Transforming power of research and development on inequality and well-being: a European Union perspective within the circular economy framework

The paper addresses growing scholarly and policy interest in income inequality, its determinants, and consequences for innovation, technological progress, and development. Existing evidence on how redistribution and inequality affect innovation is mixed, and climate change is expected to exacerbate inequality, necessitating strategies that protect vulnerable groups while aligning environmental and development policies. The study positions circular economy (CE) and human development (HD) as interlinked concepts often criticized for missing each other’s social or environmental dimensions, with innovation central to both. The research fills a gap by jointly examining income, R&D potential, and human development across EU countries grouped by CE adoption level. The objective is to quantify and evaluate relationships between general income level (Earnings), R&D potential (researchers and GERD across sectors), and economic output (HDI) in EU-27 from 2010–2020, distinguishing countries with higher vs. lower CE use. The research questions are: RQ1: Is there a relation between the general level of income and the outputs of R&D potential among European countries with a higher and lower level of CE use? RQ2: Is there a relation between the outputs of R&D potential and economic development represented by the HDI among the European countries with a higher and lower level of CE use?

Prior work shows rising income inequality globally and mixed evidence on its relation to innovation. Some studies find innovation and R&D reduce inequality, while others show higher R&D spending can correlate with increased inequality, with results sensitive to geography, measures, and pre/post-tax definitions. Patent-based innovation indicators are widely used but have limitations due to strategic patenting and commercialization barriers. Environmental innovation literature indicates that reducing inequality, economic growth, and democratization support green technological innovation; inequality can inhibit carbon emission efficiency improvements and weaken the abatement effect of renewable technological innovation above certain inequality thresholds. Studies highlight institutional quality, IP rights, sectoral structure, and access to knowledge networks as mediators between R&D, innovation, and development. The CE concept is critiqued for lacking social dimensions, while HD is critiqued for insufficient environmental sustainability, motivating frameworks that integrate CE and HD. Overall, the literature motivates examining how income relates to R&D potential and how R&D links to HDI, especially under differing CE adoption levels across countries.

Design: Quantitative, comparative analysis using EU-27 panel data (2010–2020). The study proceeds in two stages: (1) cluster analysis to classify countries by CE use; (2) panel regression analyses to test RQ1 and RQ2. Data and variables: Sources include Eurostat and UNDP. Earnings: a composite income variable created by averaging standardized (0–1) values across 13 OECD/EUROSTAT net earnings categories by household type and earnings level, justified by high inter-correlation (r>0.9). R&D potential: two indicator groups measured for four sectors (Business enterprise, Government, Higher education, Private non-profit): (i) Researchers (full-time equivalent) per 100,000 inhabitants; (ii) GERD (gross domestic expenditure on R&D) in PPS million per 100,000 inhabitants. Economic development: Human Development Index (HDI) from UNDP. CE classification variables: (i) Patents related to recycling and secondary raw materials per 1,000,000 inhabitants; (ii) Circular material use rate. Sample: EU-27 countries: AUT, BEL, BGR, HRV, CYP, CZE, DNK, EST, FIN, FRA, DEU, GRC, HUN, IRL, ITA, LVA, LTU, LUX, MLT, NLD, POL, PRT, ROU, SVK, SVN, ESP, SWE. CE clustering: Country-level averages (2010–2020) of CE indicators were used. The silhouette method indicated two optimal clusters. Partitioning Around Medoids (PAM) with Manhattan distance formed clusters. Cluster 1 (higher CE use): Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Italy, Luxembourg, Netherlands, Poland, Slovenia, Spain. Cluster 2 (lower CE use): Bulgaria, Croatia, Cyprus, Greece, Hungary, Ireland, Latvia, Lithuania, Malta, Portugal, Romania, Slovakia, Sweden. Panel regression: Separate models were estimated by cluster. Model selection used F-tests for country and year effects, robust Hausman tests to choose between fixed (within) and random effects, and Breusch–Pagan tests for heteroscedasticity. Where needed, heteroscedasticity-robust estimators were applied (White estimator for random effects; Arellano estimator for fixed effects). Software: R 4.2.1 with packages plm, lmtest, sandwich, factoextra. Robustness: Models were re-estimated on the full pooled sample (no clustering), generally supporting consistency of signs and significance patterns relative to cluster-based estimates.

Classification: Two clusters by CE use were identified. Cluster 1 (higher CE): AUT, BEL, CZE, DNK, EST, FIN, FRA, DEU, ITA, LUX, NLD, POL, SVN, ESP. Cluster 2 (lower CE): BGR, HRV, CYP, GRC, HUN, IRL, LVA, LTU, MLT, PRT, ROU, SVK, SWE. RQ1 (Income → R&D potential): - Overall, positive relationships between Earnings and R&D indicators, stronger in higher-CE countries. - Cluster 1 (higher CE): Strongest links were Earnings→R_FTE_Business_enterprise (Within; R²=0.448; β=339.641, p<0.001) and Earnings→GERD_Higher_education (Random; R²=0.466). Earnings→GERD_Business_enterprise also significant (R²=0.309; β=54.273, p<0.001). Government R&D showed weak/non-significant responses (e.g., Earnings→R_FTE_Government β=−10.875, p=0.26; Earnings→GERD_Government β=2.52, p=0.055; α significant but slope weak). - Cluster 2 (lower CE): Fewer significant relations. Significant effects in Business enterprise sector: Earnings→R_FTE_Business_enterprise (Random White; R²=0.173; β=194.223, p<0.001) and Earnings→GERD_Business_enterprise (Random White; R²=0.195; β=22.668, p<0.001). Earnings→GERD_Higher_education also significant (R²=0.094; β=5.072, p<0.001). Other sectors mostly non-significant. RQ2 (R&D potential → HDI): - Broadly positive significant associations between R&D indicators and HDI. - Cluster 1 (higher CE): Very strong explanatory power for Private non-profit R&D FTE with HDI (Random; R²=0.831; β=0.00367, p<0.001). GERD_Private_non_profit had high R²=0.746 with HDI (β=0.02073, p=0.018). Business enterprise sector also significant: R_FTE_Business_enterprise→HDI (R²=0.221; β=0.00015, p<0.001) and GERD_Business_enterprise→HDI (R²=0.238; β=0.00072, p<0.001). Higher education R&D strongly linked to HDI (R_FTE β=0.00028, p<0.001; GERD β=0.00244, p<0.001). Government R&D FTE slope not significant with HDI (β=−0.00009, p=0.306), while GERD_Government had a small but significant slope (β=0.00097, p=0.039; R²=0.014). - Cluster 2 (lower CE): Strongest HDI links appear in Business enterprise sector: GERD_Business_enterprise→HDI (Within; R²=0.316; α significant) and R_FTE_Business_enterprise→HDI (R²=0.294; β=0.00021, p<0.001). Higher education R&D also significant (R_FTE β=0.00021, p<0.001; GERD β=0.856 intercept, R²=0.284). Government R&D FTE (Within) significant (country FE), and GERD_Government shows a positive association with HDI (R²≈0.029). Private non-profit sector links largely non-significant in Cluster 2. Sectoral efficiency insight: Government R&D appears relatively inefficient in explaining HDI or responding to Earnings, especially in higher-CE countries, compared to business and higher education sectors. Robustness: Pooled-sample random-effects models generally confirm signs and significance, with patterns closer to Cluster 1 in Earnings→R&D models; no major sign reversals.

The findings directly address RQ1 and RQ2. Higher income levels are associated with stronger R&D capacity, especially in business and higher education sectors, and this linkage is notably stronger in countries with higher CE adoption, suggesting that CE-aligned economies may amplify the transformation of income into research inputs. For RQ2, R&D capacity correlates positively with human development, particularly via business enterprise and higher education sectors; in higher-CE countries the private non-profit sector’s researcher intensity also shows a very strong association with HDI. Conversely, government R&D shows weaker and often non-significant effects on both responsiveness to income and contribution to HDI, pointing to potential inefficiencies or delayed payoffs. These results underscore the relevance of sectoral composition in national innovation systems and highlight that CE adoption may enhance the effectiveness of R&D in contributing to human development outcomes. The study’s implications support policies that strengthen R&D in business and higher education, foster CE adoption, and reassess the structure and efficiency of government R&D, considering time-lags and knowledge transfer mechanisms. The relationships suggest that reorienting growth metrics toward well-being and environmental outcomes, and fostering cross-sectoral innovation networks, can better align innovation policy with human development.

The study demonstrates that across EU-27 (2010–2020), higher income levels are positively associated with stronger R&D potential, and greater R&D inputs are linked to higher human development, with effects more pronounced in countries with higher CE adoption. Business enterprise and higher education sectors are the primary channels through which income translates into research inputs and research translates into development outcomes; the private non-profit sector also plays a notable role in higher-CE countries. Government R&D appears comparatively less effective, especially in high-CE contexts, indicating room for efficiency improvements and better knowledge transfer. Contributions include integrating CE and HD perspectives with sector-disaggregated R&D analysis and moving beyond GDP-centric views toward a well-being economy framework. Policy implications include prioritizing R&D as a lever for well-being, accelerating CE transition, adopting broader progress indicators, and fostering multi-stakeholder innovation networks. Future research should refine metrics for CE-related economic outputs, examine time lags and sectoral mechanisms, and build international datasets enabling benchmarking and forecasting of well-being-centric innovation policies.

Key data limitations include missing Eurostat values, especially for the Private non-profit sector (Researchers FTE n=92 missing; GERD n=86 missing), which may affect precision but are not expected to bias conclusions substantially. Cross-country reporting differences and data transfer to international databases introduce comparability challenges. Methodologically, converting diverse earnings categories to a composite metric and relying on patent- and CE-use indicators entail measurement choices. Future research should develop improved, multi-dimensional metrics for CE-related outputs at macro and micro levels, construct international databases to enhance coverage and comparability, and analyze sectoral/national differentiation and time-lag effects to quantify causal pathways and regional disparities more robustly.