This paper investigates the nature and intensity of digitalization's impact on the corporate sector's financial security in the context of digital transformation. The study aims to comprehensively assess this impact using multidimensional econometric tools to develop well-founded scientific and practical recommendations for enhancing ifnancial resilience at the enterprise level. The research addresses several key tasks: adapting methodological approaches for quantifying financial security levels considering digital economy transformation processes; constructing classical OLS regression models to identify the basic relationships between digitalization and financial security; applying panel econometric models that account for spatial and temporal heterogeneity; and implementing a Bayesian modeling approach to obtain robust estimates under data limitations and potential violations of classical assumptions. Empirical results reveal a statistically significant negative relationship between the level of digitalization and financial security indicators of the corporate sector. This relationship is more pronounced among enterprises with higher levels of financial stability. The identified negative association should not be interpreted as an argument against digitalization, but rather as evidence of the need for a comprehensive approach to managing digital transformation, with attention to potential financial risks. The study proposes recommendations for adapting business models, improving financial monitoring systems, and strengthening the capacity of enterprises to address emerging challenges in the digital age.
The rapid development of digital technologies over the past decades has transformed production and consumption processes and fundamentally altered the financial relations structure at all economic system levels. Digitalization, acting as a catalyst for structural changes in business models, data processing and analysis technologies, and forms of interaction with clients and investors, creates fundamentally new conditions for the functioning of the corporate sector. In this context, ensuring financial security becomes particularly relevant as a fundamental basis for the sustainable development of corporate structures amid the digital transformation of the economy.
Digitalization, as a pivotal phase of the Fourth Industrial Revolution, is fundamentally reshaping all dimensions of corporate operations, including the financial security management system. In recent years, there has been a notable increase in scholarly attention to the implications of digital technologies for the financial resilience and safeguarding of enterprises operating under volatile external conditions.
Researchers emphasize several key dimensions through which digitalization influences corporate
ifnancial security. Among the most frequently cited positive efects are enhanced operational
eficiency and cost optimization. Automation and digital tools significantly streamline financial processes,
minimize human-induced errors, and reduce transaction costs, enabling corporations to operate more
eficiently and strategically allocate resources [
Digitalization also plays a critical role in expanding access to finance and promoting financial inclusion.
Digital platforms and fintech innovations facilitate broader access to financial services, particularly for
ifrms historically facing funding constraints, enabling them to secure financing and manage liquidity
more efectively [
Another significant contribution of digitalization lies in strengthening risk management and auditing
practices. Integrating digital tools provides deeper insights into corporate operations, enhances audit
quality, and enables real-time monitoring of financial transactions, reinforcing internal controls and
increasing transparency. Artificial intelligence and machine learning support advanced risk modeling
and fraud detection, contributing to developing more robust financial security systems [
Emerging empirical evidence also suggests that digital transformation may mitigate excessive
corporate financialization by incentivizing productive investments over speculative financial activities. This
shift potentially fosters more sustainable and resilient financial structures in the long term [
Ultimately, existing research supports the hypothesis that digitalization contributes to the resilience
of financial systems under crisis conditions [
At the same time, scholars increasingly highlight the risks and challenges associated with digital
transformation, which may, in some cases, undermine financial security. Chief among these are
cybersecurity risks. The expansion of digital connectivity and the exponential growth of data volumes
increase the potential attack surface for cybercriminals, thereby exposing corporations to greater risks
of data breaches, fraud, and other cyber threats [
Digitalization also introduces regulatory complexities and compliance costs, as corporations must
continuously adapt to evolving data protection, privacy, and financial reporting standards. Meeting
these requirements often demands substantial financial and human capital, which may divert resources
from innovation and transformational initiatives [
In addition to the technological and operational dimensions of financial security, scholars emphasize
the importance of institutional factors in shaping the financial sustainability of the corporate sector.
The institutional quality of the financial system, including regulatory eficiency, transparency, and the
degree of institutionalization, directly influences firms’ ability to adapt to digital transformation and
maintain financial stability. As highlighted by Y. Kovalenko, an institutional approach allows for a
more comprehensive assessment of financial sector eficiency, particularly in environments undergoing
systemic change [
Despite substantial scholarly contributions to the study of digitalization and financial security, their interrelationship remains underexplored, particularly in empirical validation regarding the nature and intensity of digital transformation’s impact on the corporate sector’s financial security indicators. Most existing studies either focus on qualitative aspects of this interaction or examine isolated sector-specific case studies. This fragmented approach hinders the development of a comprehensive understanding of the phenomenon and limits the identification of broader patterns and trends.
The primary objective of this study is to conduct a comprehensive assessment of the nature and intensity of the impact of digitalization processes on the financial security of the corporate sector. This is achieved by applying multidimensional econometric tools to formulate evidence-based scientific and practical recommendations to strengthen financial security under conditions of digital transformation.
To achieve this objective, the following scientific and practical tasks have been identified and implemented: • To select methodological approaches for the quantitative assessment of the financial security level of the corporate sector, taking into account the transformational dynamics of the digital economy • To construct and evaluate classical Ordinary Least Squares (OLS) regression models of the impact of digitalization on corporate financial security, to determine the baseline characteristics of the relationship between the relevant indicators • To develop and implement panel econometric models that incorporate the spatial and temporal heterogeneity of the examined processes, thereby enabling a more nuanced analysis of digitalization’s influence on financial security • To apply a Bayesian modeling approach to assess the impact of digitalization on corporate financial security, providing robust estimations under data limitations and potential violations of classical econometric assumptions • To formulate evidence-based scientific and policy recommendations for ensuring the financial security of the corporate sector in the context of intensified digitalization processes at both macro and micro levels, based on the identified empirical regularities.
To ensure the study’s representativeness, panel data were collected from all 27 EU member states over 2016–2023, resulting in suficient observations (n = 214) to support robust statistical analysis.
To assess the level of digitalization, the study employed the Digital Economy and Society Index (DESI)
for the years 2017 to 2024. Compiled by the European Commission [
It is important to note that the DESI methodology was revised in 2023, after which the European
Commission discontinued the publication of the composite index and shifted its focus to disaggregated
sub-indices. To preserve temporal comparability and maintain methodological consistency, this study
recalculated the composite DESI scores for 2023 and 2024 using the 2022 methodology [
Given the structure of DESI reporting, where each year’s index reflects data from the previous calendar year (e.g., DESI 2024 reflects statistics from 2023), the study applied a temporal adjustment by shifting DESI values by one year relative to financial security indicators. Thus, for instance, the DESI 2024 index is aligned with financial security indicators for 2023, ensuring accurate temporal synchronization of the analyzed data.
To measure the financial security of the corporate sector, oficial statistics from Eurostat for the
period 2016–2023 were utilized. The analysis incorporated two distinct groups of indicators:
2. Non-financial transactions indicators (dataset NASA_10_NF_TR) [
For both indicator groups, the analysis focused on sector S11 – Non-financial corporations, which
is in line with the study’s objective of evaluating the financial security of the corporate sector of the
economy. The methodology proposed by Y. Pavlyuk and O. Laktyonova [
, , , , represent the composite sub-indices of profitability, business activity, liquidity and solvency, financial stability, and investment appeal, respectively.
, , , , are the weighting coeficients assigned to each respective subcomponent in the integrated assessment.
Each sub-index (e.g., profitability, liquidity) was in turn calculated as a weighted average of selected individual indicators using the following formula:
= ∑︁ · , (2)
=1 where denotes the composite value of the -th component of financial security, is the -th indicator within the -th component, represents the weight of the -th indicator in evaluating the -th component; is the number of indicators comprising the -th component.
The structure of the financial security index and the corresponding weighting coeficients for each component and indicator are presented in Table 1.
Unlike the original methodology [
A comprehensive set of econometric models of varying complexity and assumptions was applied to assess digitalization’s impact on the financial security of the corporate sector. This approach ensured the robustness of the results and enabled the exploration of diferent dimensions of the relationship between variables: 1. Classical Linear Regression (OLS):
= 0 + 1 · + , where 0 is intercept, 1 is DESI coeficient (main parameter of interest), ∼ (0; error variance.
(3) 2), 2 is 2. Robust Linear Regression with Adjusted Standard Errors: Estimation was conducted with heteroskedasticity-consistent (HC1) robust standard errors to correct for potential heteroskedasticity in the residuals. 3. Robust Regression (M-estimation): The Huber M-estimator was applied to mitigate the influence of outliers on coeficient estimation, enhancing model reliability in the presence of non-normal disturbances. 4. Polynomial Regression: Non-linear specifications were introduced to capture possible curvilinear efects of digitalisation on financial security:
= 0 + 1 · + 2 · 2 + .
5. Quantile Regression: The model was estimated at various quantiles of the dependent variable
distribution ( = 0.25, 0.50, 0.75) to explore heterogeneous efects of digitalisation across diferent
levels of financial security [
= 0 + 1 · + + ,
where denotes the country-specific efect [
Posterior distribution for 1: ∼ (
; 2), = 0, 1, ... 2 ∼ (; ), 1|, X ∼ ( , 2
), = ︁( ∑︀ · + 1 ︁)
2
︁( ∑︀ 2 +
2
was obtained using the No-U-Turn Sampler (NUTS), an eficient implementation of the
Hamiltonian Monte Carlo (HMC) method [
The model was estimated using four independent Markov chains, each comprising 4000 iterations with a 2000-iteration warm-up period, yielding 8000 valid posterior samples for inference.
A comprehensive set of diagnostic tests was applied to evaluate the model quality and ensure the results’ robustness. These diagnostics allowed for the validation of model assumptions and the appropriateness of the panel data structure.
The Breusch-Pagan Lagrange Multiplier (LM) Test was employed to determine the presence of random
efects, thereby evaluating the suitability of the panel data framework [
The Shapiro-Wilk Normality Test was used to test for normality of residuals. The assumption of normally distributed residuals is central to classical linear regression models, particularly for the validity of standard errors, confidence intervals, and significance tests. The null hypothesis ( 0) posits that the residuals follow a normal distribution.
The Studentized Breusch-Pagan Test was applied to check for heteroskedasticity - i.e., non-constant variance of residuals across observations. Heteroskedasticity can lead to ineficient estimates and biased standard errors. The null hypothesis (0) assumes homoskedasticity, whereas the alternative hypothesis (1) indicates the presence of heteroskedasticity in the model.
The Pesaran CD Test was implemented to test for cross-sectional dependence. This test determines whether residuals are correlated across cross-sectional units (e.g., countries or regions) at a given point in time. The null hypothesis (0) states that no cross-sectional dependence exists (i.e., residuals are uncorrelated between units). In contrast, the alternative hypothesis (1) indicates that residuals are correlated, implying the existence of interdependence between countries.
To detect the presence of serial correlation in idiosyncratic errors, the Breusch-Godfrey/Wooldridge
Test was used. Serial correlation implies that residuals from one time period are correlated with those
from another. This violates a key assumption of classical regression models: residuals should be
independently distributed. The null hypothesis (0) assumes no serial correlation in the idiosyncratic
errors [
The Hausman Test was employed to decide between Fixed Efects (FE) and Random Efects (RE)
panel models. The null hypothesis (0) states that individual-specific efects are uncorrelated with the
regressors, making the RE model both consistent and eficient. If the null is rejected, the FE model is
preferred due to potential endogeneity between individual efects and explanatory variables [
Finally, the F-test for Two-Way Efects was applied to assess the significance of both individual and time-specific efects in the panel dataset. This test aims to determine whether cross-sectional (country-specific) and temporal efects should be incorporated into the model. The null hypothesis (0) states that there are no significant individual or time efects, supporting using a simpler pooled model. However, the alternative hypothesis (1) implies that a two-way efects model provides a more accurate specification by accounting for unobserved heterogeneity across both dimensions.
The econometric analysis was conducted using custom scripts developed by the authors [
For the specification and estimation of the Bayesian regression models, the brms package was utilized
[
Applying the aforementioned econometric modelling techniques, a comprehensive analysis was conducted to examine the relationship between the level of digitalization, measured by the Digital Economy and Society Index (DESI), and financial security indicators of the corporate sector. The results reveal a statistically significant negative association between these variables, consistently observed across most employed models.
The classical linear regression model (Ordinary Least Squares, OLS) yields a coeficient of digitalization on corporate financial security equal to –0.028 (p < 0.01). This suggests that a one-unit increase in the DESI score is associated with an average decrease of 0.028 units in the financial security indicator (see Table 2).
Although the model demonstrates statistical significance ( = 9.515, = 0.002), the coeficient of determination 2 is only 0.043, indicating that digitalization explains approximately 4.3% of the variation in financial security across the corporate sector. However, the primary objective of this analysis is not to construct a predictive model for corporate financial security. It is evident a priori that digitalization is not the sole determinant of financial resilience. Instead, the purpose is to evaluate the presence and direction of the efect of digitalization on financial security.
Overall, the linear regression model representing the impact of digitalization on corporate financial security in EU countries can be expressed by the following general equation: = 2.122 − 0.028 · + 2 · X i + , (10) where Xi denotes the covariates (a vector of control variables accounting for heterogeneity in financial security outcomes).
The Breusch–Pagan Lagrange Multiplier test for random individual efects (Table 4) yields a test statistic of 2 = 0.254 ( = 0.6146). Thus, the null hypothesis of no significant individual efects cannot be rejected. This finding implies that a pooled regression model may be suficient for the analysis, and incorporating country-specific efects is not critically necessary in this specification.
Model diagnostics revealed a violation of the normality assumption for the residuals, as indicated by the Shapiro–Wilk test ( = 0.700, < 0.001), suggesting a departure from one of the key assumptions of classical linear regression. On the other hand, the Breusch–Pagan test did not detect heteroskedasticity ( = 0.448, = 0.503), which represents a positive aspect of the model’s specification.
Model Classical linear regression (OLS) OLS with robust standard errors Robust regression (M-estimation) Polynomial regression (linear term) Polynomial regression (quadratic term) Quantile regression (25%) Quantile regression (50%) Quantile regression (75%) -0.028 -0.028 -0.017
Robust regression techniques were employed to account for potential issues arising from the residual distribution and the presence of outliers. The model estimated using heteroskedasticity-consistent standard errors (HC1) confirmed the negative efect of digitalization (–0.028), albeit with a slightly higher standard error (0.012) and reduced statistical significance ( = 0.015) (see Table 2).
Further robustness checks using Huber’s M-estimation yielded a coeficient of –0.017, indicating a somewhat weaker magnitude of the efect, although it remained statistically significant. Notably, the standard error decreased to 0.004, enhancing the estimate’s reliability. The use of robust estimation methods is particularly justified in light of the observed deviations from the normality assumption, reinforcing the robustness of the obtained results (Table 2).
An assessment of nonlinearity by including a quadratic term in the model did not indicate any statistically significant improvement in fit ( = 0.474, = 0.492). The coeficient on the squared DESI term (–0.0004) was statistically insignificant ( = 0.492), suggesting no evidence of a parabolic (nonlinear) relationship between digitalization and corporate financial security. This finding supports the conclusion that the relationship between the investigated variables is predominantly linear in nature.
A comparison of linear and polynomial regression models is presented in Figure 1.
In addition, quantile regression was conducted to explore how the efect of digitalization varies across diferent points in the distribution of corporate financial security. The results indicate that the magnitude of the negative efect increases when moving from lower to higher quantiles: • 25th percentile: –0.014 (not statistically significant) • 50th percentile (median): –0.017 ( < 0.05) • 75th percentile: –0.026 ( < 0.05)
These findings suggest a heterogeneous efect of digitalization: the negative impact is more pronounced among corporations with higher levels of financial security. In contrast, the efect is minor and statistically insignificant for those with lower financial security scores.
Although the Breusch–Pagan Lagrange Multiplier test failed to reject the null hypothesis of no significant individual efects, using panel data techniques remains justified due to theoretical considerations regarding unobserved heterogeneity across countries. The Hausman test (Table 5) rejected the null hypothesis at the 5% significance level, indicating that the fixed efects model is preferable to the random efects specification. This result suggests that the individual country efects are likely correlated with the explanatory variables, and thus, the fixed efects model provides more consistent estimates.
The random efects model suggests that a one-unit increase in the Digital Economy and Society Index (DESI) is associated with a 0.028-unit decrease in the corporate financial security index ( < 0.01). This result is statistically significant using conventional, robust, and Driscoll–Kraay standard errors, accounting for cross-sectional dependence. Notably, this estimate fully aligns with the findings from the classical linear regression model.
In contrast, the fixed efects model yields a weaker negative relationship ( = ˘0.010), which is not statistically significant ( = 0.433). The considerable diference between the estimates obtained from the fixed and random efects models confirms the Hausman test results, indicating a potential misspecification of the random efects model.
These findings reveal several methodological challenges, highlighting the need for more flexible and robust modeling approaches, such as Bayesian regression.
The results of the panel data analysis on the impact of digitalization on corporate financial security highlight several limitations of the classical approach-namely, violations of the normality assumption, the presence of cross-sectional dependence, and the sensitivity of results to model specification.
Bayesian regression ofers a flexible methodological alternative that addresses these issues and enables more reliable and interpretable inferences. In particular, the Bayesian approach provides: • Robustness to violations of classical assumptions • A natural framework for incorporating estimation uncertainty • The ability to integrate prior knowledge • Greater flexibility in modeling complex relationships
Thus, applying Bayesian regression to assess the impact of digitalization on corporate financial security is feasible and necessary for obtaining reliable and theoretically grounded conclusions, particularly under data constraints and complex interdependencies.
The results of the Bayesian regression analysis, conducted in conjunction with panel data methods to assess the relationship between digitalization and the financial security of the corporate sector, are presented in Table 7. An assessment of the diagnostic metrics indicates the high quality of the Bayesian model estimation: 1. The Rhat statistic equals 1.00 for all model parameters, indicating excellent convergence of the Markov chains. Rhat values close to 1 confirm that the independent chains produce consistent parameter estimates, supporting the stability and reliability of the results. 2. The efective sample size (ESS) is high across all parameters (Intercept, DESI coeficient, sigma), further supporting the robustness of the posterior estimates.
Overall, the diagnostics confirm the reliability and stability of the parameter estimates generated by the Bayesian model.
The analysis reveals a statistically significant negative efect of the level of digitalization on the ifnancial security of the corporate sector. The 95% credible interval does not include zero (Figure 2), confirming the credibility of the identified efect. A one-unit increase in the DESI index is associated with an average reduction of 0.03 units in the financial security index, all else equal. This result is 95% confidence interval (lower) consistent with the findings of the classical regression analysis, which indicated a decrease of 0.028 units.
The standard deviation of the country-level random efects reflects the extent of heterogeneity in the baseline financial security level across countries after controlling for digitalization. The estimated standard deviation of the random efects (0.19) is relatively modest compared to the residual variance parameter (1.69), suggesting a moderate cross-country variation in baseline financial security. However, the wide credible interval [0.01, 0.52] indicates some uncertainty in this estimate.
The sigma parameter represents the unexplained variation in financial security not accounted for by the model. Its relatively high value (1.69) indicates substantial residual variability, likely due to the influence of unobserved factors not included in the model. This, again, aligns with the findings from the classical regression approach.
In general, the results of the Bayesian modeling are consistent with the direction of efects found in the classical panel data analysis but provide a more robust and informative estimation framework.
The formally derived model capturing the impact of digitalization on corporate sector financial security, based on Bayesian regression integrated with a panel data structure, can be expressed as follows: ∼ ( ; 1.692), = 2.07 − 0.03 · + , ∼ (0; 0.19 2). (11)
The identified negative relationship between digitalization and the financial security of the corporate
sector may be theoretically explained by several hypotheses:
• The Technological Risk Hypothesis. Adopting digital technologies is accompanied by new risks
cybersecurity threats, dependence on technological platforms, and data breaches - which may
adversely afect financial stability. For instance, the Global Initiative Against Transnational
Organized Crime, in its latest Global Organized Crime Index 2023, has for the first time included
cyber-dependent crimes in the composition of the Organized Crime Index [
Thus, the observed negative efect may stem from the transformational costs and emergent risks associated with the digitalization process. It is important to note that the presented findings reflect shortand medium-term efects, while the long-term consequences of digitalization for corporate financial security may difer significantly.
Based on the results of the conducted study, several practical recommendations can be formulated. For enterprises, it is advisable to: • Develop comprehensive digital transformation strategies that explicitly incorporate risk management components • Implement digital solutions in a phased manner, accompanied by regular assessments of their impact on financial performance • Maintain a balanced approach to digital investment by aligning transformation eforts with the need to ensure financial stability For investors, the following considerations are particularly relevant: • Take into account the digital maturity indicators of companies when evaluating their investment attractiveness • Conduct thorough analyses of the relationship between digital transformation expenditures and ifnancial resilience indicators From the perspective of regulators, it is recommended to: • Foster an enabling regulatory environment for digital transformation that simultaneously accounts for potential financial security risks • Develop monitoring tools and early warning systems aimed at identifying emerging threats to ifnancial security arising from digitalization • Promote advancing digital competencies and infrastructure to support sustainable and secure digital development
When interpreting the findings of this study, it is essential to consider several methodological limitations that define the scope of generalizability and outline potential avenues for further scientific inquiry.
The primary limitation of this research lies in the relatively small panel dataset, which includes only 27 countries over an 8-year observation period. This characteristic of the sample imposes significant constraints on the statistical power of the analysis and reduces the robustness of the inferential procedures. In examining the relationship between digitalization and the financial security of the corporate sector, the limited temporal horizon is particularly critical, as it hampers the ability to assess potential lagged efects of digital transformation adequately.
By nature, digitalization represents a complex, long-term process, the outcomes of which may materialize over various time horizons. While short-term efects may reflect adverse outcomes due to the need for substantial upfront investments and transformation-related costs, medium- and long-term impacts are potentially positive, stemming from enhanced eficiency, streamlined business processes, and the emergence of new development opportunities. The available 8-year period - determined by changes in DESI methodology - appears insuficient to fully capture the entire spectrum of temporally distributed efects, particularly given the varying pace of digital transformation across countries.
Moreover, the limited cross-sectional dimension of the dataset (27 countries) reduces the capacity to identify and control for heterogeneity in the impact of digitalization across economies with different levels of development, institutional environments, and corporate sector structures. A larger sample size would enable more detailed segmentation and allow for cluster-based or diferentiated analysis of countries, which might reveal divergent digitalization efects depending on specific national characteristics.
These identified limitations highlight the need to expand future research longitudinally, to account for lagged efects, and cross-sectionally, to increase the sample’s representativeness and facilitate a more granular investigation of heterogeneity in the observed relationships.
The empirical findings of this study reveal a statistically significant negative association between the level of digitalization and indicators of financial security in the corporate sector. This relationship remains robust across diferent estimation methods and appears stronger for firms with higher baseline levels of financial security.
However, this negative association should not be interpreted as an argument against digitalization. Instead, it should be viewed as evidence of the need for a comprehensive approach to managing digital transformation, which explicitly considers the potential risks to financial stability. Formulating efective digitalization strategies must incorporate mechanisms for monitoring and managing financial risks and adaptive changes in business models and organizational structures to ensure that the benefits of digital transformation do not come at the expense of corporate financial resilience.
The authors have not employed any Generative AI tools.