https://dinastipub.
org/DIJEFA Vol.
No.
4, 2025
DOI: https://doi.
org/10.
38035/dijefa.
https://creativecommons.
org/licenses/by/4.
The Impact of Digital Transformation on Risk-Taking of Commercial Banks in Indonesia Muhammad Lukman Hakim1*.
Rizal Ahmad Fauzi2 Institut Pertanian Bogor University.
Jawa Barat.
Indonesia, muh.
lukman@gmail.
Institut Pertanian Bogor University.
Jawa Barat.
Indonesia, rizal.
afauzi@gmail.
Corresponding author: muh.
lukman@gmail.
Abstract: Background: Progress in digital technology has prompted commercial banks to undergo digital transformation to improve operational efficiency and consolidate risk management measures.
The empirical relationship between bank risk-taking and digital transformation is however inconclusive and subject to contingencies.
Purpose: The study aims to examine the effect of digital transformation on credit risk (Non-Performing Loan/NPL), insolvency risk (Z-scor.
, and liquidity risk (Loan-to-Deposit Ratio/LDR) at Indonesian commercial banks during the period from 2014 to 2024.
Design/methodology/approach: This study employs the quantitative technique of panel data regression of 10 Indonesian commercial The Analytic Hierarchy Process (AHP) method is also employed to determine the priority between risk indicators affected by digital transformation in order to guarantee methodological triangulation.
Findings/Result: The regression result indicates that digital transformation is statistically negatively correlated with credit risk (NPL), although the correlation is not significant at the 5% level.
No significant relationship comes out of Z-score and LDR.
AHP results identify that NPL has the highest priority weight of 0.
7445, indicating that digital transformation activities are primarily focused on mitigating credit risk.
Conclusion: Digitalization of Indonesian commercial banks could potentially reduce credit risk but needs further strengthening and testing to obtain statistically significant outcomes.
Emphasis on NPL in AHP analysis highlights the strategic importance of credit risk management in digitalization.
Originality/value (State of the ar.
: This study contributes to the exceedingly small empirical literature on the impact of digital transformation on risk-taking in emerging banking markets.
It combines econometric and decision-making approaches .
egression and AHP) for a first-of-its-kind, combined look at the way digital initiatives prioritize different types of banking risks.
Keywords: digital transformation, risk-taking.
Non-Performing Loan.
Z-score.
Loan-toDeposit Ratio.
AHP.
Indonesian commercial banks.
3732 | P a g e https://dinastipub.
org/DIJEFA Vol.
No.
4, 2025
INTRODUCTION
Digitalization has become a dominant paradigm of the global banking industry's modernization, as in Indonesia.
The emergence of mobile banking, internet banking, and application of big data and artificial intelligence has revolutionized banking operations, extended the reach of services, and enhanced the efficiency of transactions.
As highlighted by Bank Indonesia .
, digital banking transactions in Indonesia increased 158% from 2018 to 2023, indicative of robust consumer demand for digital financial services.
In line with this, the Financial Services Authority (OJK) unveiled the Digital Banking Transformation Blueprint in 2021 as a strategic blueprint to enable systemic and measurable embrace of digital technology among the national banking sector (OJK, 2.
While there are numerous benefits of digital transformation in banking efficiency and in financial inclusion, it also has implications for the financial system's stability, particularly that of the risk-taking behavior of banks.
Risk-taking is defined by Stulz .
as a strategy that increases exposure to credit risk, insolvency risk, and liquidity risk.
On the positive side, technology adoption would boost risk management channels through data processing in real time and portfolio monitoring on an auto-pilot basis.
Conversely, competitive forces and the imperative for innovation in the digital economy may compel expansionist inclinations that usher in new risks (Zhu et al.
, 2.
Previous studies have yielded inconclusive evidence on the relationship between digital transformation and risk-taking.
Hoque et al.
, .
determined that bank digitalization reduces credit and insolvency risks but has a small impact on the risk of liquidity.
Fan et al.
, .
and Li & Zhang .
corroborated this by observing that the efficiency that digital technologies encourage in operations has the ability to neutralize risk-taking behaviors.
On the other hand, studies by Dai et al.
, .
and Feng & Yu .
posit that digitalization will augment risktaking, especially within organizations that face intense competition and intense innovation These results are diametrically opposite since the impacts of digital transformation on risk-taking were found to be contingent on the quality of governance, infrastructure readiness, and organizational characteristics (Liang et al.
, 2023.
Sang et al.
, 2.
In the Indonesian case, empirical research specifically looking at the effect of digital transformation on bank risk-taking is still scarce.
Local literature mostly addresses topics such as efficiency, customer activity, or satisfaction, not addressing thoroughly how digitalization influences the three major risk typesAicredit, insolvency, and liquidityAitogether.
Besides, the prevailing methodological approaches are normally descriptive or half-quantitative, without judgmental triangulation such as the Analytic Hierarchy Process (AHP), which could potentially shed more light on risk prioritization in the digital transformation agenda (Fan et , 2024.
Hoque et al.
, 2024.
Sidorova et al.
, 2.
Against these lacunae, this study aims to provide theoretically and practically significant contributions to the body of knowledge on digitalization in banking risk management.
Theoretically, this research combines Institutional Theory (Scott, 2.
and Resource-Based View (Harrison et al.
, 2.
principles to explain the mechanism between institutional pressures from outside, organizational capabilities, and risk-taking activities.
In practice, the study seeks to deliver evidence-based policy recommendations to industry players and regulators on the line of adaptive and sustainable digital risk management frameworks.
Specifically, this study's research goals are: .
to evaluate the effect of digital transformation on Indonesian commercial bank credit risk.
to evaluate the effect of digital transformation on insolvency risk.
to evaluate the effect of digital transformation on liquidity risk.
Utilizing a quantitative panel data approach from 2014Ae2024 and corroborated using the AHP methodology, this research aims to provide a comprehensive understanding of digitalization's strategic position within providing national financial system stability.
3733 | P a g e https://dinastipub.
org/DIJEFA Vol.
No.
4, 2025
METHOD
This study employed a quantitative design using secondary panel data to analyze the impact of digital transformation on risk-taking behavior of Indonesian commercial banks.
The observations span ten commercial banks with yearly observations for 2014 through 2024, purposively chosen based on data availability and consistency over time.
Data were collected using documentation techniques from publicly available annual reports, accounts, the Indonesian Banking Statistics of the Financial Services Authority (OJK), and macroeconomic statistics provided by Bank Indonesia.
The variable of digital transformation was proxied through a composite index that comprised several observable measures, including digital transactions volume, number of mobile banking users, expenditure on IT, and ratio of branch reduction.
Furthermore, the Digital Maturity Index as constructed by OJK .
in Circular Letter No.
24/SEOJK.
03/2023
was employed where data were available.
This aligns with the institutional digital readiness measurement framework established by regulatory authorities (OJK, 2.
Three risk-taking proxies make up the dependent variables: .
credit risk as measured by the Non-Performing Loan (NPL) ratio, .
insolvency risk as measured by the Z-score, and .
liquidity risk as measured by the Loan-to-Deposit Ratio (LDR).
The Z-score was calculated through the following formula (Laeven & Levine, 2.
ycs Oe ycIycaycuycyce = yaycycycnycyc ycIycCya yaycycyceycyc yua.
cIycCy.
E(ROA) is the standard deviation of return on assets over the sample period.
Several control variables were included to prevent omitted variable bias, including firm size .
og of total asset.
Capital Adequacy Ratio (CAR).
Return on Assets (ROA), credit growth.
CASA ratio.
GDP growth, and central bank interest rate.
These controls have been widely used in the prior banking risk literature to determine the precise effect of digital transformation (Fan et al.
The main analysis method is panel data regression, both fixed effects and random effects To determine the most appropriate model, the Hausman test was applied to test whether individual effects are correlated with explanatory variables (Baltagi, 2.
The general regression equation is given below:
ycUycnyc = yuycn yuIyc yu1 yaycnyciycnycycaycoycnyc yu2 ycUycnyc yuAycnyc Where Y_it is the dependent variable for bank i at time t.
Digital_it is the digital transformation index.
X_it is the vector of control variables.
alpha_i is the unobserved bankspecific effect.
lambda_t is the year fixed effect.
and epsilon_it is the error term.
In addition to this, the current study employs the Analytic Hierarchy Process (AHP) as a complementary decision-support method to establish the relative importance of each risk AHP, which was established by Lande et al.
, .
, employs pairwise comparisons to arrive at weights for every criterionAiin this case, credit risk, insolvency risk, and liquidity riskAibased on expert opinion.
AHP presents a structured and replicable approach to decisionmaking in complex scenarios such as digital transformation in risk management (Sidorova et , 2.
Figure 1 shows this study's conceptual framework.
It displays the hypothesized relationships between digital transformation and the three types of bank risks, after adjusting for firm-specific and macroeconomic factors.
3734 | P a g e https://dinastipub.
org/DIJEFA Vol.
No.
4, 2025
Hypothesis The study employs the Institutional Theory (Scott, 2.
and the Resource-Based View (Barney, 2.
in developing its hypotheses.
With the Institutional Theory, it is external influences such as regulation and technological change that compel organizations to implement changes in their strategies, including risk management.
With the Resource-Based View, however, internal capabilitiesAisuch as digital infrastructureAimust be considered as critical resources in seeking competitive advantage while limiting exposure to risk.
More and more literature has addressed the relationship between digital transformation and bank risk-taking, and results have been diverse.
While some research indicates digital transformation translates into lower risk, some studies indicate otherwise.
Hoque et al.
confirmed that digital initiatives at Vietnamese banks reduced credit and insolvency risk significantly but had no statistically significant effect on decreasing liquidity risk.
Similarly.
Li & Zhang .
reported that digital transformation curbed overall risk-taking by means of enhanced operating efficiency and reduced management costs.
Fan et al.
also confirmed a decline in systemic risk among Chinese commercial banks due to digitalization, indirectly mediated by increased operating efficiency.
Converse results point to the probability that digital transformation could promote riskier Dai et al.
concluded that digitalization increased the risk-taking level, especially that of private and technology-intensive companies.
Feng & Yu .
also established that managerial abilities complemented the risk-taking effect of digitalization, particularly against the backdrop of policy uncertainty.
Studies by Sang et al.
, .
and Liu et al.
, .
further emphasized that digital innovation would enhance exposure to risk in settings that prioritize growth and innovation, suggesting that the effects of digital transformation may be context-dependent.
Despite the literature at hand, studies focusing on Indonesian commercial banks in particular and incorporating panel quantitative data in conjunction with evaluation methods like the Analytic Hierarchy Process (AHP) are not common, if not non-existent.
This study attempts to fill this gap through examining the differential effect of digitalization on credit risk, insolvency risk, and liquidity risk.
Therefore, hypotheses to be tested are as follows:
H1: Digital transformation has significantly an adverse effect on credit risk .
s measured by NPL) in Indonesian commercial banks, as shown by Hoque et al.
Fan et al.
and Li & Zhang .
H2: Digital transformation has significantly an adverse effect on insolvency risk .
s measured by Z-scor.
in Indonesian commercial banks, in line with Hoque et al.
and Fan et .
H3: There is a significant negative effect of digital transformation on liquidity risk .
s measured by LDR) among Indonesian commercial banks, though previous research such as Hoque et al.
show the effect is not statistically significant and there should be more research conducted in Indonesia.
RESULTS AND DISCUSSION
Descriptive Statistics The descriptive statistics presented in Table 1 provide an overview of the distribution, central tendency, and variance of the primary variables used in this study.
These are the Digital Transformation Index.
Non-Performing Loan (NPL) ratio.
Z-Score .
s a proxy for insolvency ris.
Loan-to-Deposit Ratio (LDR), and Bank Size in logarithmic form.
Observations consist of 230 for 10 commercial banks listed in Indonesia from 2014 to 2024.
3735 | P a g e https://dinastipub.
org/DIJEFA Vol.
No.
4, 2025
Table 1.
Descriptive Statistics DIGITAL_INDEX NPL
Z_SCORE
LDR
SIZE
Sum Sum Sq.
Dev.
Observations Mean Median Maximum Minimum Std.
Dev.
Skewness Kurtosis Jarque-Bera Probability The Digital Transformation Index is 0.
7058 with a standard deviation of 0.
indicating that it has moderate variation over time and banks.
The index ranges from the minimum value of 0.
460 up to the maximum of 0.
950, reflecting heterogeneity in digital technology take-up among Indonesian banks.
The distribution is nearly symmetrical, as shown through the value of the skewness of close to zero (Oe0.
, and Jarque-Bera test statistic .
p < 0.
confirms that the variable is very much not normally distributed.
The average NPL ratio is 2.
55%, which represents relatively low credit risk levels in the observed banks, but the values range from 1.
54% to 3.
The distribution is also almost symmetrical .
kewness = Oe0.
, but the kurtosis is lower than 3 .
, which indicates a flat The Jarque-Bera test .
p < 0.
also confirms non-normality in the data This indicates the presence of small outliers or heavy tails, and these can impact regression estimates if not properly handled.
For the Z-Score, which reflects insolvency risk, the mean is 46.
97 and the standard deviation is 7.
This variable is also nearly normally distributed with minimal skewness (Oe0.
and low kurtosis .
However, the Jarque-Bera test .
p < 0.
indicates non-normality, therefore, careful interpretation must be used in inferential analysis.
The greater the Z-Score, typically, the more solvency there is and less chance of bank default.
The average Loan-to-Deposit Ratio (LDR) is 90.
11%, indicating that banks lend 90% on average from their deposits.
This value falls within the acceptable regulatory limit, implying judicious handling of liquidity.
The skewness is negative (-0.
with minimal spread .
tandard deviation = 5.
Nevertheless, the Jarque-Bera test statistic value .
p < 0.
reflects the presence of non-normal characteristics.
Finally.
Bank SizeAimeasured by the natural logarithm of total assetsAihas an average 47 with a standard deviation of 0.
It varies between 16.
04 and 18.
99, reflecting minimal size variation among the sampled banks.
The distribution is nearly symmetrical .
kewness = 0.
, and Jarque-Bera test statistic .
p < 0.
suggests that the data are mildly skewed away from normality.
These statistical properties highlight the necessity for the use of sound estimation procedures in future regression models to allow for valid inference.
3736 | P a g e https://dinastipub.
org/DIJEFA Vol.
No.
4, 2025
Model Selection Tests Table 2.
Model Selection Tests Test Chow test Hausman Test Lagrange Multiplier Tests Prob.
Result Common Effect Model Random Effect Model Random Effect Model The panel model selection tests are given in Table 2.
The Chow Test provides a p-value 0589, which is slightly higher than the 5% critical value, suggesting that the fixed effect model is not appreciably better than the common effect model.
That being said, the Hausman Test has a p-value of 0.
6009 (> 0.
, suggesting that the random effects model has a better fit than the fixed effects model.
Furthermore, the Lagrange Multiplier (LM) test also produces a p-value of 0.
2888, which is further evidence in favor of the use of the random effects model over the common effect model.
Based on these results, the random effect model is selected as the most fitting estimation technique for this study.
Regression Results Table 3.
Regression Results: The Impact of Digital Transformation on Credit Risk (NPL) Variable DIGITAL_INDEX SIZE
CAR
ROA
KREDIT_GROWTH
CASA
PDB_RIIL
Coefficient 3,702 -0,500 -0,047 0,009 0,052 -0,011 -0,002 -0,027 Std.
Error
0,968
0,260
0,046
0,017
0,101
0,009
0,005
0,043
t-Statistic 3,826 -1,923 -1,011 0,537 0,519 -1,211 -0,278 -0,622 Prob.
0,000
0,056
0,313
0,592
0,605
0,227
0,782
0,534
ycUycnyc = 3.
702097 Oe 0.
500111 Oe 0.
052323 Oe 0.
Oe 0.
001503 Oe 0.
Table 3 presents the regression of the impact of digital transformation on credit risk, as a proxy by Non-Performing Loans (NPL).
The negative coefficient of the DIGITAL_INDEX .
is statistically significant at the 10% level .
= 0.
, indicating that digital transformation can decrease credit risk, though the result is not statistically significant at the conventional 5% level.
This finding is in line with Hoque et al.
, which set that digital transformation reduces credit risk through improved operating efficiency.
The remaining control variables, including SIZE.
CAR.
ROA.
CREDIT_GROWTH.
CASA, and GDP_RIIL, do not have statistically significant effects on NPL.
This aligns with the suggestion that although digitalization is positive in terms of improved credit risk management, it does not work in a standalone manner against general macroeconomic or structural factors.
Table 4.
Regression Results: The Impact of Digital Transformation on Insolvency Risk (Z-Scor.
Variable DIGITAL_INDEX SIZE
CAR
ROA
Coefficient 54,791 0,817 -0,281 -0,126 -0,492 Std.
Error
12,562
3,368
0,597
0,215
1,302
t-Statistic 4,362 0,243 -0,470 -0,586 -0,378 Prob.
0,000
0,809
0,639
0,559
0,706
3737 | P a g e https://dinastipub.
org/DIJEFA KREDIT_GROWTH CASA
PDB_RIIL
Vol.
No.
4, 2025
-0,195
0,014
-0,031
0,119
0,071
0,552
-1,643
0,198
-0,057
0,102
0,843
0,955
ycUycnyc = 54.
817084 Oe 0.
280503 Oe 0.
125831 Oe 0.
492239 Oe 0.
013994 Oe 0.
Table 4 presents the regression output for insolvency risk using the Z-Score.
The coefficient for DIGITAL_INDEX is positive .
but statistically not significant .
= .
, which indicates that digital transformation does not have a significant impact on insolvency risk during the observed period.
Similarly, no control variables are significant, including SIZE .
= 0.
CAR .
= 0.
, and ROA .
= 0.
Contrary to evidence by Fan et al.
, which observed that digitalization dampens systemic risk through operational The insignificance in this case is due to the fact of the relatively low maturity of digital infrastructure in the Indonesian banking sector or limitations to quantify long-term consequences of risk stabilization.
Table 5.
Regression Results: The Impact of Digital Transformation on Liquidity Risk (LDR) Variable DIGITAL_INDEX SIZE
CAR
ROA
KREDIT_GROWTH
CASA
PDB_RIIL
Coefficient 77,905 1,694 0,520 0,026 -0,731 -0,096 0,030 0,439 Std.
Error
9,805
2,632
0,466
0,168
1,019
0,093
0,055
0,432
t-Statistic 7,945 0,644 1,114 0,157 -0,717 -1,033 0,553 1,017 Prob.
0,000
0,521
0,267
0,875
0,474
0,303
0,581
0,310
ycUycnyc = 77.
026417 Oe 0.
730849 Oe 0.
Test for the impact of digital transformation on liquidity risk, defined in terms of the Loan-to-Deposit Ratio (LDR).
The coefficient of DIGITAL_INDEX is positive .
but statistically insignificant .
= 0.
, which means that digital transformation does not have any statistically significant impact on liquidity risk.
Control variables SIZE.
CAR.
ROA.
CREDIT_GROWTH, and CASA also show no significant relationship with LDR.
This finding is consistent with Hoque et al.
, which reported that it failed to identify a significant effect of digital transformation on liquidity risk.
One probable explanation is that technology initiatives within banking care more about customer interface and automation rather than radically changing liquidity management procedures.
Figure 1.
Analytical Hierarchy Process (AHP) Model Structure of Digital Transformation and Banking Risk Indicators 3738 | P a g e https://dinastipub.
org/DIJEFA Vol.
No.
4, 2025
Figure 1 illustrates the hierarchical structure utilized in the Analytic Hierarchy Process (AHP) where Digital Transformation is set as the principal criterion that influences three critical indicators of risk: Non-Performing Loan (NPL).
Z-Score .
isk of insolvenc.
, and Liquidity Risk.
The format of the model gives the foundation for performing pairwise comparisons to determine the relative weights of each dimension of risk with respect to digital transformation initiatives for commercial banks.
Figure 2.
Pairwise Comparison Matrix: Risk Indicators in Relation to Digital Transformation Figure 2 illustrates the pairwise comparison matrix utilized to rank the significance of the three indicators of risk relative to the digital transformation criterion.
The results were as follows: NPL was considered far more significant relative to Liquidity Risk and moderately more significant relative to Z-Score.
These qualitative evaluations are then transformed into quantitative scores through Saaty's AHP scale such that consistency testing can be performed and ultimate priority weights derived.
The inconsistency index from the matrix is 0.
which is well within the normally accepted range of 0.
10 to 0.
15, and thus the pairwise comparisons are reasonably consistent (Lande et al.
, 2.
Table 6.
Priority Weights of Risk Indicators in the Context of Digital Transformation Inconsistency Name
0,1144
Normalized Idealized Non-Performing Loan (NPL) 0,7445 1,0000 Liquidity Risk
0,1499
0,2014
Z-Score
0,1056
0,1419
As revealed by Table 6, the AHP-generated priority weights also indicate that Credit Risk (NPL) has the highest normalized value of 0.
7445, also idealized to 1.
This indicates that among the three dimensions, credit risk is the most critical area affected by digital Liquidity Risk ranks second at a normalized weightage of 0.
dealized = 2.
, while Z-Score .
nsolvency ris.
is the least significant at a normalized weightage of 1056 .
dealized = 0.
This suggests that stakeholders perceive digital transformation to have greatest impact on credit risk mitigation, followed by liquidity management, and least on insolvency measures.
Managerial Implications The results show that digital transformation has the highest perceived impact on credit risk, as revealed through both regression and AHP outputs.
This means that from the point of view of a bank manager, high consideration should be provided to making high-priority digital investments in the domains of credit evaluation, tracking loans, and early warning.
Boosting digital expertise in these domains can help reduce non-performing loans and credit quality in Though no significant effects were identified on insolvency and liquidity risk, digital technologies can indirectly provide benefits in terms of enhanced reporting and real-time financial management.
Banks are therefore required to adopt an extensive digital risk framework where technology deployment is synchronized with key risk priorities.
Regulators 3739 | P a g e https://dinastipub.
org/DIJEFA Vol.
No.
4, 2025
such as OJK and Bank Indonesia can also apply these insights to policy by encouraging banks to integrate digital instruments into credit risk management processes in a way that allows digital transformation to not only enhance efficiency but also financial system stability.
CONCLUSION
This study aimed to examine the impact of digital transformation on risk-taking behavior for Indonesian commercial banks, in particular credit risk, insolvency risk, and liquidity risk.
The findings indicate that digital transformation exerts a marginally significant negative impact on credit risk (NPL), demonstrating its potential to improve the quality of credit through improved data-monitoring and automation.
However, there was no observed effect on insolvency risk (Z-scor.
or liquidity risk (LDR), suggesting that these risk dimensions are not influenced directly by digital programs or require longer time intervals to observe observable AHP analysis also validated the importance of credit risk, which was positioned as the most strongly focused risk dimension in the context of digital transformation.
Thus, the study concludes that digital transformation is a primary force behind better risk governance, but its effect is greatest in credit risk management.
Recommendations Banks are advised to align digital transformation programs to their risk management agenda, particularly to strengthen credit risk assessment and control frameworks.
The focus needs to be on strengthening digital platforms for loan origination, monitoring, and recovery to contain NPLs effectively.
Furthermore, to harvest the broader benefits of digitalization, banks are also recommended to invest in overall risk systems spanning liquidity and solvency aspects, although their short-term impacts are not statistically significant.
Policymakers and regulators such as OJK and Bank Indonesia are called upon to develop targeted incentive and guideline frameworks that encourage responsible digital adoption, especially in credit Further studies are recommended to explore the long-term influence of digital transformation across different risk categories as well as to examine the moderating role of institutional capacity and regulatory support.
REFERENCES