Page 176 - ISC PROCEEDINGS 21.4

levels and returns accrue disproportionately to high-skilled and capital-owning groups,
the marginal distributional effect reverses (β₂ > 0).
Three gaps motivate this study. First, systematic global estimation of a statistically
significant internet–inequality turning point with cross-sectional dependence-robust
inference remains absent. Second, the extent to which the turning point varies
quantitatively across World Bank income groups has not been estimated in a unified
interaction framework. Third, a simultaneous comparison of internet penetration, fixed
broadband, and mobile cellular subscriptions within a single methodological framework
— to test whether the DKC is technology-specific — has not been undertaken. This paper
addresses all three gaps using 128 countries over 2000–2022, a quadratic fixed effects
model with Driscoll–Kraay (1998) standard errors, income-group interaction terms, and
System GMM estimation (Roodman, 2009).
2. Literature review
The theoretical foundation combines three frameworks. Kuznets (1955) posited that
structural transformation initially intensifies inequality before alleviating it; Elfaki &
Ahmed (2024) validate an analogous Technological Kuznets Curve across Asian economies.
The DKC examined here is the inverse: internet access first reduces inequality by
democratising information and market participation (β₁ < 0), then amplifies it as skill-
biased returns and platform concentration dominate (β₂ > 0). The SBTC framework
(Acemoglu, 2002; Card & DiNardo, 2002) provides the micro-level mechanism for the
post-threshold reversal, while Piketty's (2014) r > g condition explains why platform-era
capital accumulation reinforces rather than alleviates the distributional shift at high
penetration levels.
Empirical findings are context-dependent. Njangang et al. (2022), analysing 45
nations via GMM, found a positive association between ICT and wealth inequality,
moderated by democratic institutions. Ho et al. (2025) report a negative impact of
digitalisation on income inequality across 45 developing countries, with governance
quality as the key moderator. The most directly comparable evidence comes from
Ariansyah et al. (2023), who document a nonlinear relationship between mobile
broadband and inequality across 122 Indonesian regions with a turning point at
approximately 60 percent network coverage — closely corresponding to the global
threshold estimated here. Hjort & Poulsen (2019) exploit submarine cable rollouts across
Africa to identify employment gains among less-educated workers, providing
microeconomic support for the inequality-reducing phase of the DKC. At the technology
level, Gruber & Koutroumpis (2011) demonstrate asymmetric distributional effects of
fixed versus mobile infrastructure due to cost barriers, while Bahia et al. (2020) show that
mobile broadband generates inclusive consumption gains through its broad cross-income
diffusion — motivating the present study's parallel estimation across ICT types. Lee &
Hwang (2026), examining 217 countries, find that ICT's inequality-reducing effect is more
pronounced where entrepreneurial conditions are suboptimal, confirming that
institutional context mediates the income-group heterogeneity tested in Section 4.2.
3. Data and methodology
3.1. Data
This study uses an unbalanced panel of 128 countries over 2000–2022, drawn
entirely from the World Bank's World Development Indicators (World Bank, 2024). The
final sample comprises 1,346 country-year observations. The dependent variable is the
Gini coefficient (SI.POV.GINI) from the World Bank's Poverty and Inequality Platform. The

175