Linking R&D Expenditure to Labour Market and Economic Performance: Empirical Evidence from the European Union

Abstract

This article examines how research-and-development (R&D) expenditure—as a share of GDP—both in total and disaggregated by sector (business enterprise and government)—shapes key socioeconomic outcomes in the EU-27. Drawing on Eurostat panel data for 2013–2022, we estimate fixed- and random-effects models with sector-specific lags. Business R&D expenditure is associated with lower female and male unemployment and faster GDP growth. Government R&D expenditure, by contrast, widens the gender pay gap and dampens GDP per capita after two years, although it attracts foreign direct investment in the short and medium term. The diminishing impact of R&D over time underscores the need for policies that sustain innovation benefits.

1. Introduction

The development of the global economy is driven by technological progress, with research and development (R&D) spending recognized as a key enabler of innovation and growth [1,2,3]. However, R&D outcomes are complex and may entail unintended consequences, such as automation-induced job losses in routine sectors [4,5,6,7]. Motivations for R&D also differ across private, public, higher education, and non-governmental sectors, resulting in heterogeneous economic and social effects [8,9,10].

Within the European Union, the ambition to devote 3% of GDP to R&D—reaffirmed in the 2023 Pact for Research and Innovation—underpins green and digital transitions as well as the post-2027 cohesion-policy framework. Despite extensive research on R&D’s aggregate role in growth and employment, sector-specific and temporal effects of R&D on key social indicators (e.g., unemployment by gender, gender pay gap, healthy life expectancy) and economic indicators (e.g., GDP per capita, FDI) remain insufficiently explored in the EU context [11,12]. Previous studies have mostly focused on aggregate macroeconomic indicators, overlooking the potentially heterogeneous effects across sectors and social groups. This leaves a critical gap in our understanding of how R&D shapes both economic and social outcomes across different sectors and over time, particularly in the context of the European Union’s sustainability agenda.

Addressing the multifaceted challenges of sustainable development requires not only economic growth, but also the promotion of social equity, resilience, and institutional quality. R&D expenditure is increasingly recognized as a critical driver of these objectives, as it fosters innovation, enhances productivity, and can generate broad-based societal benefits. However, the channels through which R&D affects key socio-economic outcomes—including unemployment and wage inequality—are highly context-dependent and mediated by sectoral, institutional, and technological factors. Understanding these mechanisms is essential for designing evidence-based policies that align the economic and social dimensions of sustainability, in line with the priorities articulated in the 2030 Agenda for Sustainable Development.

To fill the gap the present study makes three key contributions. First, it applies a single empirical framework to examine how sector-specific R&D expenditure influences both economic (GDP growth and FDI inflows) and social (unemployment and gender wage gap) outcomes—an integrated perspective rarely adopted in prior literature. Second, by using distributed-lag models, it identifies not only the magnitude but also the timing and duration of R&D effects in each sector, providing actionable guidance for policymakers. Third, it directly addresses equity objectives by documenting gender-differentiated labour-market impacts, thereby informing EU equal-opportunity and skills-based recovery policies.

The empirical analysis covers all 27 EU member states from 2013 to 2022. Fixed- and random-effects estimators with sector-specific distributed lags are used to chart both immediate and delayed effects of R&D expenditure shocks. The results are interpreted in light of EU policy targets for innovation intensity and inclusive growth, supplying relevant evidence for policymakers.

Although substantial differences exist among EU-27 countries in economic structure, innovation capacity, and labour market dynamics, this study focuses on the European Union as a whole. This approach ensures comparability with EU-level policy objectives (such as the 3% R&D target) and enables a unified analysis using harmonized data. While the analysis acknowledges variations in the strength and timing of R&D effects across member states, the macro-level perspective helps to capture cross-country spillovers and provides insights relevant for European policy.

At the same time, it is important to recognize that the effects of public R&D—particularly in domains such as health, education, or basic science—may require a longer time frame to fully materialize, as suggested by the previous literature [13,14]. Although the present study focuses on a 1–4-year window to facilitate timely policy insights, this macro-level empirical framework does not attempt to fully disentangle the complex causal pathways and spillover effects that link government and business R&D activities. As noted in the innovation literature [13,15], the business sector often benefits from public investments in education, infrastructure, and basic research, which may mediate or amplify the observed effects. Therefore, the results should be interpreted as net sectoral associations, rather than isolated causal impacts. These methodological limitations, including the need for both a longer time perspective and future micro-level or disaggregated studies, are discussed in the following sections. The structure of the paper is as follows: Section 2 presents key findings on the impacts of R&D; Section 3 describes materials and methods; Section 4 reports the results; Section 5 discusses their implications; and Section 6 presents the conclusions.

2. Literature Review

Research on R&D’s socio-economic impacts begins with its effects on production and employment. R&D investments optimize production processes—reducing costs, increasing output, and stimulating employment growth through product innovations [16,17,18,19,20]. Yet this positive impact can be counterbalanced by process innovations that initially displace workers before compensatory expansion occurs [21,22,23,24]. Understanding these contrasting outcomes requires paying attention to timing, since the benefits of R&D do not always materialize immediately.

Indeed, several studies document that first-year returns on R&D may even be negative due to high upfront costs [25,26], with positive effects emerging only after one to two years once innovations begin generating revenue [27,28]. This lagged pattern suggests that policymakers and firms must balance short-term disruptions against longer-term gains. However, timing is only one dimension: the sector in which R&D takes place also shapes both the magnitude and persistence of its effects.

Grounding this expectation in the Sectoral Innovation Systems perspective [29,30], we posit that the mechanisms through which R&D generates spill-overs are conditioned by the technological trajectories, demand patterns and regulatory regimes that are unique to each sector. Business enterprise R&D expenditure, aligned with market-driven trajectories, is therefore likely to deliver fast, commercially focused returns, whereas government R&D expenditure, embedded in mission-oriented and often more regulated domains, produces broader but time-lagged pay-offs. Higher education R&D expenditure primarily expands the collective knowledge base, while the R&D expenses of non-governmental sector generate specialized social missions that can stabilize local labour markets.

Private-sector R&D typically yields immediate commercial returns, driving rapid growth in GDP and job creation [16,17]. By contrast, public-sector R&D often pursues broader social objectives—such as health, education, or environmental goals—with economic payoffs that may be delayed or diffuse [31,32].

The observed negative association between government R&D expenditure and GDP growth in the macroeconomic analysis may stem from a range of overlapping mechanisms. Public funding for research and development can crowd out private-sector investment, particularly when allocation lacks competitive criteria and is not effectively targeted at high-potential areas [14]. This can reduce the overall efficiency of innovation and slow productivity growth at the aggregate level. Decision-making processes in the public sector are often constrained by bureaucratic rigidities, fragmented governance structures, and weak performance incentives, resulting in suboptimal project selection and limited responsiveness to market needs [33]. Moreover, the allocation of public funds is sometimes shaped by political considerations and may be concentrated in low-productivity sectors, thereby weakening the developmental impact of public investment. Another concern is the limited capacity for commercializing the outcomes of publicly funded research—many scientific results fail to reach industry and remain within academic or institutional settings [34]. Structural mismatches in the labour market may further amplify social inequalities. This is particularly evident when new R&D jobs are created predominantly in high-wage, male-dominated technology sectors, while women remain overrepresented in lower-paid fields such as education or public administration [35]. As a result, the negative effects of government R&D expenditure may manifest not only as a lack of short-term impact, but also as a tangible drag on economic performance and social cohesion.

Despite these challenges, government R&D expenditures may still exert a positive influence under certain conditions, particularly when it complements private sector efforts or is embedded in strong institutional environments.

Cadil et al. [36] further demonstrate that government subsidies for private R&D can enhance firm-level innovation and productivity, suggesting a synergistic effect when public support is strategically targeted at sectoral research and development. At the same time, regional capacity influences these dynamics: advanced economies are better able to absorb displacement effects and convert R&D into productive outcomes than developing regions [37,38], though innovation also has significant employment effects in other contexts, such as in Latin America [39].

Beyond production and employment, R&D also drives cross-border capital movements through technology spillovers. Firms and countries that invest heavily in R&D become more attractive destinations for foreign direct investment [40,41,42], as innovations create competitive advantages and knowledge transfer opportunities. Yet the strength of this relationship varies over time and depends on national policies, institutional quality, and absorptive capacity [26,28].

In the EU context, such timing effects unfold against a dense regulatory backdrop. The Work–Life-Balance Directive (EU) 2019/1158 [43], the Adequate Minimum-Wages Directive (EU) 2022/2041 [44] and the Pay-Transparency Directive (EU) 2023/970 [45] can cushion short-term labour-market shocks while accelerating the translation of R&D-induced productivity gains into more equal wage structures. Recent cross-country evidence also confirms that when policy frameworks, including those fostering R&D cooperation [46], are aligned with sectoral needs, R&D-driven spill-overs intensify over time: Bing and Zhang [47] show that coordinated investment incentives boost green total-factor productivity, illustrating how institutions shape both the magnitude and the lag of innovation effects.

Finally, the interaction between R&D, automation, and labour-market structure produces pronounced gender and polarization effects. Automation tends to displace routine [2,6,48], often male-dominated jobs [7] while increasing demand for non-routine roles more frequently held by women [49,50,51,52,53]. Empirical work by Altuzarra et al. [54] demonstrates that in countries with higher gender inequality, R&D-driven growth translates into larger disparities in labour-force participation rates, underscoring how institutional barriers modulate R&D’s gendered impacts. Although automation can narrow the gender pay gap in some contexts, persistent occupational segregation—especially in STEM fields—limits women’s gains [35,55,56,57,58]. Social-economy initiatives also matter: Castro Núñez et al. [59] find that firms in Spain’s social-economy sector—where R&D is often community-oriented—achieve more stable employment for women and reduce the “glass ceiling”, offering a potential model for R&D policies aimed at closing gender gaps. Moreover, these structural shifts, driven by technological change and other factors like trade and outsourcing [60], can generate short-term unemployment [9,21,61] and require workforce retraining alongside an evolving demand for skills [62,63].

Taken together, these strands of literature [61,62] reveal that R&D’s impacts are multifaceted—shaped by timing, sector, region, institutional support, and workforce composition—and underscore the need for an empirical analysis that disaggregates effects by sector and lag structure within the EU context. Against this backdrop, the following hypotheses translate the study’s overarching objective into empirically testable claims:

H1. 

An increase in domestic expenditure on research and development (GERD), expressed as a percentage of GDP, by sector leads to improvements in social indicators (female and male unemployment, gender pay gap, healthy life at birth) and economic indicators (GDP per capita, inward and outward FDI flows).

H2. 

The impact of research and development expenditures varies not only by sector (business vs. government) but also by the length of the analyzed period. Sector-specific effects may differ in intensity and dynamics, and the effects of government R&D expenditures may manifest with greater delay and complexity due to the specific nature of their social objectives.

3. Materials and Methods

The empirical strategy involved estimating both fixed-effects and random-effects panel models for all European Union countries, based on Eurostat data for the period 2013–2022. The balanced panel consists of 270 observations (27 EU member states over 10 years). The independent variables included gross domestic expenditure on R&D (GERD) as a percentage of GDP, analyzed in three variants: total GERD (all sectors: business enterprise, government, higher education and non-governmental), GERD in the business enterprise sector, and GERD in the government sector. The dependent variables comprised female unemployment, male unemployment, the gender pay gap, and healthy life expectancy at birth. The selection of unemployment rates, wage inequality (proxied by the gender pay gap), and GDP per capita as key outcome variables reflects the multidimensional nature of sustainability, encompassing economic vitality, social cohesion, and equitable opportunity. These indicators, widely adopted in cross-national studies, enable robust comparative analysis and capture core elements of the Sustainable Development Goals, particularly those addressing decent work, reduced inequalities, and sustained, inclusive economic growth. All estimations were carried out in R (version 4.4.2) using the plm package (version 2.6-4).

The choice of lag lengths for the R&D variables was informed by established empirical evidence, which suggests that the effects of innovation typically emerge after a delay of 1–3 years for economic outcomes and 2–5 years for labour market indicators. Accordingly, distributed lag models with up to four years of lags were estimated, and robustness checks confirmed the stability of the main results. Model selection between fixed-effects and random-effects specifications was conducted using the Hausman specification test.

The analytical framework leverages the advantages of panel data covering all EU countries over a relatively long period, enabling the study to analyze the dynamics of change over time while accounting for unobserved heterogeneity across units. Unobserved heterogeneity refers to persistent differences between countries that are not captured by the observed variables but may influence the outcomes of interest. If such unmeasured factors are correlated with the explanatory variables, they could result in biased or inconsistent estimates if not properly addressed.

Therefore, panel data methods—specifically, fixed-effects and random-effects models—are particularly appropriate in this context. The fixed-effects model controls for unobserved heterogeneity by focusing on within-country variation over time, while the random-effects model also accounts for unobserved heterogeneity, assuming that these effects are random and uncorrelated with the explanatory variables. This approach facilitates the simultaneous analysis of both between-country differences and within-country dynamics, thereby enhancing the efficiency and interpretability of the estimation.

To assess the effects of R&D expenditure on social and economic outcomes in the EU countries (2013–2022), we estimated both fixed-effects (FEs) and random-effects (REs) panel models for each selected dependent variable using Eurostat data. This dual-modelling approach allows for robust comparison and ensures that the chosen specification best captures the relationship under study.

When both FEs and REs specifications are estimated, the Hausman consistency test is used to determine the most appropriate estimator for each equation. In our analysis of seven outcomes, the test favours the FEs estimator for GDP per capita, male unemployment, and the gender pay gap, so these results are discussed using fixed-effects coefficients. For healthy life expectancy and female unemployment, the test is inconclusive, but the main results remain robust to either estimator, as the signs and significance of the coefficients do not change. For inward and outward FDI (as a share of GDP), the test finds no systematic difference between FEs and REs, which supports the use of the more efficient random-effects specification for these models.

To test the study hypotheses, the analysis used both fixed-effects and random-effects models to estimate the impact of R&D expenditure on the standardized values of each dependent variable, following standard panel-data procedures. The overall approach ensures that the results are robust to different modelling assumptions and reflect both the within- and between-country variation present in the data.

The general form of the model is as follows:

$${\mathrm{y}}_{it}=\mathsf{\alpha }+\mathsf{\beta }{\mathrm{X}}_{it}+{\mathsf{\mu }}_i+{\mathsf{\nu }}_{it}$$

(1)

where

y_it—dependent variable for unit i at time t;

X_it—matrix of independent variables;

β—coefficient vector;

α—intercept (common constant term);

µ_i—individual effect specific to unit i (random and uncorrelated with X_it);

ν_it—a random component that varies over time and between units.

The β coefficients in the random-effects model are interpreted as the influence of the independent variables X_it on the dependent variable y_it, considering the random effect µ_i.

In contrast to the random-effects model, the fixed-effects model assumes that the individual effects (µ_i) can be correlated with the observed independent variables (X_it). The general form of the model is as follows:

$$y_{it}=\mathsf{\alpha }+\mathsf{\beta }{X}_{it}+{\mathsf{\mu }}_i+{\mathsf{\nu }}_{it}$$

(2)

The form of the model is identical to that of the random-effects model, but to eliminate time-constant individual effects (µ_i), a transformation of deviations from the mean values within an individual is applied:

$$y_{it}=\mathsf{\beta }\hspace{0.17em}\widetilde{X_{it}}+{\tilde{\mathsf{\nu }}}_{it}$$

(3)

where $\widetilde{y_{it}}=y_{it}-\overline{y_i}$, $\widetilde{X_{it}}=X_{it}-\overline{X_i}$ and ${\tilde{\nu }}_{it}={\nu }_{it}-{\overline{\nu }}_i$.

This transformation eliminates the time-constant individual effects µ_i, which allows for a precise estimation of the influence of the independent variables X_it on the dependent variable y_it.

All dependent variables are standardized, which enables the β coefficients to be interpreted as the change in the number of standard deviations of the dependent variable per unit change in the independent variable. This approach avoids ambiguity associated with direct interpretation in percentage points, particularly in international panel data and cross-country comparisons. Standardizing the outcomes enhances comparability between models and ensures clarity in interpreting the size and direction of effects. This standardization facilitates the comparison of the effects of different variables both within and across models.

Fixed-effects models are employed to control for unobserved variables that are specific to individuals and remain constant over time, whereas random-effects models are appropriate when individual effects are assumed to be uncorrelated with other variables in the model. The combined use of both approaches allows for a comprehensive analysis of the influence of independent variables on dependent variables, while also accounting for the unique characteristics of panel data.

It should be noted that, like most macro-panel analyses, the approach adopted here primarily identifies robust associations rather than definitive causal relationships.

Additionally, the macro-panel design does not enable the explicit separation of the indirect or spillover effects arising from the interdependencies between public and business R&D sectors. For example, business R&D performance may reflect advantages conferred by the broader institutional and scientific environment created by public investment. Accordingly, the empirical strategy here is best suited to identify robust sectoral patterns at the aggregate level, but not the underlying causal channels or the full sequence of mediating mechanisms. While fixed-effects models help control for unobserved heterogeneity and temporal ordering supports causal inference, the risk of omitted variable bias and reverse causality remains. Therefore, the results should be interpreted as strong correlations, and additional methods (e.g., instrumental variable approaches or natural experiments) would be needed to establish causality with higher confidence.

Comprehensive robustness checks and diagnostic tests were conducted to ensure the validity of the results. The calculations were performed using R version 4.4.2 with the plm package version 2.6-4. Recognizing the methodological challenges inherent in panel data analysis—such as endogeneity, omitted variable bias, and cross-sectional dependence—this study employs a range of estimation techniques, including fixed-effects, random-effects, and dynamic panel models. All statistical procedures and robustness checks are described in detail in the Materials and Methods section, ensuring transparency and reproducibility.

4. Results

This section presents the results of a panel study of EU countries (2013–2022), based on Eurostat data, examining the relationship between selected independent variables related to R&D expenditure and key social and economic outcomes, using both fixed-effects and random-effects models. Overall, these findings indicate that the relationship between unobserved country characteristics and the explanatory variables is outcome-specific and can vary depending on the dependent variable analyzed. In some cases, this correlation is particularly strong—such as for GDP growth, male unemployment, and the gender pay gap—supporting the use of the fixed-effects estimator for these outcomes. In other cases, such as for the FDI equations, the assumptions underlying the random-effects model are more appropriate. This variation highlights the need to tailor model selection to each equation, ensuring that policy conclusions drawn from the analysis are valid and appropriately contextualized.

The results indicate that business R&D expenditure consistently drives positive economic and social outcomes in the EU, while government R&D demonstrates more complex and sometimes adverse effects, particularly regarding the gender pay gap and long-term GDP growth.

Among the social variables, the first outcome analyzed is female unemployment. The results of the fixed-effects and random-effects models for the impact of GERD on female unemployment are presented in

Table 1. Fixed- and random-effects estimates for female unemployment in the EU, 2013–2022.

Independent Variable	Lag	Estimate (FE)	Pr (>|z|) (FE)	Estimate (RE)	Pr (>|z|) (RE)
GERD as a percentage of GDP for all sectors	L1	−0.788	0.000 ***	−0.535	0.000 ***
	L2	−0.563	0.001 ***	−0.4	0.002 ***
	L3	−0.293	0.098 ***	−0.242	0.056 *
	L4	−0.293	0.069 *	−0.234	0.047 **
GERD as a percentage of GDP for government sector	L1	1.987	0.024 **	1.76	0.026 **
	L2	1.646	0.038 **	1.489	0.039 **
	L3	1.370	0.048 **	1.259	0.049 **
	L4	0.613	0.295	0.594	0.281
GERD as a percentage of GDP for business enterprise sector	L1	−1.152	0.000 ***	−0.824	0.000 ***
	L2	−0.839	0.000 ***	−0.623	0.000 ***
	L3	−0.493	0.021 **	−0.407	0.009 ***
	L4	−0.339	0.087 *	−0.308	0.037 **

Source: own elaboration. Asterisks denote statistical significance (* p < 0.10, ** p < 0.05, *** p < 0.01).

below.

Within the set of social variables, female unemployment is examined as the initial outcome. The results indicate that a one-percentage-point increase in business-sector R&D expenditure as a share of GDP is associated with a statistically significant reduction in female unemployment by approximately 1.15 percentage points in the first year (FE = −1.152, p < 0.01). This effect remains statistically significant, although its magnitude gradually decreases in subsequent years. In contrast, a one-percentage-point increase in government-sector R&D expenditure as a share of GDP leads to a significant rise in female unemployment by about 1.99 percentage points in the first year (FE = 1.987, p = 0.024). This initial adverse effect—likely related to labour market adjustments such as automation or restructuring resulting from new public-sector R&D initiatives—gradually diminishes and becomes statistically insignificant after two to three years. The estimated coefficients from both the fixed-effects and random-effects models, illustrating the impact of GERD on female unemployment, are shown in Figure 1.

Analysis of the results reveals that an increase in research and development expenditure (GERD) affects female unemployment in a sector-dependent manner. Studies have shown that an increase in GERD across all sectors leads to a significant reduction in female unemployment, particularly in the short term, although this effect becomes less pronounced over time. In the business enterprise sector, this effect is the strongest, as an inc.rease in R&D expenditure results in a sharp decrease in female unemployment in the short term; even though the intensity of this effect diminishes over time, it remains statistically significant in the long term. In contrast, in the government sector, an increase in GERD is associated with an increase in female unemployment in the short and medium term, with this negative effect disappearing in the long term.

Male unemployment constitutes the second social variable under analysis. The results of the fixed-effects and random-effects models for the impact of GERD on female unemployment are presented in

Table 2. Fixed- and random-effects estimates for male unemployment in the EU, 2013–2022.

Independent Variable	Lag	Estimate (FE)	Pr (>|z|) (FE)	Estimate (RE)	Pr (>|z|) (RE)
GERD as a percentage of GDP for all sectors	L1	−0.993	0.000 ***	−0.497	0.000 ***
	L2	−0.665	0.003 ***	−0.360	0.005 ***
	L3	−0.283	0.191	−0.206	0.095 ***
	L4	−0.383	0.043 **	−0.231	0.044 **
GERD as a percentage of GDP for government sector	L1	1.924	0.092 ***	1.291	0.167
	L2	1.780	0.075 *	1.271	0.129
	L3	1.699	0.044 **	1.291	0.077 ***
	L4	0.411	0.551	0.330	0.593
GERD as a percentage of GDP for business enterprise sector	L1	−1.450	0.000 ***	−0.773	0.000 ***
	L2	−1.063	0.000 ***	−0.595	0.000 ***
	L3	−0.600	0.021 **	−0.391	0.012 **
	L4	−0.475	0.041 **	−0.328	0.024 **

Source: own elaboration. Asterisks denote statistical significance (* p < 0.10, ** p < 0.05, *** p < 0.01).

below.

The results indicate that increases in business R&D expenditure lead to persistent and statistically significant decreases in male unemployment across all time horizons. In contrast, government R&D does not exhibit a statistically significant effect on male unemployment.

The results show that a one-percentage-point increase in business R&D expenditure (as a share of GDP) is associated with a reduction in male unemployment of approximately 1.0–1.1 percentage points in the short term (Lag 1, p = 0.011), with significant effects persisting up to three years. By contrast, government R&D expenditure does not exhibit a statistically significant effect on male unemployment across any of the estimated lag periods.

The estimated coefficients from both the fixed-effects and random-effects models, illustrating the impact of GERD on male unemployment, are presented in Figure 2.

In the analysis of male employment, it was observed that GERD expenditure across all sectors has a positive effect in reducing male unemployment in the short term, although its impact diminishes over the long term. In contrast, R&D expenditure in the corporate sector shows a strong and lasting effect on reducing male unemployment across all periods examined, whereas government R&D expenditure has no significant impact on the level of male unemployment.

The next social variable examined is the gender pay gap. The results of the fixed-effects and random-effects models for the impact of GERD on gender pay gap are presented in

Table 3. Fixed- and random-effects estimates for gender pay gap in the EU, 2013–2022.

Independent Variable	Lag	Estimate (FE)	Pr (>|z|) (FE)	Estimate (RE)	Pr (>|z|) (RE)
GERD as a percentage of GDP for all sectors	L1	−0.302	0.023 *	−0.137	0.222
	L2	−0.032	0.824	0.067	0.572
	L3	0.315	0.054 *	0.301	0.019 **
	L4	0.649	0.000 ***	0.499	0.000 ***
GERD as a percentage of GDP for government sector	L1	0.511	0.410	0.551	0.351
	L2	1.283	0.040 **	1.245	0.035
	L3	1.928	0.002 ***	1.828	0.002 ***
	L4	1.801	0.005 ***	1.704	0.004 ***
GERD as a percentage of GDP for business enterprise sector	L1	−0.548	0.000 ***	−0.35	0.011 **
	L2	−0.338	0.051 *	−0.165	0.263
	L3	−0.052	0.792	0.063	0.695
	L4	0.305	0.175	0.3	0.082 ***

Source: own elaboration. Asterisks denote statistical significance (* p < 0.10, ** p < 0.05, *** p < 0.01).

below.

The analysis reveals that government R&D expenditure exerts the strongest impact on this indicator, significantly widening the gender pay gap in the medium and long term. Specifically, a one-unit increase in government GERD is associated with an increase in the pay gap of 1.28 units after a two-year lag (FE = 1.283, RE = 1.245), 1.93 units after three years (FE = 1.928, RE = 1.828), and 1.80 units after four years (FE = 1.801, RE = 1.704); all of these effects are statistically significant. In the short term (Lag 1), the effect is positive but not statistically significant. These findings suggest that the longer the time horizon, the stronger the impact of government R&D expenditure on gender pay inequality.

In contrast, business R&D expenditure exhibits a favourable but short-lived effect, reducing the gender pay gap in the short term (Lag 1: FE = −0.350), although this effect is only statistically significant at the 10% level. For longer lags, the coefficients remain negative but are not statistically significant, indicating that the impact diminishes over time.

For total R&D spending across all sectors, the effect on the gender pay gap is less pronounced. A significant increase in the gap is observed only in the long term, with a four-year lag, where a one-unit rise in GERD corresponds to a 0.649-unit increase in the pay gap. These results indicate that, among all sectors, government R&D expenditure has the most pronounced and persistent effect on widening the gender pay gap, while business R&D’s favourable impact is only evident in the short term.

The estimated coefficients from both the fixed-effects and random-effects models, illustrating the impact of GERD on the gender pay gap, are shown in Figure 3.

The results presented in Figure 3 clearly demonstrate distinct sectoral patterns in the relationship between R&D expenditure and the gender pay gap. Notably, business sector R&D investment is associated with a marked short-term reduction in the gender pay gap, particularly visible in the first year; however, this effect rapidly diminishes and is not sustained in subsequent years. In contrast, government-sector R&D expenditure exhibits a consistently strong and persistent increase in the gender pay gap, with the most pronounced effects emerging from the third year onward. The combined effect across all sectors is relatively weak and ambiguous compared to the distinct trends observed within individual sectors.

The final social variable considered is healthy life expectancy at birth. The results of the fixed-effects and random-effects models for the impact of GERD on the gender pay gap are presented in

Table 4. Fixed- and random-effects estimates for healthy life at birth in the EU, 2013–2022.

Independent Variable	Lag	Estimate (FE)	Pr (>|z|) (FE)	Estimate (RE)	Pr (>|z|) (RE)
GERD as a percentage of GDP for all sectors	L1	0.349	0.033 **	0.256	0.049 **
	L2	0.106	0.555	0.105	0.448
	L3	−0.261	0.244	−0.082	0.601
	L4	−0.645	0.018 **	−0.198	0.244
GERD as a percentage of GDP for government sector	L1	0.146	0.852	0.007	0.992
	L2	−0.661	0.412	−0.677	0.365
	L3	−0.351	0.689	−0.436	0.589
	L4	−0.111	0.912	−0.248	0.782
GERD as a percentage of GDP for business enterprise sector	L1	0.352	0.078 *	0.305	0.060 *
	L2	0.043	0.844	0.113	0.513
	L3	−0.490	0.069 *	−0.157	0.421
	L4	−1.000	0.003 ***	−0.301	0.165

Source: own elaboration. Asterisks denote statistical significance (* p < 0.10, ** p < 0.05, *** p < 0.01).

below.

The analysis indicates that R&D expenditure across all sectors produces a modest, short-term positive effect on healthy life expectancy; however, these gains tend to diminish or become statistically insignificant over longer time periods. Specifically, a one-unit increase in total GERD is associated with a statistically significant increase in healthy life expectancy of 0.12 units after a one-year lag (FE = 0.12, p = 0.047), while the effects at longer lags (two to four years) are positive but not statistically significant.

Disaggregating by sector, business R&D exhibits a positive effect on healthy life expectancy in the short term (Lag 1: FE = 0.08, p = 0.089), though this relationship does not retain statistical significance in the medium or long term. Government R&D expenditure shows no consistent or significant impact on healthy life expectancy at any lag.

The estimated coefficients from both the fixed-effects and random-effects models, reflecting the effects of GERD on healthy life expectancy at birth, are presented in Figure 4.

The analysis demonstrates that total R&D expenditure yields a modest and statistically significant improvement in healthy life at birth in the short term; however, these positive effects fade and lose significance over longer periods. Sectoral breakdown reveals that business R&D has a similar short-lived positive association, while government R&D shows no consistent or significant impact on healthy life expectancy at any time horizon.

GDP per capita is the first economic variable examined in the analysis. The results of the fixed-effects and random-effects models for the impact of GERD on the gender pay gap are presented in

Table 5. Fixed- and random-effects estimates for GDP per capita in the EU, 2013–2022.

Independent Variable	Lag	Estimate (FE)	Pr (>|z|) (FE)	Estimate (RE)	Pr (>|z|) (RE)
GERD as a percentage of GDP for all sectors	L1	0.769	0.000 ***	0.715	0.000 ***
	L2	0.778	0.000 ***	0.715	0.000 ***
	L3	0.437	0.024 **	0.549	0.000 ***
	L4	0.109	0.617	0.435	0.001 ***
GERD as a percentage of GDP for government sector	L1	−1.325	0.066	−0.942	0.164
	L2	−1.326	0.072 *	−0.927	0.179
	L3	−2.059	0.007 ***	−1.559	0.027
	L4	−1.971	0.012 **	−1.464	0.044 **
GERD as a percentage of GDP for business enterprise sector	L1	1.155	0.000 ***	1.031	0.000 ***
	L2	1.195	0.000 ***	1.042	0.000 ***
	L3	0.958	0.000 ***	0.880	0.000 ***
	L4	0.643	0.015 **	0.728	0.000 ***

Source: own elaboration. Asterisks denote statistical significance (* p < 0.10, ** p < 0.05, *** p < 0.01).

below.

The empirical results indicate that business R&D expenditure constitutes the principal driver of GDP growth within the European Union, yielding consistently positive and robust effects. By contrast, government R&D expenditure is associated with a negative impact on GDP per capita in the medium and long term.

The strongest and most lasting positive effect on GDP per capita is observed for R&D financing in the business enterprise sector, as expected. In contrast, the findings regarding the impact of government R&D expenditures are surprising, as they reveal a negative effect that intensifies over time. In the first (Lag 1) and second year (Lag 2) following government R&D spending, the effect in the fixed-effects (FEs) model is close to statistical significance (p = 0.066 and p = 0.072), while in the random-effects (REs) model it remains statistically insignificant (p = 0.164 and p = 0.179). However, in the third and fourth years (Lag 3 and Lag 4), the effect becomes clearly negative and statistically significant in both models. Specifically, in the FEs model, the coefficient is −2.059 (p = 0.007) for Lag 3 and −1.971 (p = 0.012) for Lag 4, while in the REs model the coefficients are −1.559 (p = 0.027) and −1.464 (p = 0.044), respectively. This indicates a long-term tendency for GDP per capita to decline as government R&D spending increases.

The estimated coefficients derived from both the fixed-effects and random-effects models, reflecting the influence of GERD on GDP per capita, are presented in Figure 5.

The analysis reveals that R&D investments in the business enterprise sector exert the strongest and most persistent positive influence on GDP per capita. In contrast, government R&D spending demonstrates an increasingly negative impact over time. While the initial effects of government R&D are weak and only marginally significant, both models show a clearly negative and statistically significant relationship in the third and fourth years following the expenditure. These results underscore a long-term tendency for increased government R&D funding to be associated with declines in GDP per capita, highlighting a sharp divergence between the outcomes of public and business-driven R&D investments.

The next economic variable analyzed is the net inflow of inward FDI as a percentage of GDP. The estimated coefficients from both the fixed-effects and random-effects models, reflecting the effects of GERD on net inflows of inward FDI (% of GDP), are presented in

Table 6. Fixed- and random-effects estimates GERD on net inflows of inward FDI (% of GDP) in the EU, 2013–2022.

Independent Variable	Lag	Estimate (FE)	Pr (>|z|) (FE)	Estimate (RE)	Pr (>|z|) (RE)
GERD as a percentage of GDP for all sectors	L1	0.091	0.308	0.046	0.582
	L2	0.060	0.579	0.005	0.960
	L3	0.029	0.824	−0.036	0.750
	L4	0.012	0.926	−0.051	0.653
GERD as a percentage of GDP for government sector	L1	1.147	0.002 ***	1.135	0.002 ***
	L2	1.200	0.005 ***	1.193	0.004 ***
	L3	1.087	0.020 **	1.093	0.016 **
	L4	0.850	0.081 *	0.869	0.063 *
GERD as a percentage of GDP for business enterprise sector	L1	0.091	0.412	0.039	0.705
	L2	0.046	0.731	−0.016	0.897
	L3	−0.021	0.895	−0.084	0.543
	L4	−0.031	0.842	−0.093	0.497

Source: own elaboration. Asterisks denote statistical significance (* p < 0.10, ** p < 0.05, *** p < 0.01).

.

The results show that government R&D expenditure exerts a positive and statistically significant effect on inward FDI in the short and medium term. Specifically, a one-unit increase in government GERD is associated with an increase in inward FDI of 0.61 percentage points of GDP after a one-year lag (FE = 0.61, p = 0.041) and 0.49 percentage points after a two-year lag (FE = 0.49, p = 0.049). However, this effect diminishes and loses statistical significance at longer lags (Lag 3 and Lag 4).

The estimated coefficients from both the fixed-effects and random-effects models, illustrating the impact of GERD on inward FDI, are presented in Figure 6.

In practical terms, even a modest rise in public R&D investment can trigger a noticeable surge in foreign investment inflows, strengthening the country’s position as an attractive destination for international capital. This short-term effect underscores the strategic importance of government R&D spending not only for innovation, but also for enhancing the country’s integration into the global economy.

In contrast, neither business R&D nor total R&D expenditure exhibit a statistically significant effect on inward FDI at any lag.

The final economic variable considered is the net outflow of outward FDI as a percentage of GDP—that is, the total value of capital that investors from the analyzed countries invest abroad as direct investments. The estimated coefficients from both the fixed-effects and random-effects models, reflecting the effects of GERD on net inflows of outward FDI (% of GDP), are presented in

Table 7. Fixed- and random-effects estimates for GERD on net inflows of outward FDI (% of GDP) in the EU, 2013–2022.

Independent Variable	Lag	Estimate (FE)	Pr (>|z|) (FE)	Estimate (RE)	Pr (>|z|) (RE)
GERD as a percentage of GDP for all sectors	L1	0.066	0.414	0.040	0.604
	L2	0.035	0.723	0.004	0.966
	L3	0.011	0.924	−0.024	0.819
	L4	0.003	0.982	−0.030	0.770
GERD as a percentage of GDP for government sector	L1	1.044	0.002 ***	1.056	0.002 ***
	L2	1.097	0.005 ***	1.118	0.003 ***
	L3	0.992	0.019 **	1.027	0.012 **
	L4	0.781	0.068 **	0.832	0.044 **
GERD as a percentage of GDP for business enterprise sector	L1	0.063	0.534	0.032	0.740
	L2	0.017	0.889	−0.018	0.871
	L3	−0.040	0.779	−0.072	0.569
	L4	−0.048	0.732	−0.077	0.535

Source: own elaboration. Asterisks denote statistical significance (** p < 0.05, *** p < 0.01).

.

The results indicate that government R&D expenditure is consistently associated with an increase in outward FDI across all time lags. For example, a one-unit increase in government GERD leads to an increase in outward FDI by 1.044 units after one year (FE = 1.044, p = 0.002; RE = 1.056, p = 0.002), by 1.097 units after two years (FE = 1.097, p = 0.005; RE = 1.118, p = 0.003), and by 0.992 units after three years (FE = 0.992, p = 0.019; RE = 1.027, p = 0.012). After four years (Lag 4), the effect remains positive but becomes marginally significant in the FE model (0.781, p = 0.068) and remains statistically significant in the RE model (0.832, p = 0.044).

The estimated coefficients from both the fixed-effects and random-effects models, illustrating the impact of GERD on outward FDI, are presented in Figure 7.

These findings demonstrate a persistent and robust relationship between government R&D spending and the international expansion of domestic firms.

This persistent association suggests that higher government R&D spending may stimulate domestic firms’ international expansion, facilitating knowledge transfer, access to new markets, and integration with global production networks. Such outward investment, while reflecting capital flows abroad, can create long-term opportunities for technological upgrading, global partnerships, and the strengthening of domestic innovation capacity.

In contrast, neither business R&D nor total R&D expenditure demonstrates a statistically significant effect on outward FDI at any lag.

To further assess the robustness of the obtained results and to illustrate the consistency of the estimated effects across different model specifications and sectors, Figure 8 presents the confidence intervals for female and male unemployment using both fixed- and random-effects models, separately for the government and business enterprise sectors.

The results display broadly consistent patterns regardless of model type, sector, or lag length. Overall, these findings suggest that the relationship between R&D expenditure and unemployment by gender remains robust for the choice of estimation method. However, the analysis also highlights sectoral differences and the increasing uncertainty of estimates for government R&D with longer lags, which may reflect the more complex and delayed impacts of public R&D investment.

A more detailed comparison of the fixed-effects and random-effects models yields the following findings:

R&D spending by companies had the strongest and most immediate impact on reducing both female and male unemployment. Although it was particularly effective in the short term, its effectiveness diminished over the longer term.

Government spending on R&D had a negative impact on female employment in the short and medium term, though this effect diminished in the long term. For male employment, the impact of government R&D spending was statistically insignificant.

R&D spending across all sectors had a positive impact on both female and male employment in the short term; however, its effectiveness gradually declined over time, remaining statistically significant in the long term only in the random-effects model.

Regarding GDP per capita, both models highlight the key role of the business enterprise sector in driving GDP growth, with the random-effects model emphasizing long-term efficiency. In contrast, GERD in the government sector exhibits a negative impact on GDP in both models, especially in the medium and long term.

As for FDI, in both models, GERD in the government sector positively influences FDI inflows and outflows in the short and medium term, but this effect disappears in the long term. In the case of R&D expenditures covering all sectors and those in the business enterprise sector, no significant impact on FDI is observed in either model.

These findings demonstrate that the fixed-effects model tends to emphasize short-term differences between sectors, whereas the random-effects model highlights the persistence of GERD effects in the corporate sector with respect to GDP per capita. Additionally, the random-effects model identifies greater long-term efficiency of government investment in FDI compared to the fixed-effects model. Overall, both models underscore the critical role of the business enterprise sector in driving economic growth, while the government sector appears to be more effective in providing short-term support for FDI.

In summary, the findings reveal a clear divide between the positive, growth-promoting role of business R&D expenditure and the more ambiguous or even adverse effects of government R&D expenses, particularly on gender equality and GDP growth. This highlights the necessity of sector-sensitive R&D policies.

5. Discussion

The analysis reveals intricate relationships between R&D spending and socio-economic performance in EU countries, confirming some of the existing literature while broadening the perspective through an examination of sectoral and temporal differences. The empirical evidence indicates that the effects of R&D expenditure on labour market and social outcomes are highly heterogeneous across sectors and time periods. Business-sector R&D is associated with more immediate gains in employment and wage equality, whereas public-sector R&D tends to generate longer-term benefits that are often mediated through broader social and institutional channels. Such results underscore the importance of adopting integrated policy approaches that leverage R&D not only for economic competitiveness, but also for advancing social inclusion, gender equality, and institutional sustainability—fundamental pillars of sustainable development.

This pattern—showing an initial positive impact on employment, followed by a subsequent reduction as automation progresses—is consistent with Vivarelli’s compensation theory [22], according to which the early employment gains from technological change may eventually be offset by the substitution of human labour with technology. Gender-specific effects further enrich the discussion. The greater reduction in female unemployment compared to male unemployment, particularly driven by R&D in the business sector, reflects a structural shift toward service-oriented occupations where women are better represented [35,52,55], a trend potentially also linked to innovation within the service sector itself [64]. However, the limited impact of government R&D on male unemployment suggests that policy interventions may need to address sector-specific barriers, such as skills mismatches in declining industries.

Notably, the statistically significant association between public R&D expenditure and female unemployment—but not male unemployment—can be attributed to structural differences in the labour market. Women are overrepresented in public sector-financed industries such as education, healthcare, and public administration, which are direct beneficiaries of government R&D projects [50,51]. Public investments in these sectors frequently lead to restructuring, digitization, or automation, which may temporarily increase the risk of job loss among existing employees—primarily women [6,7,52,56,57]. For men, who are more often employed in private or industrial sectors less directly affected by public R&D, such effects are more diffuse and do not reach statistical significance in macroeconomic analysis [9,48,55]. This is consistent with recent studies on occupational segregation by gender and the impact of public policy on female employment [50].

The sectoral analysis in the study reveals significant discrepancies within the existing literature. The dominance of the corporate sector in driving GDP growth is consistent with Bogliacino et al. [18] and other studies on innovation and firm performance [65], which found that private firms lead in commercializing innovation, particularly in technology-intensive industries. In contrast, while government R&D expenditure does not contribute positively to GDP per capita in the medium and long term, it also appears to have limited effectiveness—or even adverse effects—in promoting gender equity, as reflected in the observed widening of the gender pay gap. This observation is consistent with findings from OECD [31] and Ravšelj and Aristovnik [32], who argue that public investment in R&D often prioritizes broader societal objectives but may not always achieve the intended equity outcomes.

Taken together, these mechanisms—documented in both empirical studies and theoretical literature—help explain the negative effects of government R&D spending observed in our results. Addressing these issues requires targeted policy interventions to improve public-sector efficiency, strengthen links with industry, and ensure that public investments in R&D contribute to both inclusive and sustainable economic growth.

The limited commercial orientation or inefficiencies associated with certain public sector research investments may help explain these patterns. As highlighted by David et al. [13], public R&D often fails to generate immediate productivity gains unless effectively complemented by private sector innovation efforts, and the effectiveness and value of public R&D support can be variable [34]. This suggests that the mere allocation of public funds to R&D is insufficient to drive economic growth and social progress without strong linkages to the private sector. For instance, public funding in sectors such as healthcare or education may favour female employment in non-routine service-oriented jobs [57,58,66], which could explain a stronger reduction in the gender pay gap. This is in line with Cortes and Pan’s [51] observation that automation disproportionately displaces male-dominated routine jobs, while women continue to work in occupations that are less susceptible to automation [67].

However, while such sector-specific effects of public funding may temporarily reduce gender pay disparities by increasing female employment in non-routine, service-oriented jobs, our aggregate and long-term results indicate that government R&D expenditure is overall associated with an increase in the gender pay gap. This apparent contradiction likely reflects the low representation of women in high-wage public R&D positions, particularly in STEM fields, which drives the overall gap upward at the macroeconomic level despite positive effects in select sectors.

A key limitation of this study, and an important caveat when interpreting the results, is its fundamentally macroeconomic and correlational approach, rather than a causal one. The adopted conceptual model compares the effects of public and business R&D expenditures but does not allow for disentangling the full impact of all institutional, educational, and regulatory factors that indirectly shape the outcomes observed in the business sector. Some of the positive effects attributed to business R&D may actually reflect the benefits derived by the private sector from a broad range of public goods and services, such as an educated workforce, infrastructure, basic research, and a stable regulatory environment [13,15]. Consequently, the findings of this research should be interpreted primarily as evidence of macro-level associations rather than definitive causal relationships specific to each sector. This limitation stems from the fact that firms often benefit from public investments in areas such as education, basic research, regulatory frameworks, fiscal and monetary policy, or welfare systems—factors shaped by government action that cannot be fully disentangled in aggregate-level analyses. The use of national-level data, while valuable for cross-country comparison, involves averaging that may obscure important sectoral, regional, or group-specific differences. Moreover, it constrains the ability to identify the complex interdependencies between public policies and business innovation, which are likely to differ across sectors and institutional contexts The temporal dynamics of government R&D-related FDI inflows are also noteworthy. The short-term attractiveness of FDI is consistent with Sasidharan and Kathuria’s results [40], which showed that government R&D expenditures positively influence the ability of foreign firms to invest. However, maintaining high FDI growth rates is more complex, as it is also linked to factors beyond R&D, such as infrastructure upgrades and regulatory reforms.

Furthermore, the authors’ study challenges some previous findings. For example, the negative correlation between government R&D spending and GDP in later periods contradicts the research of Bilbao-Osorio and Rodríguez-Pose [14], who emphasized the growth potential of public R&D spending. This discrepancy may stem from methodological differences; whereas earlier studies focused on aggregate R&D spending, the authors’ sectoral breakdown reveals that government spending often targets long-term social goals (e.g., health, equality) with delayed economic returns. Similarly, the observed lack of impact—or even a short-term widening—of the gender pay gap due to corporate R&D contradicts the findings of Goos et al. [50], who linked automation to increased pay polarization between men and women. This may reflect sectoral segregation, as Blau and Kahn [35] noted that women remain underrepresented in high-paying STEM fields funded by corporate R&D, limiting their access to its benefits.

The long-term negative impact of government R&D expenditure on GDP per capita, as well as the increase in the gender pay gap, may result from a less efficient allocation of public funds for R&D. Such expenditures are often targeted at social objectives or long-term basic research, whose economic effects may only materialize after a considerable delay [13]. Furthermore, bureaucratic barriers, skills mismatches in the labour market, and structural delays in the commercialization of research outcomes can limit the short-term effectiveness of these investments

A further important distinction between government and business R&D expenditure lies in their functional roles within the innovation system. Government R&D funding is often directed towards basic research, which serves as a critical foundation for later-stage business R&D and commercial applications. The interaction and transfer of knowledge and resources between these sectors play a key role in the innovation process, with government-supported basic research frequently providing the groundwork for breakthroughs that are subsequently developed and brought to market by private enterprises. Recognizing these complementarities, future research could further explore the mechanisms and effectiveness of knowledge and resource transfer, including R&D cooperation [47], between public and private R&D activities, as well as their joint contributions to long-term socio-economic outcomes.

While this study applies distributed-lag models with up to four years of lags, it is important to acknowledge that the realization of R&D effects is inherently variable and context-dependent. The complexity of innovations, sectoral differences, market dynamics, regulatory environments, and other external factors may significantly moderate the timing and scale of the observed outcomes. As such, some effects of R&D—particularly in areas such as health or basic research—may only become evident over much longer periods, including several years or even decades. Therefore, the temporal lags used in this analysis reflect average patterns observed at the macroeconomic level across EU countries, rather than capturing the full range of possible project timelines. This limitation should be considered when interpreting the results, and future research is encouraged to explore longer-term effects and sector-specific dynamics. In addition to these considerations, a key limitation is that the corporate/business sector is treated as a single category, although it encompasses diverse industries with potentially distinct R&D dynamics and socio-economic effects. The use of aggregate sectoral data precludes analysis of intra-sectoral heterogeneity. Future research would benefit from employing more granular, industry-level R&D data to identify which segments of the corporate sector contribute most strongly to the observed macro-level outcomes [68].

Although this study focuses on the European Union, future research should extend comparative analyses to other advanced economies such as the USA, Japan, China, and South Korea. Cross-regional comparisons could reveal whether the observed sectoral and temporal patterns in R&D impacts are specific to the EU institutional context or more generalizable across different innovation systems.

To facilitate the interpretation of the main results,

Table 8. Effects of R&D expenditure by sector on major socioeconomic outcomes.

Dependent Variable	Business R&D	Government R&D	All Sectors (Total GERD)
Female unemployment	↓ (short and long term)	↑ (short, then fades)	↓ (short, then fades)
Male unemployment	↓ (short and long term)	–	↓ (short, then fades)
Gender pay gap	↓ (short term)	↑ (long term)	↑ (long term)
Healthy life expectancy	↑ (short term)	–	↑ (short, then fades)
GDP per capita	↑ (all lags)	↓ (long term)	↑ (short term)
Inward FDI	–	↑ (short/medium term)	–
Outward FDI	–	↑ (all lags)	–

Legend: ↑—positive effect; ↓—negative effect; –—no significant effect; (short/long term)—effect observed mainly in short/long time horizon; (fades)—effect diminishes over time. Source: own elaboration.

provides a qualitative summary of the sector-specific effects of R&D expenditure on the analyzed socioeconomic outcomes.

The findings suggest that a “one-size-fits-all” approach to R&D policy is insufficient. Resource allocation should be tailored not only to stimulate economic growth but also to prevent the widening of wage inequalities, ensuring that government R&D investments foster both competitiveness and labour market equity. In light of these results, policymakers are advised to consider sectoral specificity, prioritize support for female-intensive STEM fields, and promote public–private co-investment to maximize both economic and social returns. As emphasized in the recent literature [11], differences in education systems and the quality of digital infrastructure may further influence the effectiveness of R&D policies across countries. Moreover, government R&D often targets strategic or long-term objectives, such as healthcare, education, or social equity policies, that do not directly translate into immediate economic gains [22], and is typically managed in a more bureaucratic and less flexible environment than the private sector, potentially reducing the efficiency and commercial impact of public investments.

This study has several limitations. The analysis is restricted to EU member states and the 2013–2022 period, which may affect the generalizability of the results. The use of sectoral aggregate data does not capture potentially important differences within sectors or effects manifesting over longer time horizons. Future research should employ more granular, industry-level data and extended observation periods. Furthermore, the research period encompasses the COVID-19 pandemic, which may have introduced exceptional shocks to labour markets and R&D investment. Although year-specific fixed effects were included to control for such events, the specific impact of the pandemic could not be fully isolated. In addition, FDI inflows and outflows are shaped by a range of factors beyond R&D expenditures, including macroeconomic stability, institutional quality, market size, and the regulatory environment. The analysis presented here does not capture all possible determinants of FDI, which may vary significantly between countries. Consequently, the potential for omitted variable bias and cross-country heterogeneity represents a limitation of this study.

Despite certain research limitations, the presented findings offer a significant contribution to the understanding of how sectoral R&D expenditures affect key economic and social indicators. Practical implications suggest that decision-makers should consider sectoral specificity when designing innovation policies. Support for the private sector can foster rapid economic growth and reduce unemployment, while additional policy instruments are necessary in the case of the government sector to minimize negative effects such as a widening wage gap or declining GDP per capita.

From both a theoretical and public interest perspective, it is crucial to further explore additional themes present in the literature. This could involve analyzing the effectiveness of industrial policies—for example, supporting green technologies—the quality of digital infrastructure that determines R&D effectiveness [34], or differences in education systems affecting the supply of qualifications [11].

A key limitation of this research is its inability to fully disentangle the direct and indirect effects of government and business R&D expenditure. In the real-world innovation ecosystem, business-sector R&D builds upon a broad foundation of public investment—including education, basic science, regulatory frameworks, and infrastructure—provided by government and higher-education sectors. As a result, the positive effects attributed to business R&D may partly reflect public investments and enabling environments rather than purely firm-level activity. This interdependence is a well-recognized challenge in the empirical innovation literature [13,15]. The econometric models used here, based on macro-level panel data, cannot isolate these intertwined mechanisms. Our findings should thus be interpreted as the net sectoral effects, conditional on the existing institutional context, rather than as cleanly separated causal pathways. Addressing this issue would require a more granular analytical framework—such as structural equation modelling (SEM), network analysis, or the use of matched firm-level and policy data—capable of modelling spillover channels and mediating factors.

Future research should build on this work by integrating data on public goods provision, human capital, and sectoral linkages, as well as by exploiting natural experiments or longitudinal microdata. Such approaches would allow for more robust identification of causal effects and clarify the distinct contributions of public and business R&D investments to socioeconomic outcomes. Until then, sectoral comparisons such as those presented here should be interpreted with appropriate caution.

Future studies should test the sector–time–gender framework proposed here beyond the EU and investigate the effects of R&D expenditures in greater detail across industries and over longer periods. Comparative research involving non-EU countries [11,18,22,39,63,69], as well as deeper examination of the mediating roles of industrial policy, digital infrastructure, and education systems, would provide valuable insights into the mechanisms through which R&D investments affect socio-economic outcomes [11,22]. Addressing these directions will help disentangle complex causal pathways and further inform evidence-based innovation policy.

6. Conclusions

This study advances innovation economics theory by demonstrating that the socioeconomic effects of R&D expenditure in the European Union are determined by the sectoral origin of funding, the timing of observed effects, and gendered labour market structures. Drawing on the Sectoral Innovation Systems perspective, the analysis shows that business sector R&D generates immediate and robust gains for GDP growth and unemployment reduction, whereas government-sector R&D is associated with delayed, mission-oriented effects that may inadvertently widen the gender pay gap—a finding consistent with OECD [31] and Ravšelj and Aristovnik [32], who note that public R&D investment frequently prioritizes broader social objectives over short-term economic returns. Distributed-lag models reveal that business R&D effects are most pronounced within one to three years, while government R&D effects materialize after two to five years, highlighting the importance of timing in R&D effectiveness. Notably, the analysis provides the first EU-wide evidence that sector-specific R&D trajectories generate systematically different labour-market outcomes for women and men, thus addressing a gap in prior aggregate studies. The empirical results support both main hypotheses: (H1) increases in R&D expenditure as a share of GDP induce significant sectoral changes in key social and economic indicators, and (H2) these effects are highly sensitive to temporal dynamics.

Policymakers should differentiate R&D support strategies more precisely, taking into account sectoral specificities. The business sector should be supported by policies aimed at the rapid commercialization of research outcomes. In contrast, the government sector requires policies that mitigate negative social effects, for example, by investing in digital education, vocational retraining, and the development of infrastructure that facilitates cooperation between the public and business sectors.

By systematically examining the sectoral and temporal effects of R&D expenditure on economic and social outcomes, this study contributes to a deeper understanding of how innovation can support the transition to more sustainable societies. The results highlight the need for policy interventions that are sensitive to sectoral differences and that explicitly address the interplay between economic growth and social equity. Future research should further investigate the causal pathways linking R&D to the diverse dimensions of sustainability, making use of interdisciplinary frameworks and open data to inform policy and practice.