The analytical foundation of the study is the Environmental Kuznets Curve, the influential hypothesis holding that pollution rises with income during early industrialization and falls once a critical income threshold is passed, as economies shift toward cleaner technologies, service-oriented production, and stronger environmental preferences. The mechanism is conventionally decomposed into a scale effect, which raises emissions as output expands; a composition effect, which alters the sectoral structure of the economy; and a technique effect, which lowers the emissions intensity of production as incomes and abatement capacity grow. The inverted U emerges when the latter two effects eventually dominate the first. What has been missing, the authors argue, is digitalization, which acts on precisely the margins through which the curve operates, altering production technologies, the composition of output, and the intensity of energy and material use.

Theory offers reasons to expect digitalization to help and reasons to expect it to hurt. On the beneficial side sit four channels: efficiency gains from digital process control, smart grids, and predictive maintenance; dematerialization, as digital substitutes displace physically embodied goods; structural change, as digital diffusion accelerates the shift toward services; and improved monitoring and enforcement, as granular sensing lowers the cost of observing emissions. Against these stand three opposing forces: the growing electricity appetite of data centers and networks, the embodied emissions in hardware manufacturing and disposal, and the rebound effect, whereby efficiency gains lower effective costs and stimulate compensating consumption. Because these channels operate simultaneously and in opposition, the net effect cannot be resolved on theoretical grounds, which is exactly what makes the question empirical.

To measure digitalization properly, the researchers constructed a composite index rather than relying on a single indicator such as internet penetration. The index combines three measures drawn from International Telecommunication Union data: the share of individuals using the internet, mobile cellular subscriptions per hundred inhabitants, and fixed broadband subscriptions per hundred inhabitants. Using principal component analysis on pooled data from 1995 to 2022, they extracted a single underlying factor that accounts for nearly 85 percent of the variance across the three indicators, then rescaled it onto a range from zero to one hundred. Validation checks were reassuring: the index correlates at 0.99 with an equal-weighted alternative, at 0.94 with the ITU’s own ICT Development Index over the years it existed, and its internal consistency measured by Cronbach’s alpha reaches 0.92.

The sample was chosen with care. Fifteen economies, Brazil, Chile, China, Colombia, Egypt, India, Indonesia, Malaysia, Mexico, Peru, the Philippines, South Africa, Thailand, Türkiye, and Vietnam, span per capita incomes from roughly 620 dollars at the start of the period to about 14,300 dollars at its end, an interval that comfortably brackets the estimated turning point and allows the inverted U to be identified from within the data rather than extrapolated beyond it. The group also spans the full spectrum of digital diffusion, from economies where the internet was effectively absent in 1995 to those now approaching universal connectivity, and it ranges across coal-dependent, hydrocarbon-exporting, and hydro-intensive energy systems on four continents.

The methodological apparatus is deliberately demanding. Tests confirmed strong cross-sectional dependence, meaning the fifteen economies respond to common global shocks rather than evolving independently, and rejected slope homogeneity, meaning no single set of elasticities fits them all. Both findings forced the use of second-generation panel procedures robust to these features. The long-run estimates, obtained under five different estimators resting on different identifying assumptions, converge on a consistent picture: a positive income coefficient and a negative quadratic term support the inverted U, with an estimated turning point near 8,690 dollars per capita in constant 2015 terms, falling within the observed income range of the sample. Energy consumption emerges as the single most robust determinant of emissions, with an elasticity between 0.68 and 1.00 that is significant at the one percent level under every estimator, establishing the energy system as the binding constraint through which every other mechanism must operate.

Digitalization enters negatively in all five estimators, indicating that the efficiency and structural-change channels dominate the footprint and rebound effects in this sample, though the effect is small and reaches conventional significance only at the ten percent level in the pooled specification. Crucially, when the authors tested whether digitalization relocates the turning point itself by including an income-digitalization interaction, the interaction proved insignificant. The evidence therefore supports a level effect rather than a pivot: digital diffusion lowers the emissions associated with any given level of income without altering the income at which the relationship turns. The small magnitude of the net effect is consistent with prior work showing that the digital economy raises emissions directly while moderating the emissions generated by growth, two opposing channels of comparable magnitude producing a modest resultant.

The study’s most distinctive contribution lies in its treatment of adjustment dynamics. Conventional models impose a single speed at which emissions return to their long-run equilibrium path, but the economics of carbon lock-in predicts otherwise: correcting an overshoot requires retiring sunk fossil-fuel capital, which is slow and costly, while correcting an undershoot requires only fuller utilization of installed capacity, which is fast. Allowing the correction speed to depend on the direction of movement in the disequilibrium, the momentum-conditioned specification dominates its symmetric counterpart. Deviations that are widening correct at 10.8 percent per period, implying a half-life of roughly six years, while those that are narrowing correct at 24.8 percent, a half-life of about two and a half years. The system resists movement back toward equilibrium roughly two and a half times more strongly when emissions are drifting away from their long-run path than when they are returning to it. The authors are candid that bootstrap inference places the asymmetry at the ten percent significance level, making the finding suggestive rather than decisive, but its direction matches precisely what carbon lock-in theory predicts.

The policy implications follow directly. Because digitalization lowers the curve, green digital investment emerges as a decarbonization instrument compatible with the developmental imperatives that emerging economies cannot set aside, rather than a competitor for the resources those imperatives command. Because energy use remains the dominant driver of emissions, digital strategies must be pursued jointly with the decarbonization of energy supply rather than as a substitute for it, and the direct energy footprint of data centers and networks warrants explicit regulatory attention, including efficiency standards and requirements that new facilities be matched with additional renewable generation. Most pointedly, because emissions overshoots correct slowly, policymakers cannot rely on market-led reversion precisely in the regime where environmental damage is greatest. Binding standards, carbon pricing, and the deliberate removal of fossil-fuel subsidies are strengthened by the evidence, as is the case for coordinated international action given that emissions across these economies respond to common global shocks that no single country can address alone.

The study also acknowledges its limits. The digitalization index is a proxy for a multidimensional phenomenon, the modest number of countries limits the power of the bootstrap threshold test, and the analysis is associational rather than causal. Yet the central message stands with unusual clarity: the environmental consequences of growth in emerging economies are shaped not by income alone, nor even by income and digitalization together, but by the direction-dependent manner in which emissions adjust toward their long-run path. Half the world’s emissions now hang on economies whose digital transformation is flattening their pollution curve while their entrenched carbon infrastructure fights to keep it from falling fast.

Subject of Research: The relationship between digital transformation, carbon emissions, and asymmetric environmental adjustment in emerging economies

Article Title: Digital transformation, carbon emissions, and asymmetric environmental equilibrium in emerging economies

Article References: Sendi, A., Alofaysan, H., Alqayidi, A., & Mohammed, K. S. (2026). Digital transformation, carbon emissions, and asymmetric environmental equilibrium in emerging economies. Environmental and Sustainability Indicators, 32, Article 101522. https://doi.org/10.1016/j.indic.2026.101522

Image Credits: AI Generated

DOI: 10.1016/j.indic.2026.101522

Keywords: digitalization, carbon emissions, Environmental Kuznets Curve, emerging economies, carbon lock-in, ICT, energy consumption, asymmetric adjustment, cointegration, climate policy, rebound effect, green digital investment