
Digitalization is often expected to broaden economic opportunity, yet its distributional effects may change as access spreads. This paper asks whether household internet access is associated with income inequality in a stage-dependent way across Europe. Using a harmonized Eurostat country panel, we compare fixed-effects threshold models with dynamic panel and semiparametric checks, while accounting for the conventional relationship between income and inequality and for a broad set of social, labor-market, demographic, and macroeconomic factors. The findings reveal a clear asymmetry. During the early phase of digital diffusion, wider access can coincide with greater inequality, although this pattern is sensitive to how the early period is identified. Once internet access becomes widespread, further diffusion is consistently associated with lower inequality. Dynamic panel estimates provide only partial support for this mature-diffusion result, whereas the semiparametric analysis confirms the same broad shape. The evidence suggests that digital policy should evolve with the stage of diffusion: expanding basic access and affordability where connectivity remains uneven, then shifting toward skills, meaningful use, and inclusive digital services where access is already widespread.
This study examines the relationship between frequency-layer deployed LTE bandwidth intensity and regional economic performance in Indonesia's affirmative regions during 2020–2024. The indicators combine configured LTE carrier bandwidth with recorded LTE sector-cell multiplicity, aggregate deployment within Low-band (850/900 MHz), Mid-band (1800 MHz), and Upper-mid-band (2100/2300 MHz) layers, and normalise the resulting intensity by district population. Using a strongly balanced panel of 166 districts and 830 district–year observations, the study treats two-way fixed effects (FE) as the baseline estimator of conditional within-district associations and instrumental-variable (IV) estimation as acomplementary IV sensitivity analysis. The layer-specific coefficients are positive, but their ordering and precision differ across estimators: Mid-band is numerically largest in the FE specifications, whereas Low-band is numerically largest in the IV specifications. Under conventional district-clustered inference, the IV p-values are 0.1009, 0.1398, and 0.0934 for Low-band, Mid-band, and Upper-mid-band, respectively; however, the weak-instrument-robust Anderson–Rubin 95% confidence sets include zero for all three layers. The IV results therefore do not uniformly confirm the FE estimates. The reconstructed aggregate indicator is also positive but imprecisely estimated (β = 0.0525; clustered SE = 0.0303; p = 0.085), and its 95% confidence interval includes zero. The findings describe conditional associations with recorded LTE deployment intensity rather than causal or pure frequency effects. Frequency-layer and aggregate indicators consequently provide complementary, partial perspectives for telecommunications-infrastructure and spectrum-policy evaluation.
Data spaces provide infrastructure for enabling secure and trustworthy data sharing. Via different technologies and intermediary actors, data spaces allow data providers and consumers to interact. This paper explores the business models of those intermediary actors providing connections to the data space, data space connector providers. Implementing a qualitative, comparative case study methodology, four data spaces were analyzed to unravel the relationship between data space design and connector provider business models. Design choices such as the degree of standardization, centralization, and access and certification costs have major influence on how the business models of connector providers are shaped. This research contributes to the limited body of literature considering data space-related business models and aims to support the understanding of the economic potential of data spaces.
Australia's artificial intelligence (AI) policy is shifting from voluntary risk governance towards an infrastructure-centred industrial strategy. This article asks how the emerging framework reconfigures AI governance as industrial policy and whether it can advance the sovereign-capability and reindustrialisation ambitions associated with the Future Made in Australia (FMIA) agenda. It conducts an interpretive longitudinal analysis of 18 federal policy documents issued between 2019 and 15 July 2026, supplemented by electricity-market and national-accounts evidence. The analysis identifies four connected forms of reciprocity—material, cultural, capability and institutional—and conceptualises their integration as a legislated social-licence floor: enforceable minimum obligations that condition infrastructure access and regulatory facilitation without claiming that law can manufacture trust or acceptance. The policy constitutes an emerging AI–industrial strategy, but not yet sovereign AI or reindustrialisation. Its developmental effects remain contingent because AI services, compute and capital cross borders; model developers, cloud platforms and end users occupy distinct layers; and imported equipment, federal implementation, carbon leakage, additionality measurement and unsettled copyright rules can weaken domestic value retention. Five tests make the framework assessable: additionality, domestic value retention, capability deepening, diffusion, and resource and community performance. Together they distinguish investment volume from durable capability formation while incorporating consumer incidence and cross-border effects.