Written by: Dr. Marușa Beca Pescu, Babes-Bolyai University in Cluj.
Abstract
This paper investigates whether digital transformation contributes to export upgrading in Central and Eastern Europe (CEE). Using panel data from 11 CEE countries between 2007 and 2023, we estimate fixed-effects models with heteroskedasticity- and autocorrelation-consistent (HAC) standard errors to assess the impact of internet penetration and R&D expenditure on the share of high-technology exports in manufactured goods.
The empirical results show that greater internet usage is a statistically significant and robust driver of high-tech export performance. In contrast, R&D expenditure does not display a direct effect, suggesting that innovation-led trade gains may depend on institutional capacity and absorptive readiness.
Diagnostic tests confirm the presence of serial correlation and non-normality in residuals, which are mitigated through robust estimation techniques. Sensitivity checks further validate the model’s stability across alternative specifications and country exclusions.
The findings emphasize the strategic importance of expanding digital infrastructure and skills development as policy tools for trade competitiveness. Strengthening R&D governance and improving the link between research and commercialization could complement digital investments in fostering export sophistication.
1. Introduction
Digital transformation has emerged as a pivotal force reshaping global trade dynamics, particularly through its influence on production, logistics, and market access. For Central and Eastern European (CEE) countries—many of which have experienced rapid transitions and EU integration—digital connectivity offers a strategic opportunity to move up the value chain and enhance export sophistication (Grossman & Helpman, 1991; European Investment Bank, 2020).
However, progress across the region remains uneven. While frontrunners like Estonia and Poland have made significant investments in digital infrastructure and skills, others lag behind due to institutional, economic, or infrastructural constraints (OECD, 2023). These disparities raise important questions about whether digital capabilities, such as internet access, can effectively contribute to export upgrading, especially in the absence of strong innovation ecosystems.
This paper investigates whether digital transformation contributes to high-technology export performance in the CEE context. Specifically, it examines the impact of internet penetration and R&D expenditure on the share of high-tech exports in manufactured goods. Using panel data from 11 CEE countries between 2007 and 2023, we estimate fixed effects models with heteroskedasticity- and autocorrelation-consistent standard errors to explore the dynamics between connectivity and export outcomes.
The study seeks to address the following core questions:
- To what extent does increased internet penetration enhance high-technology export performance in CEE economies?
- Can digital connectivity compensate for weak or misaligned R&D investment?
- What are the policy implications of promoting digital infrastructure as a lever for export upgrading?
By addressing these questions, the research contributes to the growing literature on digitalization and trade competitiveness in emerging European economies, while offering targeted insights for policymakers seeking to advance industrial upgrading in the digital age.
2. Theoretical Foundations and Empirical Context
The relationship between technological innovation and trade performance has been extensively explored in economic literature. Classical trade theories, such as the Heckscher-Ohlin model, emphasize factor endowments in shaping comparative advantage (Heckscher & Ohlin, 1991). In contrast, contemporary approaches highlight the central role of technology, innovation, and human capital in determining export performance (Grossman & Helpman, 1991; Romer, 1990). Endogenous growth theory, in particular, links sustained economic development to investments in research, development, and knowledge diffusion.
Building on these foundations, recent empirical studies have investigated how digital transformation influences trade outcomes. Numerous cross-country analyses confirm that investments in ICT infrastructure improve productivity and competitiveness, thereby facilitating export upgrading (World Bank, 2020). In Europe, the adoption of digital production technologies has been associated with increased export intensity, particularly among manufacturing firms (OECD, 2019).
In the context of CEE, digital adoption has advanced unevenly across countries. Estonia and Poland have emerged as regional leaders in internet connectivity and technological uptake, while others face challenges related to infrastructure, digital literacy, and policy implementation (OECD, 2023). Research using spatial econometric methods suggests that high-tech export growth in one CEE country may generate positive spillover effects in neighboring economies (World Bank, 2020). However, these interdependencies fall outside the scope of our analysis, which focuses on within-country dynamics.
Our study narrows the empirical scope to two central explanatory variables: internet penetration (as a proxy for digital infrastructure) and R&D expenditure (as a proxy for innovation intensity). While broader indicators—such as ICT investment, digital skills, or regulatory maturity—are also recognized in the literature, our model concentrates on these two dimensions due to data availability and cross-country comparability. This approach is consistent with previous studies that use internet accessibility as a measurable input in digital readiness assessments.
Recent debates have also turned to emerging technologies like artificial intelligence (AI), automation, and Industry 4.0, which are seen as next-generation enablers of trade competitiveness. While these developments are especially relevant for digitally advanced economies like Estonia and Czechia, their limited adoption elsewhere in the region highlights persistent digital divides (OECD, 2022). Given data constraints, our model does not explicitly capture the effects of such technologies, though we acknowledge their potential as complementary factors in future research.
Policy frameworks such as the EU’s Digital Single Market Strategy have shaped the digital environment across CEE, but implementation remains fragmented. Regulatory disparities, coupled with weaknesses in cybersecurity and digital education, continue to constrain the full potential of ICT-led export transformation (European Commission, 2023; World Economic Forum, 2020).
In this complex landscape, our empirical investigation aims to assess whether expanding digital connectivity—measured through internet usage—can foster export sophistication, particularly in contexts where R&D systems may be underdeveloped or misaligned with commercial outcomes.
3. Data and Methodology
This study investigates the relationship between digital transformation and export upgrading in CEE countries using panel data econometric techniques. The empirical strategy combines descriptive statistics and fixed-effects regression modeling, with robust inference methods applied to correct for heteroskedasticity and serial correlation.
3.1 Data and Sources
The dataset includes annual observations from 11 CEE countries over the period 2007–2023. Data were collected from publicly available international databases, including the World Bank (for internet usage and GDP growth), OECD (for R&D expenditure), and Eurostat (for high-technology exports).
All variables were selected based on their availability across countries and years, comparability across sources, and theoretical relevance to export competitiveness and digital adoption.
3.2 Variable Description
Dependent Variable:
High-technology exports (% of manufactured exports) – sourced from Eurostat, this indicator measures the share of high-tech products in total manufactured exports and reflects export sophistication.
Key Independent Variables:
- Internet users (% of population) – a proxy for digital connectivity and infrastructure, reflecting the level of internet penetration (World Bank).
- R&D expenditure (% of GDP) – representing national investments in research and innovation (OECD) (The values for 2022 and 2023 were forecasted using the FORECAST.LINEAR() function in Excel).
3.3 Econometric Model
To analyze the impact of digitalization on export performance, we estimate a fixed-effects panel regression model. The fixed-effects estimator is chosen based on Hausman test results, which indicated correlation between unobserved country-specific effects and the regressors.
The baseline econometric specification is as follows:
HighTechExpit = αᵢ + β₁ × InternetUsersit + β₂ × RDEit + εit
Where:
- HighTechExpit : high-tech exports (% of manufactured exports) for country i in year t
- InternetUsersit: internet penetration rate
- RDEit: R&D expenditure as % of GDP
- αᵢ: unobserved country-specific fixed effects
- εit: idiosyncratic error term
To account for heteroskedasticity and autocorrelation typical in macro panel datasets, the model uses heteroskedasticity- and autocorrelation-consistent (HAC) standard errors (White, 1980).
Descriptive Statistics and Diagnostics.
Before estimation, we compute summary statistics to assess data distribution and variation across countries and time. Residual diagnostics are conducted post-estimation to evaluate potential misspecification, including:
- Residual Q–Q plots to assess normality;
- Visual checks for heteroskedasticity;
- Durbin-Watson statistic to detect autocorrelation.
Robustness Checks.
Robustness of the findings is evaluated through:
- HAC (White) standard errors to address serial correlation and cross-sectional heteroskedasticity;
- Exclusion of countries one at a time to test result stability;
- Comparison with alternative model specifications including GDP growth and additional years (2007–2021 vs. 2007–2023).
4. Results and Interpretation
This section presents the main findings from the panel regression analysis estimating the effects of digital transformation on export sophistication in 11 CEE countries over the 2007–2023 period. The dependent variable is the share of high-technology exports in manufactured exports, while the key explanatory variables include internet penetration and R&D expenditure.
4.1 Descriptive Statistics
Summary statistics reveal substantial heterogeneity in both the dependent and independent variables across countries and years:
- The average share of high-tech exports was 12.54%, with a standard deviation of 5.12, indicating meaningful variation in export sophistication.
- Internet usage ranged from 28.3% to 93.2%, with a mean of 71.77%, capturing the digital expansion across the region.
- R&D expenditure averaged 1.11% of GDP (s.d. = 0.54), suggesting modest but uneven investment in innovation infrastructure.
These patterns justify the use of panel modeling to account for within- and between-country variation over time.
4.2. Model Estimation and Specification
We estimated a series of panel data models to investigate the impact of digital transformation on export performance in Central and Eastern Europe. The initial specification included four explanatory variables: Internet usage, R&D expenditure, GDP growth, and broadband penetration. However, after evaluating multicollinearity, statistical insignificance, and overall model fit, we refined the model to retain only two core predictors: Internet_users and R&D expenditure (RDE).
The final specification is a fixed-effects panel regression with heteroskedasticity- and autocorrelation-consistent (HAC) standard errors, formulated as:
HighTechExpᵢₜ = αᵢ + β₁ × InternetUsersᵢₜ + β₂ × RDEᵢₜ + εᵢₜ
Where:
- HighTechExpᵢₜ represents the share of high-technology exports in total manufactured exports for country i at time t,
- αᵢ captures country-specific unobserved heterogeneity,
- InternetUsersᵢₜ and RDEᵢₜ are the explanatory variables,
- εᵢₜ denotes the idiosyncratic error term.
We applied the Hausman test to both the full and reduced models to determine the appropriate estimator. For the full model, the test yielded a p-value of 0.63, suggesting that the random effects estimator was acceptable. However, in the reduced specification, the Hausman test returned χ² = 9.18, p < 0.05, indicating that the fixed-effects model was more appropriate due to likely correlation between the regressors and the unobserved effects.
Consequently, we estimated Model 4 using fixed effects with robust (HAC) standard errors, as it aligns both with statistical diagnostics and the theoretical expectation of country-specific heterogeneity across CEE economies.
4.3 Main Findings
The estimation results for Model 4 are summarized below:
| Variable | Coefficient | Std. Error | t-ratio | p-value | Significance |
|---|---|---|---|---|---|
| Constant | 6.173 | 3.245 | 1.902 | 0.086 | * |
| R&D expenditure (%) | 0.480 | 1.819 | 0.264 | 0.797 | ns |
| Internet users (%) | 0.0813 | 0.0357 | 2.276 | 0.046 | ** |
- The coefficient for Internet users is positive and statistically significant at the 5% level, suggesting that a 1 percentage-point increase in internet usage is associated with a 0.08 pp increase in high-tech export share.
- R&D expenditure is not statistically significant, indicating that innovation investment alone may not directly translate into export upgrading within the current institutional framework.
The LSDV R² is 0.78, showing strong explanatory power due to fixed effects, while the within R² is modest (0.15), reflecting limited time variation. Although the LSDV R² is relatively high (0.78), indicating that the model explains a large portion of total variance including fixed effects, the within R² is more modest (0.15), reflecting the proportion of variation in high-tech exports explained solely by changes within countries over time. This discrepancy highlights the importance of country-specific fixed effects in capturing unobserved heterogeneity across the panel.
5. Residual Diagnostics and Robustness Checks
5.1 Residual Analysis
- Normality: Q-Q plots and histogram of residuals suggest moderate non-normality (skewness, heavy tails). The Chi-square test confirms this (statistic = 30.42, p < 0.0001), but inference remains valid under HAC errors.
- Autocorrelation: The Durbin-Watson statistic = 0.38, indicating significant positive serial correlation. This is expected in time series panels and corrected through HAC estimation.
- Heteroskedasticity: The Breusch-Pagan test rejects the null of homoscedasticity (p < 0.0001). Robust errors were used throughout to account for this issue.
- Functional form: No systematic pattern or heteroscedastic funnel shape in residual vs. fitted plots. No influential outliers were detected.
5.2 Robustness Checks
- Model comparison: Both random and fixed-effects models were estimated. Although the Hausman test in one specification did not reject the random effects model (p = 0.63), the presence of strong group-specific intercepts (Welch F = 105.0, p < 0.0001) justifies the fixed-effects specification.
- Reduced model: When estimating a simpler fixed-effects model with only Internet_users and RDE, the coefficient for Internet_users remained significant and stable, confirming robustness.
- Country exclusion: Re-estimating the model while removing outlier countries (e.g., Estonia, Slovenia) yielded consistent results.
Overall, results confirm that internet penetration is a robust and significant predictor of export sophistication in CEE economies, while R&D expenditure, though theoretically important, does not exhibit a direct effect in the models estimated. This implies that connectivity infrastructure and digital accessibility may be more immediately impactful than innovation spending alone, unless accompanied by complementary institutional reforms.
6. Policy Recommendations
The empirical results emphasize that digital connectivity plays a more direct and measurable role in enhancing high-tech export performance than R&D spending alone in CEE economies. In light of these findings, we propose the following policy priorities:
1.Treat digital access as core economic infrastructure.
Internet usage is a statistically significant determinant of export sophistication. Policymakers should shift the framing of broadband access from social utility to strategic industrial infrastructure. Efforts should focus on expanding high-speed internet in underserved regions—especially rural and industrial peripheries—to unlock digital-led trade competitiveness.
2. Invest in complementary digital capabilities.
Connectivity without capabilities limits impact. To leverage digital access, national strategies must invest in digital skills, STEM education, and workforce upskilling. This ensures firms and labor markets can adopt and apply digital technologies in export-oriented sectors.
3. Rethink how R&D investment is governed and linked to trade outcomes
While R&D spending levels appear statistically insignificant in our models, this likely reflects inefficiencies in translating research inputs into commercial innovation rather than an absence of effort. Therefore, policy should prioritize not just increased investment, but improving institutional channels that convert R&D into export-oriented outcomes. Policies should focus on:
- Strengthening university-industry links,
- Funding applied research with export potential,
- Incentivizing innovation in tradable sectors.
4. Foster regional coordination and digital policy alignment.
Cross-border cooperation can help scale up the impact of digital transformation. CEE governments should explore:
- Joint research infrastructures,
- Shared digital platforms and innovation ecosystems,
- Harmonized standards for data and cybersecurity.
Regional alignment would improve both market access and innovation diffusion, especially for smaller economies.
These policy directions aim to transform the current digital divide into a strategic advantage, positioning CEE economies to upgrade their export structures in the face of growing global competition and technological disruption.
7. Conclusion
This study examined the role of digital transformation in fostering export sophistication across CEE economies, focusing on two principal drivers: internet penetration and R&D expenditure. Using fixed-effects panel regressions with heteroskedasticity- and autocorrelation-consistent (HAC) standard errors on data from 11 countries (2007–2023), we find robust evidence that greater internet access is positively and significantly associated with the share of high-technology exports in manufactured goods.
In contrast, R&D expenditure does not show a statistically significant direct effect. This suggests that in many CEE countries, where innovation ecosystems remain fragmented, digital connectivity may serve as a more immediate and scalable instrument for boosting export competitiveness than traditional R&D investment.
These findings reinforce the idea that expanding internet infrastructure and access should be prioritized not only as educational or social objectives, but as core pillars of industrial policy. Simultaneously, the limited impact of R&D points to a need for structural reforms—strengthening the links between research, commercialization, and export-oriented innovation.
Future research could expand this analysis by incorporating indicators of institutional quality, adopting dynamic panel methods, or exploring potential interaction effects between digitalization and macroeconomic structures. Such directions would further illuminate how digital transformation can translate into sustained competitive advantages for emerging economies in the European context.
8. Executive Summary
This report investigates the role of digital transformation in enhancing the export performance of CEE economies, with a particular focus on high-technology exports. Despite ongoing efforts to promote innovation-driven growth, many CEE countries still face limitations in translating R&D investment into competitive export outcomes.
Using panel data from 11 CEE countries between 2007 and 2023, we apply fixed-effects models with robust standard errors to analyze how internet penetration and R&D expenditure influence the share of high-tech goods in manufactured exports. The results show that internet penetration is a consistently strong and statistically significant predictor of export sophistication, while R&D expenditure has no direct measurable impact in the current institutional context.
The findings carry important implications for policy design:
- Digital infrastructure—especially broadband coverage—should be considered a strategic lever for industrial and trade competitiveness;
- Digital skills development must accompany infrastructure expansion to unlock its full economic potential;
- Innovation policy should shift from funding volume to institutional reform, focusing on how research translates into marketable technologies.
This analysis suggests that digital accessibility offers a more immediate and scalable pathway to export upgrading than traditional innovation spending. For CEE countries aiming to transition beyond the middle-income trap, a digitally focused policy mix may deliver faster and more inclusive gains than relying on R&D investments alone.
This research offers a novel empirical contribution to understanding export dynamics in CEE, a region often underrepresented in digital transformation literature.
References
Baltagi, Badi H. 2021. Econometric Analysis of Panel Data. 6th ed. Hoboken, NJ: Wiley.
European Commission. 2020. Shaping Europe’s Digital Future. https://digital-strategy.ec.europa.eu.
European Commission. 2023. Digital Economy and Society Index (DESI). https://digital-strategy.ec.europa.eu/en/policies/desi.
European Investment Bank. 2020. Innovation Investment Report: Building a Smart and Green Europe in the COVID-19 Era. https://www.eib.org/en/publications/investment-report-2020.
Grossman, Gene M., and Elhanan Helpman. 1991. Innovation and Growth in the Global Economy. Cambridge, MA: MIT Press.
Granger, Clive W. J. 1969. “Investigating Causal Relations by Econometric Models and Cross-Spectral Methods.” Econometrica 37 (3): 424–38.
Hausman, Jerry A. 1978. “Specification Tests in Econometrics.” Econometrica 46 (6): 1251–71.
Heckscher, Eli F., and Bertil Ohlin. 1991. Heckscher-Ohlin Trade Theory. Cambridge, MA: MIT Press. (Originally published 1933)
Im, Kyung So, M. Hashem Pesaran, and Yongcheol Shin. 2003. “Testing for Unit Roots in Heterogeneous Panels.” Journal of Econometrics 115 (1): 53–74.
OECD. 2019. OECD Science, Technology and Innovation Outlook 2019. Paris: OECD Publishing. https://doi.org/10.1787/9789264314565-en.
OECD. 2022. OECD Digital Economy Outlook 2022. Paris: OECD Publishing. https://doi.org/10.1787/1b0b29ac-en.
OECD. 2023. Going Digital in Central and Eastern Europe. Paris: OECD Publishing. https://www.oecd.org/digital/going-digital/.
Romer, Paul M. 1990. “Endogenous Technological Change.” Journal of Political Economy 98 (5, Part 2): S71–S102.
White, Halbert. 1980. “A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity.” Econometrica 48 (4): 817–38.
World Bank. 2020. World Development Report 2020: Trading for Development in the Age of Global Value Chains. https://www.worldbank.org/en/publication/wdr2020.
World Bank. 2021. The Digital Transformation of the Manufacturing Sector in Central and Eastern Europe. https://www.worldbank.org/en/region/eca/publication.
World Economic Forum. 2020. The Future of Jobs Report 2020. https://www.weforum.org/reports/the-future-of-jobs-report-2020.
