The researchers reconstructed the Kaya identity for carbon emissions into eight interacting variables: GRP, carbon emissions per unit of GRP, total energy consumption, carbon emissions per unit of energy, population, carbon emissions per capita, GRP per capita, and energy consumption per unit of GRP. They then applied GDIM, a calculus-based index-decomposition technique that allows absolute and intensity factors to be considered together. Unlike a direct decomposition of a fixed multiplicative chain, the method explicitly links the variables through constraints and calculates each factor’s contribution to the change in emissions. The contribution of all eight factors reproduced the observed annual change in carbon emissions for each city, with residuals at approximately 10^-12.

City-level fossil carbon dioxide emissions came from the EDGAR 2024 greenhouse-gas database, released in October 2024. EDGAR compiles emissions through 2023 using international statistics and a consistent methodology based on the IPCC 2006 Guidelines for National Greenhouse Gas Inventories. Researchers obtained Chengdu and Chongqing emissions by aggregating 0.1-degree by 0.1-degree gridded cells within each city’s administrative boundaries. The study did not independently calculate the city emissions from fuel use; the common carbon-emission factors reported for coal, oil and natural gas were provided for comparison with bottom-up studies. Population and GRP information came from official statistical yearbooks for Chengdu, Chongqing and Sichuan.

Chengdu’s energy consumption required an additional estimate because city-level energy statistics were unavailable. The researchers followed the approach described in the Sichuan Statistical Yearbook, allocating provincial energy consumption according to Chengdu’s share of provincial GRP. This preserves consistency between the estimated energy series and measured economic activity, but it assumes that Chengdu’s share of provincial energy use matches its share of provincial output. That assumption may be imperfect because the city has a distinctive service sector, industrial composition and energy mix. The authors acknowledge that the principal uncertainty affects the estimated level of Chengdu’s energy use and recommend replacing the allocation with official city energy balance tables and sectoral data when those become available.

Demographic change produced a notably different result in the two cities. Population growth contributed positively to emissions in both Chengdu and Chongqing, whose permanent-resident populations reached 21.40 million and 31.91 million, respectively, in 2023. Per-capita carbon emissions behaved differently: the factor had a positive average contribution in Chongqing but a negative contribution in Chengdu. This means the change in emissions per person added to Chongqing’s total but restrained the increase in Chengdu over the study period. The authors associate this contrast with differences in energy-consumption patterns and urbanization, and suggest that public transport use, green consumption, service-sector development and lifestyle changes may be relevant in Chengdu. Those proposed mechanisms were not directly tested and remain hypotheses rather than causal conclusions.

The energy decomposition followed a “one pull, two restraints” pattern in both cities. Increasing total energy consumption drove emissions upward, whereas falling carbon emissions per unit of energy and falling energy use per unit of GRP restrained the increase. Chongqing’s energy consumption was much larger, rising from 66.04 million tons of standard coal in 2014 to 84.26 million tons in 2023, and the authors describe its scale effect as stronger and more sustained. Chengdu’s total use was lower and more moderate, producing a smaller positive contribution. The two mitigating measures nevertheless had comparatively small effects, so the expansion of total energy use remained central to the regional emissions trajectory.

Economic scale emerged as the largest overall driver. Cumulative GRP contributions reached 29.23% for Chongqing and 29.14% for Chengdu by 2023, although the marginal contribution of economic growth gradually weakened over the period. At the same time, GRP per capita and carbon emissions per unit of GRP both restrained emissions in the two cities. The carbon-intensity factor had the strongest cumulative inhibitory effect among the decomposed economic variables, with reported cumulative contributions of −18.75% for Chongqing and −26.58% for Chengdu. These results indicate that output growth continued to carry an emissions cost even as efficiency improved. The authors interpret the weakening marginal contribution cautiously, describing it as an early sign of reduced dependence on energy consumption and environmental pressure rather than proof of complete decoupling.

One year stands out in the annual results. Chongqing’s contribution from carbon emissions per unit of GRP was negative from 2014 through 2022 but turned positive at 0.0051 in 2023, meaning emissions briefly grew faster than GRP. The authors suggest several possible explanations: a post-pandemic rebound in energy-intensive production, reduced hydropower availability during extreme drought and heat in the Sichuan–Chongqing region, greater reliance on coal-fired electricity, and diminishing gains from energy saving. Because the reversal appears in a single year, the study does not establish whether it represents a lasting structural shift. The broader pattern remained one in which rising output and energy use were opposed by improvements in emissions intensity and energy efficiency.

On this evidence, the metropolitan area’s emissions future depends on whether improvements in output carbon intensity can continue to offset expansion in population, energy demand and GRP. The study’s recommendations focus on reducing the marginal emissions impact of economic growth, accelerating industrial restructuring and low-carbon technology, and improving efficiency in industry, buildings and transport. It also proposes cleaner energy supplies, including hydropower, natural gas and renewables, although the analysis does not evaluate individual policy instruments. For the two cities, it suggests different priorities: Chongqing’s larger energy-intensive industrial base and continuing urbanization require attention to industrial retrofits, transport electrification and cleaner buildings, while Chengdu’s service-oriented economy and public-transit advantages provide a base for low-carbon consumption, rail-oriented mobility and digital emissions management.

Regional coordination is also central to the authors’ proposed response. The cities have different economic structures, energy profiles and urbanization stages, yet coordinated governance has remained incomplete. The study recommends compatible carbon-accounting standards, jointly established reduction targets and cross-jurisdictional cooperation so that Chongqing’s industrial transition and Chengdu’s lower per-capita emissions pattern can contribute to a shared metropolitan framework. These are policy interpretations of the decomposition rather than interventions tested by the research. The analysis is also limited by the estimated Chengdu energy series, the use of gridded EDGAR emissions and the absence of sector-level decomposition or econometric testing of the proposed mechanisms. The result is best understood as a quantitative map of the factors associated with emissions change in 2014–2023, not a complete causal account of the region’s low-carbon transition.

Subject of Research: Decomposition of carbon-emission drivers in the Chengdu–Chongqing metropolitan area

Article Title: Factor decomposition of carbon emissions in the chengdu-chongqing metropolitan area using the generalized divisia index decomposition method

Article References: Wan, Y., Nie, S., Song, X., & Wen, J. (2026). Factor decomposition of carbon emissions in the chengdu-chongqing metropolitan area using the generalized divisia index decomposition method. Frontiers in Environmental Science, 14, 1962372. https://doi.org/10.3389/fenvs.2026.1962372

Image Credits: AI Generated

DOI: 10.3389/fenvs.2026.1962372

Keywords: carbon emissions, Chengdu, Chongqing, GDIM, energy consumption, economic growth, carbon intensity, urbanization, decomposition analysis, emissions drivers, Factor, decomposition