Digitalization May Narrow Income Inequality Without Directly Raising Wages

Digitalization has transformed economies, expanded connectivity and reshaped employment, but its ability to deliver higher wages remains far less certain. Evidence from 38 OECD countries suggests that technological expansion is more closely associated with workers' earnings through broader economic growth than through any consistently identifiable direct wage effect. The findings challenge the assumption that increasing internet penetration automatically translates into better-paid employment.

Titled "Does Digitalization Pay? Evidence of an Indirect Wage Association Through GDP per Capita in OECD Countries," the research was conducted by Murat Ahmet Doğan of Samsun University, Türkiye and published in the MDPI journal Economies. Drawing on 950 country-year observations across 38 OECD economies between 2000 and 2024, the study examines whether internet adoption and broadband expansion are associated with higher wages, whether those relationships change as digitalization advances, and how labour-market institutions and income inequality enter the picture.

The findings complicate the economic case often attached to digital expansion. Direct associations between digitalization and wages are statistically unstable, while a stronger relationship emerges through GDP per capita. Internet penetration also displays a nonlinear association with earnings, and higher digital adoption is consistently associated with lower income inequality. Together, these results suggest that digital infrastructure and improved labour-market outcomes should be assessed separately rather than treated as interchangeable measures of economic progress.

Why digital expansion does not automatically translate into higher wages

Digital technologies can improve productivity, reorganise work and change demand for particular skills, but those changes do not necessarily produce uniform wage increases. The research builds on competing explanations of technological change: digital capital may complement skilled labour and increase its productivity, while automation can replace routine tasks and alter employment opportunities across occupations. National wage trends reflect these overlapping forces rather than a single technological effect.

The author measures digitalization principally through the share of people using the internet and fixed-broadband subscriptions per 100 inhabitants. The analysis incorporates GDP per capita, unemployment and inflation while accounting for differences between countries and changes over time. Wage-data coverage was expanded from 86% to 99.5% by supplementing international statistics with national sources from Germany, Japan and New Zealand, strengthening the geographical and historical coverage of the analysis.

Despite the breadth of the dataset, the direct relationship between digitalization and earnings remains weak. Under conventional country-clustered standard errors, neither internet use nor broadband penetration demonstrates a statistically significant association with average wages. Broadband becomes significant under Driscoll–Kraay standard errors, which account for statistical dependence across countries and time, while internet use reaches only marginal significance. The underlying estimates remain unchanged; their statistical interpretation changes with the estimation method.

A more consistent pattern appears when GDP per capita enters the analysis. A formal mediation test identifies a statistically significant indirect association between internet penetration and wages through GDP per capita, producing a test statistic of 2.82 and a probability value of 0.005. A country-clustered bootstrap confirms the result, with a 95% confidence interval ranging from 0.0022 to 0.0110.

However, the GDP relationship requires an important qualification. Because wages form part of national income, their close association with GDP can partly reflect the structure of national accounting rather than an independent economic transmission mechanism. Shared currency-conversion effects can also strengthen the apparent relationship. When purchasing-power-adjusted wages replace exchange-rate-converted nominal earnings, the GDP-per-capita coefficient declines from 1.137 to 0.491 but remains statistically significant.

The evidence does not establish that digitalization causes economic growth, which subsequently raises wages. Instead, it demonstrates that connectivity and earnings move together largely alongside national income, while a separate direct wage relationship cannot be reliably isolated.

The digital wage paradox: Connectivity, thresholds and inequality

The relationship between internet adoption and wages becomes more complicated when the analysis allows it to change at different levels of digital development. Rather than identifying a consistent upward trajectory, the study detects a U-shaped association, with a negative slope at lower levels of internet penetration and a positive slope at higher levels under its preferred statistical specification.

An initial calculation places the turning point near 76.6% internet penetration. Yet further testing reveals considerable uncertainty around that estimate, preventing the research from establishing a precise threshold at which the wage relationship changes direction. A separate comparison of countries above and below 90% internet penetration also fails to reproduce the pattern consistently across statistical methods.

Broadband connectivity provides another qualification. Among observations with internet penetration below 90%, fixed broadband displays a positive association with wages under Driscoll–Kraay errors, while the corresponding relationship is not statistically significant in the higher-penetration group. The result suggests that basic infrastructure expansion may carry different wage associations depending on how widely connectivity has already spread, although the study treats this evidence as exploratory.

Income inequality produces a clearer statistical pattern. Internet penetration is negatively associated with the Gini coefficient, a measure of disposable-income inequality, with an estimated coefficient of −0.000640 that remains statistically significant under both principal estimation methods. A separate measure comparing higher- and lower-income groups also shows a negative association with internet use under fixed-effects estimation.

The contrast between average wages and income distribution is particularly revealing. Digitalization's relationship with inequality is more consistently detectable than its relationship with aggregate earnings, indicating that changes in connectivity can be associated with wage levels and income distribution in different ways. The findings do not establish that digitalization reduces inequality causally, nor do they identify precisely which workers or households experience the observed distributional changes.

Labour institutions fail to explain the wage connection

Labour-market institutions have frequently featured in explanations of how technological gains are distributed among workers. Collective bargaining can influence wage negotiations, while union representation may shape workers' participation in technological change. Doğan examines whether these institutional arrangements systematically alter the relationship between digitalization and average earnings across OECD economies.

The evidence does not support that proposition. An earlier specification suggested that collective bargaining coverage might influence the digitalization–wage relationship, but the result disappeared after country and year effects and statistical corrections were applied consistently. Interactions involving union density were also statistically insignificant, leaving the proposed institutional-moderation relationship unsupported in the analysed dataset.

Data availability creates an additional complication. Observations lacking collective bargaining information systematically exhibit lower wages, lower internet penetration and lower GDP per capita than observations with complete information. Consequently, the institutional analysis relies on a smaller sample disproportionately representing wealthier and more digitalized economies, limiting the conclusions that can be drawn from it.

Additional tests involving tertiary education offer similarly qualified results. Higher tertiary enrollment is negatively associated with wages after controlling for GDP per capita and digitalization, but the interaction between enrollment and internet penetration is not statistically significant. The research therefore finds no evidence for the specific proposition that higher educational enrollment systematically strengthens digitalization's wage association, while leaving the negative enrollment relationship unexplained.

The results also demonstrate the importance of distinguishing theoretical expectations from statistically supported relationships. Collective bargaining, education and technological adoption may each influence labour markets through multiple mechanisms, but the study cannot establish the proposed interaction effects from its aggregate evidence. Its strongest conclusions concern the GDP-linked wage association and the negative relationship between internet penetration and income inequality.

Beyond internet access: What digital economic policy still needs to prove

The research carries a policy warning for governments investing in connectivity as part of economic development strategies. Increasing internet access may expand technological capacity, but the evidence does not justify treating additional subscriptions or higher penetration rates as reliable predictors of wage improvement. Broader economic performance remains closely associated with earnings, while the direct role of basic digital infrastructure is difficult to isolate.

For countries approaching widespread internet adoption, the study points toward the relevance of broadband quality, digital skills and productivity-enhancing measures rather than an exclusive emphasis on expanding access. Its findings also suggest that policies concerned with income distribution deserve attention alongside aggregate wage objectives, given the comparatively consistent negative association between internet penetration and inequality.

Important limits prevent these findings from becoming universal prescriptions. The analysis covers OECD economies and relies primarily on first-generation digitalization indicators, including internet access and broadband connectivity. It does not directly examine artificial intelligence readiness, robot density or other advanced technologies increasingly associated with changes in employment and production.

Methodological caution is equally necessary. Although the research employs alternative wage measures, robustness checks and dynamic-panel estimation, it cannot eliminate reverse causality or all unobserved economic influences. Digitalization coefficients are not statistically significant in the preferred dynamic-panel checks, and statistical confirmation of an indirect association does not establish a causal pathway.

Future investigation could expand comparable datasets to include AI adoption and automation intensity, examine why collective bargaining statistics are missing across particular countries and periods, and explore the unexpected negative association between tertiary enrollment and earnings. More direct evidence on the mechanisms linking technological change, productivity and wages would help clarify when digital investment translates into better-paid work.

The economic promise of digitalization cannot be judged by how many people gain internet access alone; it also demands evidence that technological expansion is accompanied by improvements in workers' economic position.