Digital Production Investments and Financial Outcomes: A Baltic and Rest of Europe Comparison

Abstract

This study provides new evidence on how production digitalization investment affects firm financial performance across diverse European regions. A panel of 14,935 firm-year observations from 30 countries (2012–2022), including a focused Baltic subsample, is used alongside a refined digital capital intensity metric based on depreciated plant and machinery value. The results indicate a positive association between digital investment and operating revenue across Europe, with significantly stronger effects observed in the Baltic region. Interaction models reveal higher marginal returns for Baltic firms, suggesting that digital capital delivers amplified value in economies with lower digital saturation but greater absorptive urgency. Employee-related costs consistently predict revenue outcomes, underscoring their role in translating digital assets into performance. Intangible fixed assets exhibit a positive impact in Baltic labor-scale models but weaker effects elsewhere, indicating that institutional maturity mediates knowledge capital productivity. Implications: (1) digital investment yields asymmetric returns; (2) workforce investment enhances digital ROI; and (3) policy should prioritize organizational readiness alongside infrastructure. This study contributes by introducing a replicable proxy for production-level digitalization and by providing rare comparative evidence on digital returns in transitional versus mature European economies.

1. Introduction

Digitalization has become a critical driver of corporate performance and industrial transformation across sectors and regions development (Tang et al., 2025; Bhatia et al., 2024; Nasiri et al., 2020; Horvat et al., 2019, 2018). Through the integration of robotics, smart production systems, artificial intelligence, and data analytics, firms are reshaping their operational and strategic capabilities (Varriale et al., 2025; Ogbeibu et al., 2024; Cette et al., 2022). The role of production digitalization in driving efficiency, responsiveness, and revenue generation has been increasingly emphasized in both academic and policy debates (Zaman et al., 2025; Cette et al., 2022; Peng & Tao, 2022). However, despite substantial investment in digital technologies, the financial outcomes of these investments remain questionable. While a number of empirical studies confirm a positive relationship between digital investment and firm-level financial performance (Zaman et al., 2025; Lastauskaite & Krusinskas, 2024; Ghosh et al., 2022), others identify a lag in financial returns, insufficient internal capabilities, or systemic inefficiencies—phenomena often described as the digitalization paradox (C. Yang et al., 2025; Zeng et al., 2022; Guo et al., 2023; Gebauer et al., 2020; Kohtamäki et al., 2020).

A persistent limitation in the existing literature concerns the inconsistent and conceptually narrow measurement of digital investment. Prior empirical studies have often operationalized digitalization using proxies such as the adoption of specific technologies, including cloud computing, big data analytics, and robotics, which typically are captured through surveys or categorical technology classifications (Cette et al., 2022; Horvat et al., 2019). While these indicators provide insight into the presence of digital tools, they inadequately reflect the scale, financial commitment, and embedded nature of digital transformation within production systems. In particular, they tend to overlook capital expenditures associated with physical assets that now routinely integrate digital functionalities.

Alternative approaches, including digital adoption indices, digital service expenditures, and self-reported usage metrics, are likewise constrained by methodological shortcomings. These include reliance on perception-based data, susceptibility to response bias, and limited comparability across firms, industries, and national contexts (Zeng et al., 2022; Kohtamäki et al., 2020). Furthermore, such measures typically capture nominal access or availability rather than the intensity or strategic depth of digital investment.

This study addresses these limitations by proposing a capital-based proxy for production-level digitalization, grounded in financial accounting data. Specifically, changes in plant and machinery (P&M) value, net of depreciation, are employed to reflect firms’ investment in digitally embedded physical assets. Although the P&M category in the Orbis database does not exclusively capture digital technologies, contemporary production equipment increasingly incorporates automation, connectivity, and intelligent control systems as standard features. As such, this proxy offers a conceptually grounded and empirically robust measure of digital capital deepening. It enhances analytical precision, supports cross-country and longitudinal comparisons, and better aligns with theoretical frameworks that link digitalization to firm-level productivity and financial performance.

The validity of the proposed proxy is further supported by several robustness checks conducted in the empirical analysis. These include alternative normalization strategies (by total assets and number of employees), fixed effects for year, country, and industry to control for unobserved heterogeneity, and the use of the RESET test to assess model specification. Additionally, Variance Inflation Factors (VIFs) confirm the absence of multicollinearity among explanatory variables, and alternative dependent variables (e.g., ROA, net profit) were tested to verify the consistency of the results. Interaction terms were also employed to examine regional moderating effects, reinforcing the contextual sensitivity of digital investment returns.

Another persistent limitation is the insufficient attention to regional heterogeneity in digital returns. The impact of production digitalization is not uniform and is often shaped by contextual factors such as national digital infrastructure, labor market readiness, and historical development trajectories (Jovanović et al., 2018). This study focuses on the Baltic region as a key comparison group because it represents a transitional economic bloc characterized by strong digital ambitions but institutional and structural constraints that differ from mature European economies. The Baltic region offers a unique empirical setting for testing regional variation. Following decades of technological stagnation during the Soviet era, these economies have undergone accelerated digital adaptation post-independence, resulting in high digital ambition but uneven firm-level digital maturity (Rindzevičiūtė, 2021). In contrast, Western European countries experienced more gradual and stable digital growth. This divergence offers a strong foundation for examining whether returns on digital investments differ systematically across regions, with particular interest in transitional economies such as the Baltics, where the marginal gains from digital investment may be amplified due to structural catch-up effects.

Accordingly, the core objective of this paper is to explain the financial performance differences associated with digital production investment across European regions. This study particularly seeks to understand how these effects manifest in the Baltic region compared to the broader European landscape. This study investigates the relationship between production digital investment and corporate financial performance, with a comparative focus on the Baltic region and the rest of Europe. Specifically, this research explores how capital expenditures in digitally embedded production assets affect operating revenue, and whether this effect is moderated by region, firm size, and complementary expenditures such as employee-related costs and intangible fixed assets. The empirical analysis is based on a firm-level dataset of 14,935 observations from the ORBIS database covering the period 2012–2022, including 126 observations from the Baltic states. To capture scale-adjusted performance, all financial variables are normalized by total assets and number of employees, and the regression analysis incorporates fixed effects for year, country, and sector to control unobserved heterogeneity.

This study makes three contributions to the literature. First, it introduces a novel method for quantifying production digital investment using balance-sheet asset data, offering a replicable and scalable alternative to survey-based or technology-specific indicators. Second, it provides firm-level comparative evidence on the digitalization–performance nexus in the structurally under-researched Baltic region, set against the broader European context. Third, it demonstrates that the performance impact of digital capital is stronger in transitional economies due to latent efficiency gaps and technology catch-up dynamics, and it highlights the complementary role of employee-related costs in enhancing digital returns.

This analysis is particularly timely given the acceleration of digital transformation following the COVID-19 pandemic. Many firms, driven by operational constraints and market shifts, increased their digital investment without adequate preparation or long-term strategic alignment (Amankwah-Amoah et al., 2021; Sostero et al., 2020). Understanding whether and under what conditions these investments yielded measurable financial benefits is essential for both managerial decision-making and public policy design.

The remainder of this paper is structured as follows: Section 2 reviews the literature and hypotheses’ development; Section 3 outlines the data and methodological approach; Section 4 presents the empirical results; Section 5 reports the conclusion with a discussion of key implications and directions for future research.

2. Literature Review and Hypotheses’ Development

Digital investment in production assets significantly enhances business financial performance (Zaman et al., 2025; Lastauskaite & Krusinskas, 2024; Liu et al., 2023; Ghosh et al., 2022). The integration of digital technologies streamlines operations and fosters innovation, leading to improved financial outcomes for business. According to Toumia et al. (2023), digital competitiveness positively influences financial performance, particularly in the post-COVID-19 era, where firms that adopted digital tools saw improved returns on assets. Digital investment, encompassing data-driven digital technologies like analytics, artificial intelligence (AI), big data, and cloud computing, is widely believed to increase firms’ growth opportunities and productivity (Cheng et al., 2025; Teng et al., 2022).

However, Chen and Srinivasan (2024) show some evidence of interim productivity increases (e.g., higher ROA and asset turnover for firms engaging in digital activities), there is limited evidence supporting improved long-term productivity, with profit margins showing no significant differences and sales growth sometimes declining. Recent empirical studies reinforce and deepen this narrative, where Lastauskaite and Krusinskas (2024) demonstrate that production digitalization investments significantly improve operating revenue in European manufacturing, with large and Eastern European firms showing the strongest returns. Meanwhile, a 2025 Chinese panel study finds that digitalization bolsters fixed-asset turnover and enhances operational efficiency, though its effect on overall profitability remains contingent on firm-level managerial capabilities and absorptive capacity (Benedek et al., 2025).

These findings suggest that digital investment enhances firm performance through mechanisms such as productivity gains, scale efficiency, and innovation enablement. However, the realized financial outcomes are conditional upon the firm’s ability to absorb, adapt, and strategically integrate digital technologies. This interpretation aligns with recent theoretical developments in digital absorptive capacity and dynamic capability frameworks, which emphasize the importance of internal readiness and process integration (Filatotchev et al., 2025; Zeng et al., 2022).

Based on the above evidence, the following hypothesis is proposed:

H1a. 

Digital investment in production assets has a positive impact on business financial performance.

While digitalization is broadly associated with performance gains, its impact is not uniform across countries or regions. The impact is influenced by factors such as industrial structure and regional economic conditions (Li et al., 2024). Such structural and contextual differences shape how firms capitalize on digital investments, making it necessary to investigate regional variation. Differences in investment levels in digital assets among Europe countries (including Baltic Region) can suggest varying financial performance outcomes, highlighting the need for tailored strategies to enhance digital competitiveness in each region. Recent empirical evidence indicates significant regional heterogeneity. A 2024 study examining China’s industrial digitalization found that digital investment boosted industrial efficiency and green innovation most strongly in less developed central and western provinces, reflecting the moderating role of industrial structure and infrastructure capacity (Wang et al., 2024). In the European context, similar regional asymmetries have been identified, particularly between Western and Eastern member states, where disparities in digital infrastructure, workforce digital skills, and institutional capacity affect outcomes. This pattern is consistent with emerging evidence on catch-up dynamics and regional digital convergence, whereby structurally lagging regions derive higher marginal returns due to larger initial efficiency gaps and technology diffusion potential (Jiang et al., 2025; Tang et al., 2025). The Baltic region, marked by digital ambition but uneven institutional development, exemplifies such transitional dynamics.

Focusing on the Baltic region, a DESI-based assessment highlights persistent gaps in digital maturity across Estonia, Latvia, and Lithuania, with implications for uneven corporate adoption of digital production assets and varied returns (Česnauskė, 2019). Another study by Eremina et al. (2019) indicate that digital investment, or digital maturity, in the Baltic region indicates a positive relationship with several financial indicators, notably sales growth, return on equity (ROE), and gross profit over assets. However, cross-country differences in institutional support mechanisms and digital absorption capacities may explain why performance gains are not equally distributed.

Further supporting this contextual interpretation, Scalamonti (2024) emphasizes that governance quality and socio-political institutions in transitional economies play a decisive role in shaping developmental outcomes and digital transformation trajectories. Complementary findings by But et al. (2024) illustrate how sectoral upgrading, such as in Czechia’s health tourism sector, depends on institutional coordination and policy alignment, factors equally relevant in the Baltic digitalization context.

These findings point to the importance of geographical and institutional context when assessing the link between digital investment and firm-level outcomes. In particular, the Baltic region, with its mixed levels of digital readiness, offers a valuable empirical setting to examine how location moderates the performance impact of digital transformation. Accordingly, the following hypothesis is proposed:

H1b. 

The performance impact of digital investment is stronger in the Baltic region than in the rest of Europe.

Employee-related costs as wages, training, and employee welfare are increasingly seen as strategic investments that enhance business financial performance (Wen et al., 2022). Recent studies emphasize that these expenditures contribute to productivity, innovation, and efficiency, especially when aligned with digital transformation and human capital development (Liang et al., 2023).

Guo et al. (2023) found that digital transformation, though resource-intensive, improves firm performance when supported by investment in employee capabilities. Similarly, Murugesan et al. (2023) demonstrated that AI-driven HR digitalization enhances workforce efficiency, strengthening financial outcomes. Liang et al. (2023) further showed that employee welfare is significantly linked to better financial performance, particularly in developing regions where labor market responsiveness is high. These results are consistent with the emerging view that human capital acts as a critical enabler of digital value realization, especially in contexts where operational agility and technology absorption are pivotal (Jiang et al., 2025).

However, the magnitude of these effects may differ across regional contexts due to institutional maturity, wage structures, and labor adaptability. Specifically, firms operating in transitional economies often face more pronounced labor constraints, which makes targeted employee investment a performance enhancer and a structural necessity (Setyadi et al., 2025). In particular, the Baltic region, characterized by smaller firms and transitional economies, may exhibit stronger performance gains from labor investments. Accordingly, the following hypotheses are proposed:

H2a. 

Employee-related costs have a positive impact on business financial performance.

H2b. 

The positive effect of employee-related costs on business financial performance is stronger in the Baltic region than in the rest of Europe.

Intangible fixed assets, such as patents, software, trademarks, organizational copyrights, licenses, and employee know-how, play a critical role in enhancing business financial performance (Purnamawati et al., 2022). Firms that invest in intangible capital often benefit from improved productivity, profitability, customer loyalty, and adaptability in competitive markets (Corrado et al., 2022; Lopes & Carvalho, 2021).

The strength of the relationship between intangible assets and business financial performance varies across regions and countries (Teng et al., 2022; Büchi et al., 2020). The difference arises from the maturity of innovative ecosystems (Leceta & Könnölä, 2021), institutional infrastructure, and capital access in Western and Central Europe, which allow firms to leverage intangible assets more effectively. In contrast, transitional economies frequently encounter barriers such as weak intellectual property enforcement, fragmented innovation networks, and limited commercialization pathways, all of which can dilute the effectiveness of intangible capital investments (Scalamonti, 2024). Consequently, while intangibles may still contribute positively in these regions, their returns are often delayed or conditional upon complementary institutional reforms and market sophistication (But et al., 2024).

Accordingly, the following hypotheses are proposed:

H3a. 

Intangible fixed assets have a positive impact on business financial performance.

H3b. 

The impact of intangible fixed assets on business financial performance is stronger in the European market compared to the Baltic region.

3. Data and Methodology

For the analysis, company-level data sourced from the Orbis database were utilized, comprising a sample size of 14,935 observations, with 126 observations representing businesses in the Baltic region. The analysis was conducted using STATA statistical software, where the data were thoroughly cleaned, and regression models were calculated to evaluate the impact of production digitalization investment. The dataset was aggregated and winsorized at the 1st and 99th percentiles to mitigate the effects of outliers and ensure robust results. To clarify the sampling process, the Orbis database was used in full scope for all 30 European countries available in its coverage. No sectoral restrictions were applied, meaning that firms across industries were included as long as they were incorporated corporate entities. The time horizon (2012–2022) reflects the maximum span of consistently reported financial data. Observations with missing values for the main variables (plant and machinery, depreciation, employee-related costs, and intangible assets) were excluded.

The dataset was compiled from the Orbis database, covering all available firm-year observations across 30 European countries between 2012 and 2022. Firms of all sizes and sectors were included, provided they reported complete financial data on key variables relevant to production digitalization as plant and machinery (P&M), depreciation, employee-related costs, and intangible fixed assets. After excluding entries with missing data, the final sample consists of 14,935 firm-year observations representing 2852 unique firms. Of these, 126 observations pertain to firms located in the Baltic region (Estonia, Latvia, and Lithuania).

While the Baltic subsample is relatively small, it was retained due to its analytical relevance for examining regional heterogeneity in digital investment outcomes. To address concerns about statistical power, the analysis avoids standalone subsample regressions and instead incorporates interaction terms within the full sample. This design allows for meaningful comparisons based on the direction and magnitude of effects, without over-relying on statistical significance or overgeneralizing from a limited regional subset.

The data spans a decade (2012–2022), providing a basis for analyzing long-term trends. This period also includes the COVID-19 pandemic, which played a crucial role in accelerating digitization trends (Amankwah-Amoah et al., 2021; Pinzaru et al., 2020; Pira & Fleet, 2025; Setyadi et al., 2025).

The selection of variables was guided by the existing literature on assessing the impact of digitalization and digital transformation on business financial performance (Jiang et al., 2025; Filatotchev et al., 2025; Zeng et al., 2022; Sanchez-Riofrio et al., 2022). The definitions of variables, along with their calculation components, are provided in Appendix A. In this research, production digitalization influence business performance through profit margins or variables such as gross profit, operating revenue, net income, or return on assets and others (Eyvazi et al., 2025; Heredia et al., 2022; Guo et al., 2023; Martín-Peña et al., 2019). At the outset of the analysis, constructing the regression equation required systematically testing multiple configurations of the chosen variables to identify the optimal model. The selection process was guided by key criteria, including statistical significance, explanatory strength, and the mitigation of multicollinearity among independent variables. Various dependent variables were examined across different model specifications, including return on assets, asset turnover ratio, net profit, earnings before interest and taxes, profit or loss before tax, net sales and operating revenue.

Among the possible financial performance indicators (e.g., return on assets, asset turnover ratio, net profit, EBIT), operating revenue was selected as the primary dependent variable because it aligns closely with the theoretical focus on value generation from digital capital. Prior research highlights operating revenue as a sensitive measure of firms’ ability to translate technological investment into market performance (e.g., Martín-Peña et al., 2019; Ghosh et al., 2022). Alternative specifications with profitability and return-based measures were also estimated as robustness checks; while the results were generally consistent in direction, operating revenue provided the most stable and interpretable outcomes across models.

Digital investment in production assets (DA) is the main independent variables in the analysis expressed as follows:

$${DA}_{i,t}={P\&M}_{i,t}-\left({P\&M}_{i,t-1}-{P\&MDEP}_{i,t}\right)$$

(1)

where P&M is Plant and Machinery value in thousands of Euros for an individual company i and time period t; P&MDEP represents Plant and Machinery Depreciation value in thousands of Euros. It is acknowledged that the P&M category is not exclusively digital. However, consistent with recent research (G. Yang et al., 2023; Cette et al., 2022; Horvat et al., 2019), production asset investment is treated as a digitalization proxy because digital functionalities are increasingly embedded in physical capital. This provides a replicable, scalable alternative to survey-based indicators that often lack comparability across countries and years.

The objective of quantifying digital investment in production assets presents inherent challenges, particularly due to the difficulty in obtaining sensitive financial data from firms. This study goes beyond merely evaluating digitalization initiatives by systematically measuring the actual capital allocated to plant, equipment, and machinery acquisitions, as well as corresponding changes in asset depreciation values. This research underscores those modern long-term investments in fixed assets, such as production machinery and industrial equipment, are increasingly aligned with digital transformation objectives. This study explicitly focuses on capital expenditures related to physical hardware and production technology, deliberately excluding expenses related to equipment leasing or workforce training for digital skill development.

This study investigates the impact of digital investment in production assets on firms’ financial performance. Although the dataset exhibits a panel structure, pooled OLS with fixed controls for year, industry, and country is employed as the baseline specification. Coefficient estimates across pooled OLS, fixed effects, and random effects models remain consistent in both direction and magnitude, supporting the robustness of the chosen approach.

To further address potential endogeneity concerns, such as reverse causality between digital investment and firm performance, additional model specifications were tested during the model setup phase. Specifically, lagged digital investment variables were used as regressors to explore the temporal direction of causality. The results from these lagged models were directionally consistent with the main specifications and produced comparable coefficients, supporting the validity of the contemporaneous approach. However, due to slightly lower explanatory power and similar patterns, these models were not retained in the final regression tables. Moreover, fixed-effects estimators and random-effects models were applied as robustness checks to account for time-invariant unobserved heterogeneity. While the baseline relies on pooled OLS with fixed controls, these supplementary models confirmed that the main findings are stable across different estimation strategies.

The analysis is based on the following ordinary least squares (OLS) regression models: regression function (2) serves as the basis for Models 1, 2, and 5, while regression function (3) underpins Models 3, 4, and 6:

$$\begin{gathered} {OP}_{i,t}=\alpha +{\beta }_1{DA}_{i,t}+{\beta }_2{INTF}_{i,t}+{\beta }_3{\mathrm{R}\&D}_{i,t}+{\beta }_4{\mathrm{C}\mathrm{O}\mathrm{S}\mathrm{T}}_{i,t}+{\beta }_5{\mathrm{N}\mathrm{E}\mathrm{T}\mathrm{P}}_{i,t}+{\delta }_1{LOGN}_{i,t}+{\gamma }_1{COUNTRY}_i+{\gamma }_2{YEAR}_t \\ +{\gamma }_3{SECTOR}_i+{\epsilon }_{i,t} \end{gathered}$$

(2)

$$\begin{gathered} {OP}_{i,t}=\alpha +{\beta }_1{DA}_{i,t}+{\beta }_2{INTF}_{i,t}+{\beta }_3{\mathrm{R}\&D}_{i,t}+{\beta }_4{\mathrm{C}\mathrm{O}\mathrm{S}\mathrm{T}}_{i,t}+{\beta }_5{\mathrm{N}\mathrm{E}\mathrm{T}\mathrm{P}}_{i,t}+{\beta }_6{\mathrm{T}\mathrm{A}\mathrm{N}\mathrm{F}}_{i,t}+{\delta }_1{LOGN}_{i,t}+{\gamma }_1{COUNTRY}_i \\ +{\gamma }_2{YEAR}_t+{\gamma }_3{SECTOR}_i+{\epsilon }_{i,t} \end{gathered}$$

(3)

It was chosen to divide all variables by number of employees per business unit (N) in model 1, 2, 5 and by the total assets (T) in model 3, 4, 6. In line with this methodology, all variables in the result tables are expressed as ratios. This is a widely used practice in research for evaluating corporate financial performance (Cappa et al., 2021), as it provides a standardized scale that facilitates comparisons across companies of different sizes. By normalizing the variables, differences in firm size are effectively accounted for, improving the clarity and interpretability of the regression results.

When variables are normalized by the number of employees in a company (N), the resulting ratios are widely recognized in research as productivity indicators (Guo et al., 2023). This approach facilitates the assessment of how efficiently human capital is utilized in generating value. It offers insights into a firm’s labour efficiency, highlighting the extent to which employees contribute to overall performance.

Similarly, when variables are normalized by total assets (T), they provide a broader perspective on the company’s asset structure and financial health. Ratios based on total assets help evaluate the capital intensity of a business.

Furthermore, the models incorporated fixed effects for country, year, and industry to account for unobserved heterogeneity across these dimensions. To address within-group correlation—where observations from the same firm over time may be interdependent—robust standard errors clustered at the firm-level were applied. OLS was retained as the baseline estimator due to limited within-firm variation in some variables (particularly in the Baltic subsample), while robustness checks with fixed and random effects yielded consistent coefficients. The selection of the regression method was guided by the studies of Zeng et al. (2022), Guo et al. (2023), and Sanchez-Riofrio et al. (2022).

To enhance the model’s robustness and identify potential nonlinearities, the Regression Specification Error Test (RESET) was performed. Additionally, to assess potential multicollinearity, Variance Inflation Factors (VIFs) were computed for the key explanatory variables. The results indicated that all VIF values were below 3, suggesting that multicollinearity is not a concern in the regression model.

Finally, for robustness check of the model, the Models 5 and 6 include interaction terms between the regional dummy variable and the digital investment in production assets variable. This methodological approach aims to account for variations across different regions, recognizing that regional-specific factors may significantly influence the relationship between digital investment and financial outcomes.

4. Empirical Results

The empirical results highlight three central findings. First, digital investment has a significant positive impact on firm performance in Europe overall, while in the Baltic subsample, the effect is larger in magnitude though not always statistically significant, likely due to sample size constraints. This finding should be interpreted with caution, as the relatively small Baltic subsample (N = 126) reduces the statistical power of regression estimates. Nonetheless, the larger coefficient size suggests potential economic relevance, especially in transitional settings where digital capital may yield higher marginal returns.

Second, employee-related costs consistently emerge as the strongest and most robust predictor across all model specifications, underscoring the complementarity between human capital and digital capital. This indicates that human resource intensity, particularly in skilled or technical roles, likely amplifies the productivity of digital assets. The finding is especially pronounced in the Baltic context, where smaller firm sizes may associate with more visible gains from labor-intensive digitalization.

Third, intangible assets exhibit more conditional effects, with their influence varying across specifications, suggesting context-dependent roles in translating digital investment into performance. In the Baltics, intangible capital appears to contribute more meaningfully to outcomes when normalized per employee, likely due to the direct role of knowledge and intellectual property in labor productivity. In Europe overall, the effect is mixed, hinting at differing capabilities to monetize or integrate intangible assets.

Taken together, the results point to important substantive differences in how digital production investment interacts with other firm-level inputs across regions. While Europe shows more stable significance across variables, the Baltic region reveals patterns of potentially high but uneven returns, shaped by firm size, labor structure, and stages of digital maturity. The detailed regression outputs reported below provide further evidence and robustness checks that support these core findings.

Table 1 and Table 2 provide an overview of the descriptive statistics for the full sample (N = 14,935), as well as for the Baltic region subsample (N = 126), under two normalization strategies: by total assets and by number of employees. When normalized by total assets (Table 1), firms in the Baltic region exhibit notably lower average levels of operating revenue (mean = 0.556) relative to their European counterparts (mean = 0.653), indicating comparatively lower revenue generation efficiency per unit of asset base. However, their digital investment intensity (DA) is marginally higher (mean = 0.264 vs. 0.242), suggesting that Baltic firms may be allocating proportionally more of their capital stock toward modernizing production infrastructure.

Table 1. Variables definition and descriptive statistics (normalized by total assets).

		OP	DA	INTF	R&D	COST	NETP	LOGN
Baltic	Obs.	126	126	126	126	126	126	126
	Mean	0.556	0.264	0.105	0.010	0.158	0.120	1.625
	Median	1.033	0.330	0.056	0.001	0.219	0.016	5.966
	Std. Dev.	0.992	0.234	0.008	0	0.182	0.0313	6.163
	Min	0.056	0.007	0	0	0.014	−0.538	0.693
	Max	2.984	1.237	0.561	0.101	0.948	0.270	8.897
Europe	Obs.	14,809	14,809	14,809	14,809	14,809	14,809	14,809
	Mean	0.653	0.242	0.186	0.047	0.178	0.197	2.494
	Median	0.910	0.206	0.155	0.015	0.219	−0.028	6.022
	Std. Dev.	0.828	0.115	0.079	0	0.178	0.021	5.983
	Min	0.002	0.0003	0	0	0.0071	−1.064	0.693
	Max	3.633	1.237	0.772	0.342	1.055	0.270	11.812
Total	Obs.	14,935	14,935	14,935	14,935	14,935	14,935	14,935
	Mean	0.652	0.243	0.186	0.047	0.177	0.197	2.488
	Median	0.911	0.207	0.155	0.015	0.219	−0.028	6.022
	Std. Dev.	0.830	0.116	0.078	0	0.178	0.021	5.983
	Min	0.002	0.0003	0	0	0.007	−1.064	0.693
	Max	3.633	1.237	0.772	0.342	1.055	0.270	11.812

Source: Authors’ calculations based on data from the ORBIS database. Note: The variables in this table have been adjusted to reflect relative size metrics. For instance, ‘Operating Revenue’ is expressed as ‘Operating Revenue per Total Assets.’ All values have been standardized by dividing each figure by the respective company’s total assets.

Table 2. Variables definition and descriptive statistics (normalized by number of employees).

		OP	DA	INTF	R&D	COST	NETP	TANF	LOGN
Baltic	Obs.	126	126	126	126	126	126	126	126
	Mean	75.651	52.817	18.761	2.725	20.396	5.299	101.997	1.625
	Median	111.729	40.230	8.240	0.477	18.209	4.051	28.424	5.966
	Std. Dev.	105.094	19.686	0.886	0	4.438	−57.782	1.926	6.163
	Min	12	1.214	0	0	51.857	96.053	1140.194	0.693
	Max	350.196	315.325	124.073	25.181	0.630	1.321	3.843	8.897
Europe	Obs.	14,809	14,809	14,809	14,809	14,809	14,809	14,809	14,809
	Mean	303.132	96.756	145.846	34.710	35.447	105.964	162.599	2.494
	Median	242.640	56.474	65.600	8.045	48.971	−16.950	85.965	6.022
	Std. Dev.	168.596	23.147	15.985	0	45.721	3.454	35.757	5.983
	Min	2.192	0.133	0	0	4.438	−662.833	0.250	0.693
	Max	2163.919	637.418	1029.017	269.545	201.454	200.775	1140.194	11.812
Total	Obs.	14,935	14,935	14,935	14,935	14,935	14,935	14,935	14,935
	Mean	302.167	96.480	145.334	34.571	35.407	105.552	163.401	2.488
	Median	241.536	56.337	65.116	7.981	48.730	−16.762	86.100	6.022
	Std. Dev.	167.763	23.133	15.684	0	45.381	3.461	35.673	5.983
	Min	2.192	0.133	0	0	4.4388	−662.833	0.250	0.693
	Max	2163.919	637.418	1029.017	269.545	201.454	200.775	1140.194	11.812

Source: Authors’ calculations based on data from the ORBIS database. Note: The variables in this table have been adjusted to reflect relative size metrics. For instance, ‘Operating Revenue’ is expressed as ‘Operating Revenue per Number of Employees.’ All values have been standardized by dividing each figure by the respective company’s number of employees.

Intangible fixed assets (INTF), a key proxy for innovation and knowledge-based capital, are substantially lower in the Baltic sample (mean = 0.105) compared to the European average (mean = 0.186), which may reflect differences in innovation ecosystems or intellectual property infrastructure. The variable employee-related costs (COST) presents a comparable mean across groups (~0.158 in the Baltics vs. 0.178 in Europe), but the distribution in the Baltic sample appears more dispersed, implying that a few firms may be responsible for significantly higher labor costs than the rest. Notably, the corporate size variable (LOGN), measured as the log of number of employees, is lower in the Baltic group (mean = 1.625) than in Europe (mean = 2.494), reflecting the smaller scale of operations.

Table 2 reports the same set of variables normalized by number of employees, which allows fir assessment of labour productivity and capital intensity.

Baltic firms report a mean operating revenue per employee of 75.65, which is approximately one-quarter the European average (303.13). Likewise, average intangible assets per employee are considerably lower in the Baltic region (18.76) compared to the rest of Europe (145.85), reinforcing the prior observation of lower intangible capital formation. The mean digital investment per employee (DA = 52.82 in Baltics vs. 96.76 in Europe) is also lower in absolute terms, though the relative capital deepening observed in Table 1 persists. However, digital investment in the Baltic sample displays substantial dispersion, with a small number of firms reporting disproportionately high values. This indicates a divergence in adoption levels, where digital modernization is concentrated among a limited set of firms.

Interestingly, despite lower firm size and resource intensity, net profit per employee is positive in the Baltic sample (mean = 5.30), while the European mean (105.96) masks a distribution with extreme outliers, possibly reflecting divergent cost structures and profit models.

Across all variables, Baltic firms show wider dispersion and more pronounced skewness and kurtosis, especially for R&D and cost-related measures. This supports the hypotheses that the region is characterized by greater heterogeneity and volatility in digital and financial performance indicators, possibly due to smaller market sizes, uneven policy support, or varying stages of digital transformation.

The correlation matrices (Table 3 and Table 4) offer preliminary insights into the relationships among the core variables used in the regression models. Across the full sample normalized by total assets (Table 3), the dependent variable Operating Revenue (OP) indicates a positive and statistically significant correlation with digital investment (DA) (r = 0.113, p < 0.01), lending initial support to hypothesis H1a. OP is also strongly and positively correlated with employee-related costs (COST) (r = 0.410, p < 0.01), aligning with H2a. Intangible fixed assets (INTF) exhibit a weak negative correlation with OP (r = −0.116, p < 0.01), suggesting that the performance gains from intangible capital may require complementary capabilities or longer-term horizons.

Table 3. The Pearson Correlation Matrix for variables used in regression analysis (normalized by total assets).

Baltic Region
No.	Var.	1	2	3	4	5	6	7
1	OP	1.000
2	DA	0.163 *	1.000
3	INTF	0.085	−0.328 ***	1.000
4	R&D	−0.048	0.046	−0.054	1.000
5	COST	0.728 ***	0.168 *	0.017	−0.133	1.000
6	NETP	−0.060	−0.098	−0.170 *	0.123	−0.272 ***	1.000
7	LOGN	0.194 **	−0.424 ***	−0.032	−0.323 ***	0.041	0.294 ***	1.000
Europe Region
No.	Var.	1	2	3	4	5	6	7
1	OP	1.000
2	DA	0.113 ***	1.000
3	INTF	−0.116 ***	−0.239 ***	1.000
4	R&D	−0.152 ***	−0.060 ***	0.047 ***	1.000
5	COST	0.410 ***	0.103 ***	0.068 ***	0.061 ***	1.000
6	NETP	0.293 ***	0.074 ***	−0.105 ***	−0.307 ***	−0.145 ***	1.000
7	LOGN	0.222 ***	0.157 ***	0.102 ***	−0.038 ***	0.054 ***	0.391 ***	1.000

Source: Authors’ calculation using data from ORBIS database. Note: *** p < 0.01, ** p < 0.05, * p < 0.1.

Table 4. The Pearson Correlation Matrix for Europe region (normalized by number of employees).

Baltic Region
No.	Var.	1	2	3	4	5	6	7	8
1	OP	1.000
2	DA	0.486 ***	1.000
3	INTF	0.384 ***	−0.079	1.000
4	R&D	0.314 ***	0.307 ***	−0.015	1.000
5	COST	0.669 ***	0.323 ***	0.427 ***	0.075	1.000
6	NETP	0.461 ***	0.555 ***	−0.303 ***	0.280 ***	0.190 **	1.000
7	TANF	0.483 ***	0.746 ***	−0.050	0.113	0.440 ***	0.652 ***	1.000
8	LOGN	−0.022	−0.365 ***	−0.145 *	−0.342 ***	−0.079	0.134	0.036	1.000
Europe Region
No.	Var.	1	2	3	4	5	6	7	8
1	OP	1.000
2	DA	0.338 ***	1.000
3	INTF	0.188 ***	0.055 ***	1.000
4	R&D	0.002	−0.003	0.156 ***	1.000
5	COST	0.286 ***	0.159 ***	0.347 ***	0.438 ***	1.000
6	NETP	0.232 ***	0.096 ***	−0.181 ***	−0.459 ***	−0.419 ***	1.000
7	TANF	0.296 ***	0.577 ***	0.036 ***	−0.031 ***	0.057 ***	0.071 ***	1.000
8	LOGN	0.057 ***	0.105 ***	−0.063 ***	−0.100 ***	−0.101 ***	0.343 ***	−0.038 **	1.000

Source: Authors’ calculation using data from ORBIS database. Note: *** p < 0.01, ** p < 0.05, * p < 0.1.

The Baltic subsample normalized by number of employees (Table 4) reveals stronger and more concentrated positive correlations. OP is highly correlated with both DA (r = 0.486, p < 0.01) and COST (r = 0.669, p < 0.01), supporting H1a and H2a with greater regional intensity—thus tentatively validating H1b and H2b. Notably, OP also correlates significantly with INTF (r = 0.384, p < 0.01), a contrast to the weak or negative association observed in the full sample. This implies that for smaller, digitally advancing economies, intangible capital may have a more direct linkage with financial outcomes.

For the European region, OP maintains a significant correlation with DA (r = 0.338, p < 0.01) and COST (r = 0.286, p < 0.01), but the association with INTF (r = 0.188, p < 0.01) is less pronounced than in the Baltic region.

Importantly, across all regions and specifications, the correlation coefficients among independent variables are generally modest. Variance Inflation Factors (VIFs) computed separately confirm that multicollinearity is not a concern in the model structure.

Table 5 presents the results of the baseline OLS regressions for the Baltic region (Models 1 and 3) and the broader European sample (Models 2 and 4), using two normalization strategies: by total assets (Models 1–2) and by number of employees (Models 3–4).

Table 5. Regression analysis results for Baltic and Europe Region.

Var.	Baltic Region Model 1 (Coeff. [Std. Error])	Europe Region Model 2 (Coeff. [Std. Error])	Baltic Region Model 3 (Coeff. [Std. Error])	Europe Region Model 4 (Coeff. [Std. Error])
DA	0.228 [0.174]	0.037 [0.043]	0.308 [0.239]	0.419 *** [0.085]
INTF	0.476 [0.674]	−0.594 *** [0.055]	1.246 ** [0.470]	0.214 *** [0.041]
R&D	2.175 [2.022]	−1.017 *** [0.125]	3.962 ** [1.670]	0.050 [0.144]
COST	1.951 *** [0.274]	1.649 *** [0.065]	3.519 *** [0.933]	3.638 *** [0.217]
NETP	0.021 [0.261]	0.945 *** [0.045]	1.364 *** [0.324]	1.133 *** [0.073]
TANF			−0.011 [0.054]	0.313 *** [0.051]
LOGN	0.105 *** [0.036]	0.024 *** [0.005]	8.648 [7.415]	−4.863 * [2.611]
CONST	0.008 [0.354]	0.558 *** [0.065]	−33.263 [64.544]	33.519 [27.412]
Model Summary
Observations	126	14,809	126	14,809
Ad. R-squared	0.768	0.431	0.722	0.365
RESET Test
SS	31.298	2738.951	551,426.205	488,055,104
df	21	47	22	48
MS	1.490	58.27	25,064.72	1,039,681.3
F	20.74	239.84	15.75	178.10
Prob > F	0.00	0.00	0.00	0.00
R-squared	0.8073	0.433	0.7708	0.3668
Adj R-squared	0.7683	0.431	0.7219	0.3647
Root MSE	0.2680	0.4929	39.898	241.61

Note: *** p < 0.01, ** p < 0.05, * p < 0.1. Source: Authors’ calculation based on data.

The effect of digital investment (DA) on operating revenue is positive but statistically insignificant in the Baltic region under both specifications (β = 0.228 in Model 1; β = 0.308 in Model 3), likely due to the small sample size (N = 126) and higher within-group variability. In contrast, the effect is positive and statistically significant in the European sample—particularly when normalized by number of employees (Model 4: β = 0.419, p < 0.01). This provides strong support for H1a, confirming that digital investment enhances financial performance across European firms, while H1b is partially supported, as the effect size is numerically higher in the Baltics, but not statistically robust.

The relationship between intangible fixed assets (INTF) and firm performance differs markedly across regions. In the Baltic region, INTF indicates a positive and statistically significant effect only in the employee-normalized model (Model 3: β = 1.246, p < 0.05), indicating that intangible fixed assets contribute meaningfully to labour productivity and financial outcomes when scaled to firm workforce. In contrast, Europe exhibits mixed results as INTF is negatively associated with performance in Model 2 (β = −0.594, p < 0.01) but becomes significantly positive in Model 4 (β = 0.214, p < 0.01). These findings partially confirm H3a, while H3b is not supported, as the Baltic region indicates stronger coefficients.

Employee-related costs (COST) have a strong and consistently positive effect on firm performance across all models and regions. The effect is statistically significant at the 1% level in both the Baltic region (β = 1.951 in Model 1; β = 3.519 in Model 3) and the European sample (β = 1.649 in Model 2; β = 3.638 in Model 4). Notably, the coefficients are larger in magnitude in the Baltic models, lending clear empirical support for both H2a and H2b. This underscores the strategic role of human capital investment, especially in smaller, transitional economies where labor-related investments are more impactful relative to firm scale.

Net profit (NETP) is significant only in the European sample, while its effect is weak or statistically insignificant in the Baltics. This suggests that profit margins may not fully capture performance variability in smaller firms undergoing digital transition.

To formally test whether the impact of digital investment on financial performance differs between the Baltic region and the rest of Europe, interaction terms were introduced in Models 5 and 6 (Table 6). The models include a regional dummy variable (REGION) and its interaction with digital investment (REG_DA) to capture moderated effects.

Table 6. Interaction Effect Analysis for Baltic and Europe Region.

Var.	Baltic Region Model 1 (Coeff. [Std. Error])	Europe Region Model 2 (Coeff. [Std. Error])
DA	0.000 [0.368]	−2.092 *** [0.588]
INTF	−0.591 *** [0.055]	0.214 *** [0.041]
REGION	−0.563 *** [0.110]	−110.731 *** [20.565]
REG_DA	0.018 [0.187]	1.256 *** [0.303]
R&D	−1.023 *** [0.125]	0.050 [0.144]
COST	1.655 *** [0.064]	3.693 *** [0.217]
NETP	0.941 *** [0.045]	1.133 *** [0.073]
TANF		0.311 *** [0.050]
LOGN	0.024 *** [0.005]	−4.857 * [2.599]
CONST.	1.683 *** [0.3210]	255.241 *** [35.065]
Model Summary
Observations	14,935	14,935
Ad. R-squared	0.433	0.366
RESET Test
SS	501,482,904	2766.15154
df	52	51
MS	9,643,902	54.23826
F	166.48	244.23
Prob > F	0.00	0.00
R-squared	0.3678	0.4345
Adj R-squared	0.3656	0.4326
Root MSE	240.68	0.49182

Note: *** p < 0.01, * p < 0.1. Source: Authors’ calculation based on data.

In Model 2, the coefficient of the interaction term REG_DA is positive and highly significant (β = 1.256, p < 0.01), indicating that the effect of digital investment on operating revenue is significantly stronger for firms in the Baltic region compared to the rest of Europe. This provides robust empirical confirmation of H1b, which posits a stronger performance impact of digital investment in the Baltic region relative to the rest of Europe. This effect likely reflects the amplifying role of regional context, structural convergence dynamics, marginal productivity of capital, and policy incentives for digital adoption in peripheral economies.

In contrast, DA on its own is negatively signed and significant (β = −2.092, p < 0.01) in the Europe-wide model. In Model 1 (Baltic only), neither the direct effect of DA (β = 0.000, n.s.) nor the interaction term (REG_DA, β = 0.018, n.s.) is statistically significant, likely due to the small sample size and reduced variability within the regional subset. Still, the Europe-wide interaction model offers clear evidence of differential digital return profiles between regions.

Across both models, employee-related costs (COST) and net profit (NETP) remain strongly and positively associated with operating revenue (p < 0.01), affirming the robustness of H2a. The effect of intangible assets (INTF) is again divergent: negative in Model 1 (β = −0.591, p < 0.01), but positive in Model 2 (β = 0.214, p < 0.01), consistent with earlier findings and supporting H3a’s conditionality on structural maturity. Model diagnostics are strong: the RESET test validates model specification (Prob > F = 0.00 in both models), and the adjusted R-squared values (0.433 and 0.366) indicate substantial explanatory power.

5. Discussion

This study examined the impact of digital investment in production assets on firms’ financial performance, with a comparative focus on the Baltic region and the broader European context. Using firm-level panel data from 30 European countries over the period 2012–2022, the analysis evaluated how digital capital formation, employee-related costs, and intangible fixed asset intensity influence operating revenue while accounting for regional heterogeneity.

The empirical evidence confirms that digital investment in production assets is positively associated with improved financial performance, particularly when variables are normalized by total assets or number of employees. However, the strength and statistical significance of this effect differ across regions. In the broader European sample, digital investment indicates a consistently significant and positive effect, while in the Baltic region, the relationship is economically meaningful but statistically weaker due to smaller sample size and higher within-group variability. Interaction models reveal that the positive performance impact of digital investment is significantly stronger for firms in the Baltic region, supporting H1b.

This result may reflect several underlying structural features specific to the Baltic economies. First, firms in transitional contexts often face greater inefficiencies and lower baseline digitalization levels, which creates room for higher marginal returns when digital technologies are introduced (Scalamonti, 2024; Jiang et al., 2025). Second, policy incentives, such as EU-funded digitization grants and innovation subsidies targeted at SMEs, have disproportionately supported digital upgrading in the Baltics, amplifying investment impact (But et al., 2024). Third, the firm size distribution in the region, characterized by a large share of small and medium-sized enterprises, implies that even modest digital investments can have concentrated effects on operational efficiency and revenue growth (Pira & Fleet, 2025). These factors combine to create an environment where digital investment yields comparatively higher productivity and financial returns, despite lower overall capital intensity.

The results also demonstrate that employee-related costs have a strong and robust positive association with business financial performance in both the Baltic and European contexts. These findings highlight the complementarity between digital transformation and human capital investment, particularly in smaller or transitional economies where labor-related expenditures have a more immediate and concentrated effect.

In contrast, the effect of intangible fixed assets is shown to be more nuanced. In the Baltic region, intangible investments are significantly associated with improved financial performance when normalized by number of employees per business. In the broader European sample, the relationship is positive under certain specifications but also exhibits variation in direction and magnitude. These findings indicate that the returns to intangible capital are dependent on context.

The strong and consistent positive coefficients on employee-related costs (COST) across nearly all specifications may not only reflect human capital enhancement but also broader structural characteristics of the labor environment. For instance, higher labor cost shares may correspond to firm size, sectoral composition, or country-level wage dynamics, particularly in SME-intensive or service-oriented economies like the Baltics, where smaller firms tend to rely more heavily on labor inputs (Stundziene & Baliute, 2022; Li & Lin, 2024). This suggests that employee-related expenditures may function both as productivity enablers and as proxies for firm formalization, capacity development, or compliance with evolving labor standards.

In contrast, the mixed signs observed for intangible fixed assets (INTF) across model specifications highlight their context-dependent financial impact. Intangible capital often requires a foundation of complementary capabilities, including digital maturity, skilled personnel, and effective internal governance (Erjavec et al., 2023). Where such enabling conditions are weak or uneven, returns on intangible investment may be diluted or delayed. Moreover, in transitional economies like the Baltics, the monetization of intangible assets is often constrained by underdeveloped commercialization mechanisms, limited access to capital, and weaker intellectual property enforcement (Khouilla & Bastidon, 2024). These institutional frictions may explain the variation in coefficient direction and significance observed across regions.

It is also important to acknowledge that cross-country differences in depreciation rules, accounting practices, and firm-level capital structures may introduce measurement variation in the reported values of plant and machinery. While normalization by total assets and employees mitigates some comparability issues, these structural accounting differences may still influence the interpretation of capital investment intensity across regions.

These insights have implications for both firms and policymakers. For firms, especially those in transitional economies, the findings suggest that performance improvements depend not only on acquiring digital technology but also on integrating it with complementary human capital strategies. Managers should ensure that digital adoption is matched with appropriate investment in workforce skills, organizational readiness, and process adaptation. For example, firms investing in digitally enabled machinery may improve outcomes by also training staff in data analytics or equipment maintenance. For policymakers, the findings emphasize the need to go beyond infrastructure funding. In regions such as the Baltics, public support should also include targeted upskilling initiatives, digital advisory services, and management development programs, which can help SMEs absorb and operationalize digital investments more effectively. More broadly, a coordinated mix of financial incentives and capability-building programs may be necessary to unlock the full value of production digitalization.

Beyond firm-level and policy implications, this study also touches on broader social considerations. The consistent association between employee-related costs and firm performance suggests that workforce investment remains central to value creation in a digital economy. This highlights the social potential of digital transformation strategies that prioritize job quality and skill development over labor cost reduction. In smaller, digitally advancing economies, aligning digital modernization with inclusive workforce strategies can foster greater regional cohesion, mitigate skill gaps, and improve long-term economic resilience. While these outcomes are not explicitly measured in this paper, they point to important directions for future research into the societal dimensions of production digitalization.

Finally, while this study employed fixed effects and a range of robustness checks to reduce omitted variable bias, potential endogeneity remains a limitation. Future research could strengthen causal inference by using instrumental variable techniques or dynamic panel models. Moreover, expanding the analytical scope to include ESG metrics, firm-level innovation outputs, or long-term performance horizons could offer a more comprehensive understanding of how digital investments influence both economic and social outcomes.