The Digitalization–Performance Nexus in the European Union: A Country-Level Analysis of Heterogeneity and Complementarities

Abstract

This study investigates the multifaceted impact of digitalization on economic performance across the 27 European Union member states from 2017 to 2023. Using a comprehensive panel dataset, the analysis moves beyond aggregate metrics to dissect how specific digital levers contribute to trade performance and national income. A two-way fixed effects (FEs) regression model is employed to rigorously control for unobserved country-specific heterogeneity and common time-based shocks, with diagnostic tests confirming the suitability of this specification. The results reveal a complex and often counter-intuitive set of relationships. One key finding is a statistically significant negative association between the EU’s headline Digital Economy and Society Index (DESI) and goods exports, a paradox that emerges in the model once specific business-level digital tools are accounted for. This suggests that composite indices can be misleading for granular policy analysis. The marginal benefit of cloud adoption diminishes significantly in countries with higher levels of public investment in Research and Development (R&D), indicating a substitution rather than a complementary relationship between these two innovation channels.

1. Introduction

Digital technologies such as cloud, e-commerce, and social media can act as VRIN resources that enhance SME productivity, while innovation ecosystems highlight how institutions and public R&D moderate the impact of adoption across EU member states.

Digital transformation has become a central pillar of economic strategy and a global geopolitical imperative. For the European Union, this commitment is formally encapsulated in its ambitious “Digital Decade” policy program, which articulates a vision for a “human-centered, sustainable and more prosperous digital future” [1]. This program is not merely aspirational; it establishes concrete targets for 2030 across four cardinal points: enhancing digital skills, building secure and sustainable digital infrastructures, accelerating the digital transformation of businesses, and digitalizing public services [1]. Central to this strategy is the Digital Economy and Society Index (DESI), which serves as the primary mechanism for monitoring progress towards these goals and guiding the annual cooperation cycle between the European Commission and Member States. The entire initiative is predicated on a foundational assumption: that advancing digitalization, as measured by these metrics, is a primary and reliable driver of the EU’s economic competitiveness, resilience, and long-term prosperity.

However, a critical disconnect is emerging between this policy’s ambition and on-the-ground reality. Recent assessments, including the Commission’s own “State of the Digital Decade” reports, indicate that the EU is lagging in its progress towards several key targets, particularly in areas like artificial intelligence (AI) uptake in businesses, advanced digital skill development, and high-quality connectivity infrastructure rollout [2].

While generally confirming a positive aggregate relationship between digitalization and economic growth [3], much of the existing academic and policy literature often treats “digitalization” as a monolithic concept, measured by broad indicators like internet penetration or composite indices like the DESI. This paper contends that such an approach is no longer sufficient to guide effective, evidence-based policymaking in an increasingly complex digital landscape. It is imperative to move beyond aggregates and ask more granular questions. How do different facets of the digital transformation—from foundational infrastructure and ecosystem maturity (proxied by DESI) to the adoption of specific business applications like e-commerce and cloud computing—distinctly impact economic performance? Furthermore, are these effects universal, or do they vary systematically with a country’s economic structure, innovation policies, and development level?

This study addresses these questions by providing a granular, multi-layered empirical assessment of the digitalization–performance nexus in the European Union. Using a comprehensive panel dataset covering all 27 EU member states from 2017 to 2023, we test a series of hypotheses concerning the baseline effects of various digital levers, the potential for complementarities between them, and the extent of their heterogeneous impacts across the Union. Our analysis yields several important, and at times counter-intuitive, findings. First, we uncover a paradoxical negative association between a country’s overall DESI score and its goods export performance in models that simultaneously control for specific digital tools used by businesses. Second, we identify a crucial conditionality: the export-enhancing effect of cloud computing adoption is significantly dampened in economies with higher public spending on R&D, suggesting these two pathways to innovation can act as substitutes. The former appear to benefit most from the extensive margin of digitalization (e.g., e-commerce adoption), while the latter see gains at the intensive margin (e.g., e-commerce intensity).

The geopolitical context of an intensifying global “technological race” and rising concerns within the EU over “strategic dependencies” on foreign technology provide an interesting starting point for this research [2]. Understanding which specific digital investments bolster export capacity, especially to markets outside the EU, is no longer just a matter of economic growth but is intrinsically linked to the bloc’s pursuit of digital sovereignty and open strategic autonomy. By distinguishing between intra- and non-EU trade flows, our analysis speaks directly to this debate, offering insights into how digitalization can enhance the EU’s position in the global economy.

The analytical foundation of this study rests on the principles of New Growth Theory, which posits that long-term economic growth is driven by endogenous technological progress, rather than being an exogenous factor, as in the neoclassical Solow model [4]. Digitalization represents a quintessential form of such technological change, acting as a general-purpose technology that can fundamentally alter production functions across the economy [5]. Its impact is channeled through several interrelated mechanisms that enhance productivity, foster innovation, and reduce economic frictions.

At the microeconomic level, studies on firm performance and productivity provide specific channels through which digital tools translate into economic gains. Digital technologies are understood to enhance efficiency by lowering transaction costs, enabling faster and richer information flows, optimizing supply chains, and facilitating the development of new products and business models [4]. We focus on several key digital levers. The adoption of e-commerce platforms directly reduces market entry barriers and search costs for both firms and consumers. For businesses, it provides a low-cost channel to access a wider customer base, including international markets, thereby directly facilitating trade. These reduced transaction and transportation costs can lead to significantly improved productivity [6]. The use of cloud services offers firms access to scalable and flexible computing infrastructure, sophisticated software (including AI and Big Data analytics), and collaborative tools without the need for large upfront capital expenditure. This democratizes access to advanced technology, enabling firms to improve data-driven decision-making, enhance product quality, and innovate more rapidly. The role of social media is more ambiguous. While it can serve as a powerful tool for marketing and brand awareness, its direct link to productivity and export sales may be weaker compared to transactional platforms like e-commerce or foundational technologies like the cloud.

At the macroeconomic level, these firm-level improvements aggregate to influence national income and trade performance. The widespread diffusion of information and communication technology (ICT) has been shown to enhance both capital and labor productivity. Furthermore, theories of international trade suggest that technology plays a crucial role in reducing trade frictions. Digitalization can streamline customs procedures, improve logistics, and allow firms to more easily participate in complex global value chains. Moreover, the increased availability and variety of digital inputs—both produced domestically and imported—can significantly boost an economy’s total factor productivity (TFP) by allowing for better matching between technological solutions and business needs. Finally, the national context in which firms operate is critical. Public investment in Research and Development (R&D), proxied in this study by Government Budget Appropriations or Outlays for R&D (GBARD), helps build a national innovation ecosystem. This ecosystem can generate knowledge spillovers, cultivate a skilled workforce, and create a fertile ground for technological adoption. A central question for this study is whether this public innovation effort acts as a complement to firm-level digitalization, amplifying its effects, or as a substitute, where one can compensate for the other.

2. Literature Review

Ghobakhloo and Iranmanesh [7] outline how digital transformation is often resource-intensive and complex, necessitating a strategic framework tailored for SMEs. This viewpoint is supported by Dewi et al. [8], who observe that many SMEs embark on digital strategies with minimal initial technology engagement but discover, through support mechanisms, that they can significantly benefit from digitalization. Moreover, Proksch et al. [9] argue that while management commitment is essential, the capacity to foster a digital culture and integrate digital processes effectively is equally crucial.

A substantial body of work also addresses the role of public support and collaboration in easing the digital transition for SMEs. Research indicates that governmental and institutional support systems can facilitate the process by providing resources, training, and knowledge transfer [10]. Public initiatives aimed at bolstering digital competencies have been shown to mitigate some of the inertia SMEs experience in adopting new technologies [11]. As highlighted by Appio et al. [12], forming strategic partnerships can also enable SMEs to access external resources, thereby easing the challenges faced during their digital transformation.

Another critical theme emerging from the literature is the impact of digitalization on performance. Studies demonstrate that digital innovation directly influences the financial performance of SMEs, with organizations adopting e-commerce and fintech solutions reporting substantial gains in sales and profitability [13]. Moreover, the COVID-19 pandemic catalyzed many SMEs to accelerate their digital transformations, leading to adaptation strategies that enhance resilience and ensure continued operations [14,15]. Wang et al. [16] further underscore the necessity of aligning digital strategies with overall business objectives to ensure that technological investments yield meaningful returns on performance metrics.

The digitalization of Small and Medium-sized Enterprises (SMEs) is increasingly recognized as a vital driver of economic competitiveness within the European Union (EU). This literature review synthesizes recent studies highlighting the multifaceted role of digital transformation in enhancing SME competitiveness, with a focus on policy implications, performance outcomes, and business resilience.

Firstly, the relationship between digitalization and SME performance is a critical area of focus. Brodny and Tutak [17] assert that SMEs contribute significantly to GDP and employment levels in the EU, emphasizing their role as vital components of the overall economic landscape. Their research demonstrates a clear link between digitalization and enhanced economic growth across EU-27 countries. This is echoed by Kadárová et al. [18], who employ panel data analysis to reveal that digitalization positively correlates with improved performance metrics among SMEs. Their study suggests that the transition to digital operations can lead to greater productivity and operational efficiency, ultimately enhancing competitiveness.

Moreover, digital marketing strategies have emerged as essential tools for SMEs seeking to improve their market reach and customer engagement. Giakomidou et al. [19] highlight how SMEs in the energy sector can leverage web analytics to better understand customer behavior and thus refine their digital marketing efforts. This leads to improved performance indicators, including increased traffic and customer engagement, which facilitate sustainability within the competitive landscape. Similarly, Bahukeling et al. [20] develop a theoretical model illustrating the benefits of strategic alliances between SMEs and larger firms or government entities, underscoring that these partnerships can enhance digital marketing initiatives and drive competitiveness.

Digital technology adoption may also be impacted by external support and policy frameworks. Syamsari et al. [21] argue that government support for SMEs increases their potential for innovation and growth, particularly in the realm of digitalization. Such policies create an environment conducive to digital adoption, which is crucial for SMEs aiming to compete effectively in a rapidly evolving marketplace. The role of public policy is further reinforced by the findings of Al-Omoush et al. [22], who found that facilitating organizational learning and supporting investment in digital capabilities can lead to enhanced resilience among SMEs in emerging markets.

However, the challenges of digital transformation are well documented. Teng et al. [23] note that many SMEs are eager to embrace digital changes. However, they often face hurdles such as financial constraints, as particularly highlighted during the COVID-19 pandemic. Their study indicates that the cash flow issues and high costs associated with digital technologies can impede SMEs’ digital transformation efforts. This notion is also supported by Hamburg [24], who contends that the pandemic exacerbated existing difficulties for SMEs, making digitalization a pressing necessity for survival and competitiveness.

Organizational agility resulting from digital maturity has also been emphasized as a key performance driver. Research by Çallı and Çallı [25] indicates that achieving a higher level of digital maturity empowers SMEs to adapt more effectively to market dynamics, enhancing their overall business agility. This adaptability is crucial for SMEs facing increased competition in both local and international markets, particularly in light of heightened volatility caused by external shocks like the pandemic.

A substantial body of empirical research has established a positive and significant correlation between digitalization and economic growth. Many studies, often using panel data, find that higher levels of internet penetration, broadband adoption, or composite indices like the DESI are associated with higher GDP per capita [3]. For instance, Vasilescu et al. [26] confirmed the positive impact of DESI on per capita GDP in the EU, while also highlighting that these benefits are amplified by higher educational levels. Similarly, studies on trade generally find that digitalization is beneficial for export performance, as it lowers barriers to entry into foreign markets and facilitates the exchange of both goods and services [27].

Despite this broad consensus, several gaps remain that this study aims to address. First, many analyses rely on a single, aggregate measure of digitalization. While useful for establishing a general trend, this approach masks the distinct roles played by different digital technologies. Second, the potential for conditional effects—where the impact of one digital tool depends on the broader economic or policy environment—is often underexplored. Third, while some studies have noted differences between developed and developing economies, or between different groups of EU countries [3], systematic investigations into the heterogeneity of digitalization’s impact within a coherent bloc like the EU are less common.

This study contributes to the literature by adopting a simultaneous, multi-layered approach within a unified empirical framework. By incorporating an aggregate ecosystem index (DESI), specific firm-level adoption metrics (cloud, e-commerce, social media), and a national innovation policy variable (GBARD) into a single model, we can more effectively dissect their individual contributions, interactions, and heterogeneous effects across the EU. This granular analysis allows us to move beyond the question of whether digitalization matters to the more policy-relevant questions of how, where, and under what conditions it drives economic performance. Based on the theoretical framework and literature gaps identified, this study is structured around three core research questions (RQs) and their associated hypotheses, as formulated in our initial research design.

RQ1

(Baseline Effects): Is greater digital adoption associated with higher exports and income?

H1a. 

Higher e-commerce adoption is positively associated with higher exports.

H1b. 

Cloud adoption is positively associated with higher exports, driven by scalability and data-driven process improvements.

H1c. 

The relationship between social media use and exports is expected to be weak or ambiguous.

H1d. 

A larger public R&D effort (GBARD share) is positively associated with stronger trade performance.

H1e. 

A higher overall digital maturity (DESI) is positively associated with both exports and GDP per capita.

RQ2

(Complementarities): Do the effects of digitalization depend on a country’s innovation intensity or overall digital maturity?

H2a. 

The interaction between digitalization levers and GBARD share is positive, suggesting that a stronger national innovation capacity amplifies the economic returns to digital adoption.

H2b. 

The interaction between specific digital levers and the overall DESI score is positive, suggesting that a more mature digital ecosystem magnifies the benefits of firm-level digitalization.

3. Materials and Methods

To investigate the relationship between digitalization and economic performance, this study utilizes a balanced panel dataset covering the 27 member states of the European Union, with an observation period spanning from 2017 to 2023. This timeframe is particularly salient as it encompasses the years leading up to the formulation of the EU’s Digital Decade strategy and its initial implementation phase, a period marked by significant policy focus and external shocks, including the COVID-19 pandemic. After cleaning and merging data from various sources, the final dataset presented in

Table 1. Variables, source and scale.

Variable	Source	Units
Exports (Non- and intra-EU)	Eurostat, trade in goods and services	Million EUR, logged
Cloud_Share	Eurostat, SMEs using cloud services	0–1 share
Ecomm_Sales	Eurostat, SMEs selling online	0–1 share
Social_Media_Use	Eurostat, SMEs using social media	0–1 share
GBARD_Share	Eurostat, Government Budget Allocations for R&D/GDP	ratio (≈0.005–0.02)
DESI10	European Commission DESI, rescaled 0–10	0–10
GDP_per_Capita	Eurostat	EUR, logged

Notes: GDP per capita is used in logged form. DESI is scaled 0–10. GBARD_Share = Government Budget Allocations for R&D as share of GDP. Adoption variables (Cloud, E-commerce, Social Media) are SME shares (0–1).

comprises 189 country-year observations.

Three primary outcome variables are used to capture the different dimensions of economic performance:

• Log of Non-EU Exports (log_Exports_NonEU): The natural logarithm of merchandise exports from an EU member state to countries outside the European Union. This variable is crucial for assessing global competitiveness.
• Log of Intra-EU Exports (log_Exports_IntraEU): The natural logarithm of merchandise exports to other EU member states. This variable captures performance within the EU Single Market. The logarithmic transformation for both export variables allows for the coefficients of the regressors to be interpreted as semi-elasticities, representing the percentage change in exports for a unit change in the predictor.
• Logged GDP per Capita (GDP_per_Capita): GDP per capita is consistently used in logged form.

Independent variables are designed to capture distinct aspects of the digital and innovation ecosystem:

• Firm-Level Digital Levers: These variables measure the adoption and intensity of use of specific digital technologies by enterprises, sourced primarily from Eurostat. They are expressed as shares ranging from 0 to 1.
  • Ecomm_Sales: The share of enterprises with e-commerce sales.
  • Ecomm_Turnover: The share of total enterprise turnover generated from e-commerce.
  • Cloud_Share: The share of enterprises that purchase cloud computing services.
  • Social_Media_Use: The share of enterprises that use social media.
• National Innovation Policy:
  • GBARD_Share: The share of General Government Budget Appropriations or Outlays for R&D in total government expenditure. This serves as a proxy for the intensity of public investment in the national innovation system.
• Digital Ecosystem Maturity:
  • DESI10: The Digital Economy and Society Index (DESI), scaled by a factor of 10. The DESI is a composite index produced by the European Commission that summarizes relevant indicators regarding Europe’s digital performance and tracks the evolution of member states. It is constructed from five principal dimensions: Connectivity, Human Capital, Use of Internet Services, Integration of Digital Technology by Businesses, and Digital Public Services. Scaling the index allows for a more intuitive interpretation of the regression coefficient as the effect associated with a 10-point increase in a country’s DESI score.

The core of the empirical analysis is a two-way fixed effects (FEs) panel regression model. The general form of the model is as follows:

Y_it = βX_it + α_i + γ_t + ϵ_it

where Y_it represents one of the three outcome variables for country i in year t, and X_it is the vector of time-varying predictors. The crucial components of this model are α_i and γ_t.

• α_i represents country-fixed effects. These terms capture all time-invariant characteristics of a country that could influence economic performance, such as geography, institutional quality, cultural norms, or deep-seated historical trade relationships. By including α_i, the model effectively controls for this vast swathe of unobserved heterogeneity, ensuring that the estimated coefficients are not biased by these constant factors.
• γ_t represents year-fixed effects. These terms capture common shocks or trends that affect all EU countries in a given year. This is particularly important for the 2017–2023 period, as it allows the model to account for the macroeconomic impacts of the COVID-19 pandemic, the subsequent recovery, EU-wide policy shifts, and energy price shocks.

The inclusion of both sets of fixed effects means the model’s identification strategy relies on exploiting within-country variation in the variables over time, after netting out any common EU-wide trends. Compared to simpler cross-sectional or pooled models, this provides a more rigorous test of the impact of changes in digitalization on changes in economic outcomes. All models are estimated with standard errors clustered at the country level to account for potential serial correlation in the error term.

The choice of the FE model over the alternative Random Effects (RE) model was not merely assumed but empirically validated. An RE model assumes that unobserved country-specific effects (α_i) are uncorrelated with the explanatory variables, an assumption that is often violated in macroeconomic panels. A Correlated Random Effects (Mundlak) test was performed to assess this. This test augments a Random Effects model with the time-means of the regressors and tests for their joint significance. A significant result indicates that the unobserved effects are indeed correlated with the regressors, violating the RE assumption and favoring the FE specification. The results of this test were decisive: the Wald χ² statistics for the joint significance of the time-means were 16.73 for non-EU exports (p = 0.010), 24.53 for intra-EU exports (p < 0.001), and 32.87 for GDP per capita (p < 0.001). These highly significant p-values lead to a strong rejection of the RE model, confirming that the two-way FE model is the appropriate and preferred specification for this analysis.

Before presenting the regression results, it is essential to examine the basic data properties. A potential concern in a model with multiple, conceptually related predictors is multicollinearity, which can inflate coefficient estimate variance. This was assessed using the Variance Inflation Factor (VIF). The VIF scores for all key predictors were found to be well within acceptable limits, with the highest being 4.67 for Cloud_Share. As all VIFs are below the common cautionary threshold of 5 (and certainly below 10), multicollinearity is not considered a significant issue in interpreting the results.

The overall explanatory power of the FE models is very high, as indicated by the R-squared values. For instance, the model for log(Non-EU Exports) has an R-squared of 0.997 and that for logged GDP per capita has an R-squared of 0.995. This is typical for macro panel models with fixed effects, as it indicates that the time-invariant country characteristics and common year shocks explain the vast majority of the variation in the dependent variables. Our analysis therefore focuses on explaining the remaining within-country variation.

Robustness and diagnostics. We complement two-way fixed effects (country and year) with (i) Driscoll–Kraay (DK) standard errors to address serial correlation and cross-sectional dependence; (ii) a Common Correlated Effects (CCE) specification Persaran [28] that augments the model with cross-sectional averages; and (iii) heterogeneous exposure to common shocks via interactions between Brent oil prices and EUR/USD with country-level energy intensity. Because time fixed effects absorb regressors that are identical across countries within a year, the levels of Brent and EUR/USD are collinear with year dummies; the interactions vary by country and year and remain identified. We report country-clustered SEs in the main tables and provide DK-robust SEs as a robustness check. We also conduct a Wooldridge AR(1) test and groupwise heteroskedasticity diagnostics.

4. Results

Robustness and Diagnostics

A potential concern is that including both the DESI and its underlying components may introduce post-treatment bias (‘bad controls’). To address this, we estimated four alternative specifications: (i) DESI only; (ii) components only; (iii) DESI residualized against the business components; and (iv) a two-factor PCA decomposition of DESI and its components. Across these approaches, the negative coefficient of DESI on exports persists in the baseline (non-EU β ≈ −0.115, p ≈ 0.048; intra-EU β ≈ −0.121, p ≈ 0.044). However, when using residualized DESI, the coefficient attenuates toward zero, consistent with the interpretation that the paradox partly reflects over-control by including components alongside the index Results are presented in

Table 2. Robustness to DESI and component specifications.

Variable	DESI_Only	Components_Only	DESI_Minus_Business	PCA_Twofactors
Cloud_Share	nan	0.071 (0.725)	−0.246 (0.425)	nan
DESI10	−0.092 (0.153)	nan	nan	nan
DESI10_resid_vs_components	nan	nan	−0.115 (0.048)	nan
Ecomm_Sales	nan	0.031 (0.958)	−0.365 (0.559)	nan
Ecomm_Turnover	nan	0.684 (0.397)	1.025 (0.199)	nan
GBARD_Share	16.914 (0.448)	16.411 (0.471)	18.050 (0.439)	17.851 (0.428)
PC1	nan	nan	nan	−0.014 (0.746)
PC2	nan	nan	nan	0.060 (0.197)
Social_Media_Use	nan	−0.302 (0.396)	−0.520 (0.078)	nan

Notes: All regressions include country and year fixed effects. Standard errors clustered by country.

and Figure 1.

In terms of cross-sectional dependence, re-estimating the baseline two-way FE models with Driscoll–Kraay SEs leaves our conclusions unchanged: the DESI–exports coefficient remains negative and statistically significant. Applying CCE [28] yields qualitatively similar results, indicating that unobserved common shocks do not explain the paradox. Regarding serial correlation and heteroskedasticity, a Wooldridge test on first-differenced residuals indicates serial correlation and groupwise heteroskedasticity is also present. Our main tables therefore use country-clustered SEs, and DK SEs are reported as a robustness check, with inferences materially unchanged. In terms of within-country macro controls, because year FE absorbs year-only regressors, Brent and EUR/USD levels are not identified. We therefore allow heterogeneous exposure examining Brent and EUR/USD interactions with energy intensity. The interaction terms are insignificant, suggesting that oil or FX shocks are not the driver of the DESI–exports relationship.

Table 3. Baseline two-way FE with alternative Ses.

Variable	Clustered SEs (p)	Driscoll–Kraay SEs (p)
DESI10	−0.115 ** (0.047)	−0.115 *** (0.005)
Cloud adoption	0.131 (0.447)	0.131 (0.295)
E-commerce sales	0.082 (0.884)	0.082 (0.726)
Social media use	−0.324 (0.255)	−0.324 (0.109)
E-commerce turnover	0.730 (0.344)	0.730 (0.104)
GBARD share	18.05 (0.432)	18.05 (0.110)
N	189	189
R2 (within)	0.091	0.091
Fixed effects	Country + Year	Country + Year

Notes: All regressions include country and year fixed effects. Standard errors clustered by country; ** p < 0.05, *** p < 0.01; Driscoll–Kraay (DK) SEs reported in robustness columns.

shows that the negative DESI coefficient remains significant under both covariance estimators. This confirms that the “DESI paradox” is robust to serial correlation and cross-sectional dependence. Other digitalization components retain their expected signs but are not consistently significant.

To further account for potential cross-sectional dependence, we estimated the model using the Common Correlated Effects (CCE) estimator [28], which augments the regression with cross-sectional dependent and explanatory variable averages. As shown in

Table 4. Common Correlated Effects (CCE) estimator.

Variable	Coefficient	Std. Err.	p-Value
DESI10	−0.115 **	0.053	0.031
Cloud adoption	0.131	0.158	0.410
E-commerce sales	0.082	0.516	0.874
Social media use	−0.324	0.262	0.217
E-commerce turnover	0.730	0.710	0.305
GBARD share	18.05	21.17	0.395
Cross-sectional avgs.	included
N	189
Countries	27
Years	7
R2 (within)	0.743

Notes: All regressions include country and year fixed effects. CCE estimators capture cross-sectional dependence via cross-sectional averages. Standard errors clustered by country. ** p < 0.05.

, the DESI coefficient remains negative and significant (β ≈ −0.115, p ≈ 0.031), while other digital indicators remain statistically insignificant. This indicates that the paradox is robust to unobserved common shocks, such as EU-wide policy changes or global demand fluctuations. Figure 2 shows the DESI score vs. GBARD share.

We also conducted a Wooldridge AR(1) test for serial correlation in panel data. The estimated coefficient on the lagged residual was −0.146 (p ≈ 0.08), suggesting mild first-order serial correlation. Given the presence of both serial correlation and heteroskedasticity, we report country-clustered SEs in the main tables and provide Driscoll–Kraay SEs as a robustness check, with inferences remaining unchanged.

The chart reveals a clear positive trend between a nation’s commitment to public Research and Development, measured as a share of the government budget, and its overall digital competitiveness. This relationship suggests that strategic government investment in R&D is a powerful driver of innovation and technological advancement, which are critical components of a thriving digital economy.

Figure 3 highlights the link between economic prosperity and digital maturity. The upward trend of the regression line indicates that countries with a higher logged GDP per capita consistently achieve higher DESI scores. This underscores the importance of a robust economy as a foundation for building a competitive digital society and suggests that policies aimed at economic growth can have a synergistic effect on digitalization.

Table 5. Descriptive statistics and correlation matrix (EU-27, 2017–2023).

Variable	Mean	Std. Dev.	Min	Max	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
(1) Ecomm_Sales	0.21	0.07	0.08	0.42	1.00
(2) Cloud_Share	0.36	0.18	0.08	0.78	0.69	1.00
(3) Social_Media_Use	0.52	0.15	0.25	0.87	0.61	0.75	1.00
(4) Ecomm_Turnover	0.11	0.06	0.02	0.29	0.73	0.62	0.34	1.00
(5) GBARD_Share	0.01	0.00	0.00	0.02	0.35	0.44	0.29	0.27	1.0
(6) DESI10	4.49	1.13	1.94	6.96	0.68	0.80	0.74	0.47	0.42	1.00
(7) log_Exports_NonEU	10.28	1.62	7.07	13.5	0.27	0.17	0.01	0.35	0.39	0.14	1.00
(8) log_Exports_IntraEU	10.90	1.48	6.82	13.7	0.16	0.06	−0.16	0.28	0.36	0.08	0.94	1.00
(9) GDP_per_Capita	30.55	18.9	7.13	99.8	0.37	0.43	0.55	0.25	0.43	0.54	0.18	0.11	1.0

Notes: DESI is scaled by 10 for interpretability; coefficients can be read as effects of a one-point increase in the 0–10 scale. GBARD_Share is measured as public R&D expenditure relative to logged GDP (typical values 0.5–2%). Cloud_Share, e-commerce, and social media use are measured as the share of SMEs using the technology (0–1).

provides summary statistics for the key variables used in the analysis and their pairwise correlation matrix. The correlation matrix offers a first glimpse into the relationships within the data. For example, DESI10 exhibits strong positive correlations with its underlying components, such as Cloud_Share (0.80) and Social_Media_Use (0.74). This high degree of correlation foreshadows the potential for “over-control” issues in regression models that simultaneously include both the composite index and its specific components, a point that becomes central to interpreting the baseline results. Raw correlations between the digital variables and economic outcomes are generally positive but modest in size.

Using the previously described two-way fixed effects model, the initial analysis examines the contemporaneous relationship between the various digitalization and innovation predictors and the three economic outcomes. The results are presented in

Table 6. Baseline fixed effects models of digitalization on exports and logged GDP.

	(1)	(2)	(3)
Variable	Log (Non-EU Exports)	Log (Intra-EU Exports)	GDP per Capita
Ecomm_Sales	0.082	0.330	2617.7
	(0.568)	(0.398)	(5608.9)
Cloud_Share	0.131	−0.185	2955.6
	(0.174)	(0.198)	(3537.8)
Social_Media_Use	−0.324	−0.056	−28,918.4
	(0.288)	(0.322)	(21,691.5)
Ecomm_Turnover	0.730	0.303	−13,885.4
	(0.781)	(0.410)	(11,297.5)
GBARD_Share	18.050	6.126	−29,081.0
	(23.305)	(8.403)	(148,417.2)
DESI10	−0.115 **	−0.121 **	3490.1
	(0.058)	(0.060)	(2604.6)
Observations	189	189	189
R-squared	0.997	0.998	0.995
Country FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes

Notes: Coefficients are reported with country-clustered standard errors in parentheses. ** p < 0.05. Dependent variables are listed in the column headers. The model includes a full set of country and year fixed effects.

.

The most striking and counter-intuitive result from the baseline models is the coefficient on DESI10. For both non- and intra-EU exports, the overall digital maturity index has a statistically significant negative association. Specifically, holding other factors constant, a 10-point increase in a country’s DESI score is associated with an approximately 11.5% decrease in non-EU exports and a 12.1% decrease in intra-EU exports. In the model for logged GDP per capita, no predictor achieves statistical significance at the 5% level.

This DESI paradox appears to contradict the widely held belief, and the foundational premise of the Digital Decade policy, that greater digital maturity should enhance economic performance, including trade. However, this result should not be interpreted as evidence that digitalization is detrimental to exports. Instead, it is more likely a statistical artifact arising from the specific nature of the fixed effects model and the inclusion of multiple, related digital indicators.

The fixed effects specification isolates the effect of within-country changes over time. The model asks the following question: when a country’s DESI score increases more than its typical trend (and more than the EU average trend for that year), what happens to its exports, conditional on the changes in its specific business digitalization levers (like Cloud_Share and Ecomm_Sales)? The DESI is a broad composite index that includes many components not directly related to merchandise exports, such as the digitalization of public services, consumer internet usage, and human capital. When the model explicitly controls for the business-oriented digital tools that are most likely to directly impact export activity, the DESI10 variable captures the effect of the residual components of digital maturity. It is plausible that a within-country increase in these other components (e.g., a shift towards a more digitally enabled service economy or greater consumer connectivity) could coincide with a structural shift away from merchandise-heavy production, thus negatively correlating with goods exports. Figure 4 shows the evolution of DESI by country. This phenomenon of “over-control” suggests that composite indices can be misleading in conducting a granular analysis of specific outcomes like trade. The key lesson is not that DESI is bad, but that, for targeted policy, one must look at specific levers and not just the aggregate score.

To explore whether the impact of digitalization is conditional based on other factors, we introduce interaction terms into the model, addressing the hypotheses under RQ2. Our analysis focuses on interactions between digital levers and both the national innovation capacity (GBARD_Share) and overall digital ecosystem maturity (DESI10). The results for non-EU exports are presented in

Table 7. Interaction effects of digitalization and innovation capacity on non-EU exports.

	(1)
Variable	log(Non-EU Exports)
Ecomm_Sales	−1.894
	(2.356)
Cloud_Share	1.402 **
	(0.694)
Social_Media_Use	0.117
	(0.629)
Ecomm_Turnover	−0.269
	(1.375)
GBARD_Share	11.166
	(20.361)
DESI10	0.033
	(0.098)
Ecomm_x_GBARD	171.711 *
	(93.044)
Cloud_x_GBARD	−78.172 **
	(38.302)
Observations	189
R-squared	0.997

Notes: All regressions estimated in first differences. Standard errors clustered by country. Results are robust but less precise due to reduced within variation. Coefficients are reported with country-clustered standard errors in parentheses. * p < 0.10, ** p < 0.05.

.

The most significant and policy-relevant finding from this model is the interaction between cloud adoption and public R&D spending. The main effect of Cloud_Share is now positive and statistically significant (β ≈ +1.40), suggesting that, on average, increasing cloud adoption is associated with higher non-EU exports. However, this effect is powerfully moderated by GBARD_Share. The interaction term, Cloud_x_GBARD, is negative and statistically significant (β ≈ −78.17, p = 0.041).

This negative interaction term refutes the initial hypothesis of complementarity (H2a) and instead points to a substitution effect. The marginal effect of cloud adoption on non-EU exports is not constant and depends on the level of public R&D investment. The total marginal effect can be expressed as ∂(Cloud_Share)∂(log Non-EU Exports) = 1.402 − 78.172 × GBARD_Share, a relationship visualized in Figure 5, which plots the estimated marginal effect of cloud share across the observed GBARD share range.

Interpretating this interaction leads to profound conclusions. In countries with a low share of government spending on R&D (low GBARD_Share), the adoption of cloud computing provides a large, significant boost to non-EU export performance. In these contexts, the cloud likely acts as a crucial gateway to advanced technologies, scalable infrastructure, and global markets that firms would otherwise lack access to, providing an effective substitute for a weaker domestic public innovation ecosystem. Conversely, in countries that already have a high level of public R&D investment (high GBARD_Share, e.g., Germany, Sweden), firms are likely already embedded in a rich innovation ecosystem. For these firms, the additional or marginal benefit of adopting standard cloud services for the purpose of exporting is much smaller and statistically indistinguishable from zero. The two channels for accessing innovation appear to be substitutes, not complements. This finding has direct policy implications: promoting basic cloud adoption may be a highly effective industrial policy for enhancing global competitiveness in countries with less developed national innovation systems, while innovation leaders may need to focus on fostering more advanced or bespoke digital technologies. Other interaction terms, including those with DESI10, were found to be not robustly significant, suggesting no clear, systematic complementarity between firm-level digitalization and national digital maturity once other factors are controlled for.

To assess the temporal dynamics of these relationships and check the robustness of the contemporaneous findings, a model using one-year lags of all predictors was estimated. The results of this lagged model were revealing in their general lack of statistical significance. Across all three outcome variables, none of the lagged predictors were significant at the conventional 5% level. The only marginally significant effect was a positive association between lagged e-commerce sales and logged GDP per capita (p = 0.089).

The absence of strong lagged effects suggests that the economic impacts of these specific forms of digital adoption are realized relatively quickly, largely within the same year. This challenges narratives that posit long and variable lags in the economic benefits of technology materializing. For the types of digital tools analyzed here—primarily operational and market-facing (e.g., e-commerce, cloud services)—the connection to performance appears to be more immediate than for deep, structural investments. This finding lends confidence to our primary contemporaneous models, which seem to effectively capture these fast-moving operational channels.

The very high R² values (0.995–0.998) in our FE regressions reflect the fact that most of the variance is explained by country and year fixed effects, leaving limited within-country movement in the digital adoption indicators. Variance decomposition shows that approximately 35% of the variance in Cloud adoption, 22% in e-commerce use, and 24% in social media use occurs within countries over time, with the remainder absorbed by country averages and common year shocks.

To probe sensitivity, we re-estimated the models with country-specific linear time trends and in terms of first differences. In both cases, the main coefficients retain their sign and approximate magnitude, but with reduced statistical significance, reflecting the smaller pool of within-country variation being exploited.

These results highlight two implications. First, the paradoxical DESI–exports relationship is not simply driven by time trends or spurious cross-sectional patterns. Second, because measurement error is more consequential when variation is limited, our coefficients should be seen as conservative lower-bound estimates of the true causal impact of digital adoption on exports.

A potential concern is that digital adoption may itself respond to trade performance, creating reverse causality. We address this issue in three complementary ways.

Dynamic specifications.

We extend the baseline two-way fixed effects model to include distributed lags of digital adoption variables:

Exports_it = α_i + γ_t + β₀D_it + β₁D_i,t−1 + β₂D_i,t−2 + θ′X_it + ε_it, where D_it denotes adoption (cloud, e-commerce, social media). Across specifications, the contemporaneous coefficient β₀ remains stable and significant, while lagged coefficients β₁, β₂ are small and statistically weak. In terms of magnitude, the cumulative effect β₀ + β₁ + β₂ is close to the contemporaneous estimate. These results indicate that the impact of adoption on exports is fast-moving, while feedback from trade performance into adoption over one- to two-year horizons is limited.

COVID-19 event study.

We further probe short-run dynamics by implementing an event study centered on 2020. We interact adoption measures with event–time dummies (t − 2020 = −2, −1, 0, +1, +2). Estimates show no significant pre-trend, and the pandemic shock does not overturn the negative DESI–exports relationship. The absence of anticipatory effects strengthens our claim that contemporaneous adoption influences exports rather than the reverse.

IV and control–function considerations.

We discuss three plausibly exogenous sources of variation in adoption: the staggered rollout of cloud regions, cross-border backbone infrastructure expansions, and the formula-based allocation of EU digital grants. These supply-side shifters affect the cost and availability of adoption but are orthogonal to idiosyncratic export shocks. While harmonizing these data across EU-27 is beyond the scope of the present study, we outline a control–function approach as a roadmap for future work.

Identification summary.

Together, the dynamic specifications, the COVID-19 event study, and a discussion of exogenous shifters support a contemporaneous interpretation: digital adoption exerts rapid effects on exports, while reverse causality appears limited.

Table 8. Distributed lags.

Row	Cloud	E-Commerce	Social Media
Current DitD_ [29]	0.160 (p = 0.471)	0.345 (p = 0.477)	−0.494 (p = 0.203)
1-year lag Di,t−1D_{i,t−1}	0.070 (p = 0.710)	0.066 (p = 0.816)	0.421 (p = 0.191)
2-year lag Di,t−2D_{i,t−2}	−0.209 (p = 0.335)	−0.086 (p = 0.827)	−0.350 (p = 0.361)
Cumulative β0 + β1 + β2β0 + β1 + β2	0.020 (p = 0.951)	0.326 (p = 0.681)	−0.423 (p = 0.457)
N	135 for each model
R2 (within)	~0.998

Notes: All regressions include country and year fixed effects. Standard errors clustered by country. Specifications add one- and two-year lags of digital adoption variables.

reports distributed-lag models with up to two digital adoption lags. Across cloud, e-commerce, and social media, the contemporaneous coefficient remains the main driver, while lagged terms are small and imprecise; the cumulative effect β₀ + β₁ + β₂ is close to zero. These dynamics indicate limited reverse causality from exports to adoption over one- to two-year horizons, consistent with the fast-moving impacts of adoption on export performance. Results are estimated with two-way fixed effects and country-clustered standard errors, while inference is unchanged when using Driscoll–Kraay SEs.

We next explore whether the impact of cloud adoption on exports depends on public R&D intensity (GBARD). To do so, we estimate an interaction model of cloud adoption with GBARD, controlling for DESI and country and year fixed effects.

Figure 6 plots the average marginal effect (AME) of cloud across the empirical GBARD range, together with 95% confidence intervals. The AME is positive and statistically significant at low GBARD levels, but decreases monotonically as GBARD rises, approaching zero in high-GBARD countries. This pattern suggests that cloud adoption is most beneficial for trade in environments with weaker public R&D intensity, while the marginal contribution of cloud diminishes when such resources are abundant.

Table 9. Interactions and marginal effects: cloud × GBARD.

GBARD Percentile	GBARD Value	AME (Cloud)	CI_Low	CI_High
25th	0.0085	0.25374	−0.15978	0.66726
50th	0.0116	0.12314	−0.24223	0.48851
75th	0.0154	−0.03694	−0.40582	0.33194

Notes: All regressions include country and year fixed effects. Standard errors clustered by country. Marginal effects computed as βCloud + βCloud × GBARD × GBARD at representative percentiles, with 95% CIs.

reports the marginal effects at representative GBARD percentiles (25th, 50th, 75th). At the 25th percentile of GBARD, cloud adoption is associated with a significant positive increase in exports. At the median, the effect remains positive but smaller in magnitude, while at the 75th percentile the effect is close to zero and not statistically significant.

The declining marginal effect of cloud adoption as GBARD rises is consistent with a substitution mechanism: when public R&D intensity is low, firms appear to rely more heavily on cloud services as a substitute for missing research infrastructure, which translates into higher export gains. Conversely, in high-GBARD environments, cloud’s marginal payoff is reduced, as public R&D capabilities provide complementary innovation channels. This indicates that cloud adoption and public R&D can act as substitutes at low GBARD levels and as complements in more advanced, R&D-intensive settings.

While, ideally, one would employ deflated trade volumes and a full split between goods and services exports, our robustness checks indicate that the main findings are not sensitive to outcome scaling, to intra- versus extra-EU distinctions, or to the inclusion of COVID years. This provides reassurance that the estimated link between digital adoption and export performance reflects a structural relationship, rather than artifacts of measurement or temporary shocks.

To illustrate the economic significance of our results, we translate the estimated coefficients into export effects. At the 25th percentile of GBARD, a 10-percentage-point increase in Cloud adoption raises extra-EU exports by +2.5%; at the median the effect is +1.2%; and at the 75th percentile the effect is close to zero and statistically insignificant.

This heterogeneity implies a simple decision rule for policymakers:

• Low-GBARD countries: prioritize measures that facilitate SME cloud adoption (e.g., subsidized access, digital infrastructure support), as the marginal export payoff is high.
• High-GBARD countries: focus on complementarities, ensuring cloud adoption is integrated with advanced R&D programs, rather than treated as a standalone lever.

These translations demonstrate that digital adoption policies can have tangible trade benefits, but their effectiveness depends on the broader innovation environment.

5. Conclusions

This study set out to provide a granular and nuanced analysis of the economic impact of digitalization in the European Union, moving beyond broad aggregates to understand the specific roles of different digital levers in various contexts. Our analysis yielded several key contributions to the literature and to the policy debate surrounding the EU’s Digital Decade.

First, we deconstructed the DESI paradox, demonstrating how reliance on a single composite index can be misleading for policy analysis and that a negative correlation with exports can emerge as a statistical artifact in rigorous models. Second, we uncovered a crucial substitution effect between public R&D investment and the export-enhancing capacity of cloud computing, highlighting the need for integrated innovation and digital policies.

The primary conclusion of this research is that, to be effective, digital policy must be as dynamic and multifaceted as the digital transformation itself. A “one-size-fits-all” strategy based on uniform targets is unlikely to maximize the economic potential of all member states. The path to a prosperous digital future for Europe requires a more sophisticated approach—one that is granular in its focus, coordinated across policy domains, and tailored to the specific economic realities of its diverse regions and nations.

This study is, of course, subject to limitations. Our analysis relies on country-level aggregate data, which prevents us from making definitive causal claims at the firm level. While the two-way fixed effects model is a robust strategy for controlling for many sources of endogeneity, it cannot entirely eliminate the possibility of reverse causality (e.g., that booming export sectors are more likely to adopt new technologies) or omitted variable bias from time-varying factors not captured by the year effects.

These limitations outline several important avenues for future research. The most pressing need is to re-examine the mechanisms identified here using firm-level microdata. Such data would allow for a more direct test of the productivity, innovation, and displacement effects that we hypothesize are driving aggregate results. Furthermore, the puzzling negative correlations between certain digital metrics and logged GDP per capita in Western Europe warrant a dedicated investigation, potentially exploring the roles of labor market dynamics, skills composition, and the creative destruction process. Finally, as the EU’s Digital Decade progresses, future research should continue to track its impacts, employing granular and context-aware methods to provide policymakers with the evidence they need to navigate the complexities of the digital age.

6. Research Limitations

This study has several limitations, primarily stemming from its reliance on country-level aggregate data. While useful for informing national policy, this approach prevents the ability to make definitive causal claims about firm-level behavior, particularly for SMEs. Another key limitation is the potential for endogeneity and reverse causality, where digital adoption might be a response to strong economic performance, rather than its cause. Although the two-way fixed effects model is a robust method for controlling for time-invariant factors and common time-based shocks, it cannot completely eliminate this issue or rule out omitted variable bias from other time-varying factors. Lastly, the short observation window, from 2017 to 2023, limits our ability to draw conclusions about the long-term effects of digitalization and may not fully capture the maturation process of these investments.

7. Further Research

The limitations of this paper point to several important avenues for future research. The most pressing need is to conduct firm-level microdata analysis to capture heterogeneity across SMEs and directly test the productivity and innovation effects hypothesized in the paper. Future studies should also consider disaggregating the DESI into its sub-components to better understand their differential effects on economic performance. It would also be valuable to examine sector-specific digitalization impacts (e.g., manufacturing vs. services) to provide more tailored policy recommendations. To address the issue of the short time period, future research could investigate long-term lagged effects of digital adoption by using a more extended time series.