Research Information System
Journal of the Knowledge Economy, 2026 (Scopus)
The relationship between digitalization—particularly broadband technologies—and productivity remains an open empirical question. While many studies report productivity-enhancing effects of digitalization, others find limited or no impact, giving rise to the so-called “productivity paradox.” This paradox is especially pronounced in less developed economies, where firm-level evidence is considerably scarcer than in advanced countries. Addressing this gap, this study investigates the impact of broadband adoption on firm productivity using microdata from Turkish firms operating in the manufacturing, services, and ICT sectors between 2012 and 2015. To account for selection bias and endogeneity in adoption decisions, we employ an endogenous switching regression model. Our findings indicate that DSL connections are not associated with significant productivity gains, whereas the adoption of other fixed broadband technologies positively influences total factor productivity (TFP). We reveal a clear sectoral ranking in TFP gains, with ICT firms benefiting the most, followed by manufacturing and services. Mobile broadband adoption also contributes to productivity improvements, particularly among service-sector firms, where flexible and decentralized workflows may amplify its effects. These results highlight an important insight aligned with the productivity paradox: the diffusion of digital infrastructure does not automatically translate into productivity gains. Instead, such gains depend on firm- and sector-specific capabilities to absorb and utilize digital tools effectively. This study offers four main contributions: (1) it disaggregates the analysis by manufacturing, services, and ICT sectors to explore sector-specific digitalization dynamics; (2) it provides novel evidence on broadband-related productivity gains within ICT industries, where digital complementarities are most pronounced; (3) it differentiates broadband technologies to reflect variation in infrastructure type, speed, and usage characteristics; and (4) it applies an endogenous switching regression approach to address both selection into adoption and potential endogeneity between broadband use and firm performance. Together, these contributions yield new insights into the firm-level mechanisms through which broadband adoption can foster knowledge diffusion and productivity in less digitally mature economies.