Firm’s
Performance: Evidence from Malaysia”. Journal of Business Studies Quarterly, 4(1), 91-110.
6. Ball, R., (2013). Accounting informs investors and earnings management is rife: two questionable
beliefs. Account. Horiz. 27, 847e853.
7. Beatty, A. L., & Harris, D. G. (1999). The effects of taxes, agency costs and information asymmetry on
earnings management: A comparison of public and private firms. Review of Accounting Studies, 4(3),
299– 326.
8. Beatty, A. L., Ke, B., & Petroni, K. R. (2002). Earnings management to avoid earnings declines across
publicly and privately held banks. The Accounting Review, 77(3), 547–5.
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2539
www.rsisinternational.org
9. Beaver, H. W. and Engel, E. E. (1996). “Discretionary behavior with respect to allowances for loan
losses and the behavior of security prices”. Journal of Accounting and Economics, 22 (1-3), 177-206.
10. Bernhardt, D., Wan, C., Xiao, Z., (2016). The reluctant analyst. J. Account. Res. 54, 987e1040.
11. Bertomeu, J., Ma, P., Marinovic, I., (2018). How Often Do Managers Withhold Information? University
of California at San Diego. Working paper.
12. Ben Othman, K. and Mersni, H. (2014). “The use of discretionary loan loss provisions by Islamic banks
and conventional banks in the Middle East region: A comparative study”. Studies in Economics and
Finance, 31(1), 106-128.
13. Bornemann, S., Kick, T., Memmel, C., and Pfingsten, A. (2012). “Are banks using hidden reserves to
beat earnings benchmarks? Evidence from Germany”. Journal of Banking and Finance, 36(8), 2403-
2415.
14. Burstahler, D., and Dichev, I. (1997). “Earnings management to avoid earnings decreases and losses”.
Journal of Accounting and Economics, 24 (1), 99-126.
15. Cheng, Q, Warfield, T and Ye, M. (2011). “Equity Incentives and Earnings Management: Evidence
from the Banking Industry‟, Journal of Accounting, Auditing and Finance, 26 (2), 317-349.
16. Christensen, P., Feltham, G., (2003). Economics of Accounting e Volume I, Information in Markets.
Springer Series in Accounting Scholarship. Springer, Heidelberg and New York.
17. Cimini R. (2015). “How has the financial crisis affected earnings management? A European study”.
Applied Economics, 47, 302-317.
18. Claessens, S., & Van Horen, N. (2015). The impact of the global financial crisis on banking
globalization. IMF Economic Review, 63(4), 868-918.
19. Cohen, D. A., and P. Zarowin (2010), ‘Accrual-based and Real Earnings Management Activities
Around Seasoned Equity Offerings’, Journal of Accounting and Economics, Vol.50, No.1, pp. 2-19.
20. Collins, J. H., D. A. Shackelford, and J. M. Wahlen. (1995). ‘‘Bank Differences in the Coordination of
Regulatory Capital, Earnings, and Taxes”. Journal of Accounting Research, 33, 263–291.
21. Darjezi, J. I. (2016). "The role of accrual estimation errors to determine accrual and earnings quality",
International Journal of Accounting and Information Management, Vol. 24 Issue: 2, pp.98-115.
22. Debnath, P., (2017). “As saying the Impact of Firm's Growth and Performance on Earnings
Management: An Empirical Observation of Indian Economy”. International Journal of Research in
Business Studies and Management, 4(2), 30-40.
23. Dechow, P. (1994). “Accounting earnings and cash flows as measures of firm performance: the role of
accounting accruals”. Journal of Accounting and Economics, 18, 3–42.
24. Dechow, M., and Skinner, D. J. (2000). “Earnings management: Reconciling the views of accounting
academics, practitioners, and regulators”. Accounting Horizons, 14 (2), 235-250.
25. Dechow, P., Dichev, I., (2002). The quality of accruals and earnings: the role of accrual estimation
errors. Account. Rev. 77, 35e59.
26. Dechow, P., Ge, W., Schrand, C., (2010). Understanding earnings quality: a review of the proxies, their
determinants, and their consequences. J. Account. Econ. 50, 344e401.
27. Dichev, I., Graham, J., Harvey, C., Rajgopal, S., (2013). Earnings quality: evidence from the field. J.
Account. Econ. 56, 1e33.
28. Du, K., 2019. Investor expectations, earnings management, and asset prices. J. Econ. Dyn. Control 105,
134e157.
29. El Sood, H. A. (2012). “Loan loss provisioning and income smoothing in US banks pre and post the
financial crisis”. International Review of Financial Analysis, 25, 64-72.
30. Färe, R., Grosskopf, S., & Lovell, C. A. K. (1994). Production frontiers. Cambridge University Press.
31. Faulkender, M., & Yang, J. (2010). Inside the black box: The role and composition of compensation
peer groups. Journal of Financial Economics, 96(2), 257-270.
32. Filip, A., and Raffournier, B, (2014). “Financial Crisis and Earnings Management: The European
Evidence”. The International Journal of Accounting, 49 (4), 455-478.
33. Gelman, A., (2006). Prior distributions for variance parameters in hierarchical models (comment on
article by Browne and Draper). Bayesian Anal. 1, 515e534.
34. Gelman, A., Carlin, J., Stern, H., Dunson, D., Vehtari, A., Rubin, D., (2013). Bayesian Data Analysis,
3rd ed. Chapman and Hall/CRC, Boca Raton, FL.
35. Goulart, A.M.C. (2007), “Gerenciamento de resultados contábeis em instituições financeiras no Brasil.
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2540
www.rsisinternational.org
36. 211 fls. (Tese Doutorado em Ciências Contábeis)”, Universidade de São Paulo, São Paulo/SP, Brazil.
37. Greenland, S., (2000). Principles of multilevel modelling. Int. J. Epidemiol. 29, 158e167.
38. Guay, W., Kothari, S.P., Watts, R., (1996). Market-based evaluation of discretionary accrual models. J.
Account. Res. 34, 83e105.
39. Healy, P. and Wahlen, J., (1999). “A review of the earnings management literature and its implications
for standard settings”. Accounting Horizons, 13: 365-383.
40. Hegde, S. P., & Kozlowski, S. E. (2021). Discretionary loan loss provisioning and bank stock returns:
The Role of economic booms and busts. Journal of Banking & Finance, 130, 106186.
41. Hribar, P., Kravet, T., Wilson, R., (2014). A new measure of accounting quality. Rev. Account. Stud. 19,
506e538.
42. Kanagaretnam, K., Lobo, G., and Mathieu, R. (2004). “Earnings Management to Reduce Earnings
Variability: Evidence from Bank Loan Loss Provisions”. Review of Accounting and Finance, 3(1),
128148.
43. Kanagaretnam, K., Lobo, G. J., and Yang, D. (2005). “Determinants of signaling by banks through loan
loss provisions”. Journal of Business Research, 58(3), 312-320.
44. Kothari, S. P., Mizik, N., and Roychowdhury, R. (2016) Managing for the Moment: The Role of
Earnings Management via Real Activities versus Accruals in SEO Valuation. The Accounting Review,
91, 559- 586.
45. Lipe, R., Bryant, L., Widener, S., (1998). Do nonlinearity, firm-specific coefficients, and losses
represent distinct factors in the relation between stock returns and accounting earnings? J. Account.
Econ. 25, 195e214.
46. Liu, C. C., & Ryan, S. G. (2006). Income smoothing over the business cycle: Changes in banks’
coordinated management of provisions for loan losses and loan charge-offs from the pre-1990 bust to
the 1990s boom. The Accounting Review, 81(2), 421–441.
47. Ljubisavljević, A., & Jakobsson, C. (2022). Earnings Management during the Covid-19 Pandemic:
Evidence from Sweden.
48. Mostafa, W., & Ibrahim, M. (2019). Firm performance and earnings management. Academy of
Accounting and Financial Studies Journal, 23(3), 1-11.
49. Moyer, S. E. (1990). “Capital Adequacy Ratio Regulations and Accounting Choices in Commercial
Banks.’’ Journal of Accounting and Economics, 13, 123–154.
50. Nikolaev, V., (2017). Identifying Accounting quality. University of Chicago. Working Paper.
51. Ongore,V.O and Kusa, G.B. (2013).“Determinants of Financial Performance of Commercial Banks in
Kenya”. International Journal of Economics and Financial Issues, 3 (1), 237- 252.
52. Owens, E., Wu, J., Zimmerman, J., (2017). Idiosyncratic shocks to firm underlying economics and
abnormal accruals. Account. Rev. 92, 183e219.
53. Pérez, D., Salas-Fumás, V., & Saurina, J. (2008). Earnings and capital management in alternative loan
loss provision regulatory regimes. European Accounting Review, 17(3), 423–445.
54. Primont, D. (1993). Efficiency measures and input aggregation. In W. E. Diewert, K. Spremann, & F.
Stehling (Eds.), Mathematical modelling in economics: Essays in honor of Wolfgang Eichhorn (pp.
288– 294). Springer.
55. Rossi, P., Allenby, G., McCulloch, R., 2005. Bayesian Statistics and Marketing. John Wiley & Sons,
New York.
56. Salas, V., & Saurina, J. (2002). Credit risk in two institutional regimes: Spanish commercial and
savings banks. Journal of Financial Services Research, 22(3), 203–224
57. Schipper, K., (1989). “Commentary on Earnings Management”. Accounting Horizons, 3, 91- 102.
58. Tabassum, N., Kaleem, A., Nazir, M.S. (2015). “Real earnings management and future performance”.
Global Business Review, 16 (1), 1-14.
59. Teoh, S.H., Welch, I., and Wong, T.J. (1998). “Earnings management and the long‐run market
performance of initial public offerings”. The Journal of Finance, 53(6), 1935-1974.
60. Wen, W. (2010). “Ownership Structure and Banking Performance: New Evidence in China”.
Universitat Autònoma de Barcelona Departament D’economia de L’empresa
61. Zang, A. (2011). ‘‘Evidence on the trade-off between real activities manipulation and accrual-based
earnings management”. The Accounting Review, 87(2), 67.
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025
Page 2541
www.rsisinternational.org
62. Zhou, F. S. (2021). Disclosure dynamics and investor learning. Management Science, 67(6), 3429-
3446.
63. Zoubi, T., and Al‐Khazali, O. (2007). “Empirical testing of the loss provisions of banks in the GCC
region”. Managerial Finance, 33 (7), 500-511.