This study focuses on five provinces—Guangdong, Guangxi, Yunnan, Guizhou, and Hainan—which together constitute the geographical coverage of the China Southern Power Grid (CSPG). It is important to clarify that the CSPG, as defined by the operational and administrative boundaries of China Southern Power Grid Company Limited, differs from the broader geographic or cultural notion of “southern China.” While the latter may include provinces such as Fujian, Jiangxi, or Hunan, these are served by the State Grid Corporation of China and therefore fall outside the scope of this study. The five selected provinces are interconnected through shared infrastructure, electricity dispatch mechanisms, and emission responsibilities under the CSPG framework. Analyzing them collectively allows for more accurate attribution of electricity-related carbon emissions and enables insights into intra-grid policy coordination.

Previous studies pointed out that when a total amount index is decomposed, the additive LMDI method is the preferred method (Ang, 2005; 2015). Therefore, this study utilizes this method to decompose carbon dioxide emissions from the electricity sector in five southern provinces (Guangdong, Guangxi, Guizhou, Hainan, and Yunnan), involving eleven influencing factors, which are classified into four categories by the electricity carbon emission features and five categories according to the electricity generation and consumption stages. The decomposing formula as Equation 1. C O 2 j = C O 2 j F C j · F C j G j · G j T G j · T G j E S j · E S j G D P j · G D P j P j · C O 2 j P j · P j E S j · E S j E C j · E C j C O 2 j · P j = U F j · P F j · F I j · P S j · E S I j · G D P p C j · C O 2 p C j · E S P j · E S D j · E C C j · P j (1)

Where, j is an integer ranging from 1 to 5, represents different regions, namely, Guangdong, Guangxi, Guizhou, Hainan, and Yunnan. The data in Equation 1 are sourced from the Southern Power Grid. Specifically, CO2 _j represents the carbon dioxide emissions from electricity generation in region j; FC_j represents the fossil fuels consumption in region j; G _j represents the total electricity generation from fossil fuels in region j; TG _j represents the total electricity generation in region j beside fossil fuels, such as solar, wind, and other types of renewable energy; ES _j represents the electricity supply in region j; EC _j represents the electricity consumption in region j; GDP _j represents the Gross Domestic Product (GDP) in region j; P _j represents the population in region j.

The eleven influencing factors of carbon dioxide emissions from electricity sector are as follows: UF _j is Fossil Fuel Carbon Emission Factor, carbon dioxide emission factor of fossil fuels in region j, calculated as the carbon emissions from electricity generation in region j divided by the consumption of fossil fuels in region j; PF _j is Fossil Fuel Power Generation Efficiency, efficiency of fossil fuel electricity generation in region j, calculated as the consumption of fossil fuels in region j divided by the electricity generation from fossil fuels in region j. FI _j is Fossil Fuel Power Generation Ratio, ratio of fossil fuel electricity generation in region j, calculated as the total electricity generation from fossil fuels in region j divided by the total electricity generation in region j, including hydroelectric, wind, and other renewable energy sources; PS _j is Electricity Self–Sufficiency Rate, self–sufficiency rate of electricity in region j, calculated as the total electricity generation in region j divided by the electricity supply in region j; ESI _j is Electricity Supply Intensity, intensity of electricity supply in region j, calculated as the electricity supply in region j divided by the economic output in region j; GDPpC_j is GDP per capita in region j; CO₂pC _j is Carbon Dioxide Emission per capita in region j; ESP _j is Available Population Size of Unit Electricity Supply in region j, calculated as the population in region j divided by the electricity supply in region j; ESD _j is Electricity Supply–Demand Ratio in region j, calculated as the electricity supply in region j divided by the electricity consumption in region j; ECC _j is Electricity Consumption Carbon Emission Factor, carbon emission factor of electricity consumption in region j, calculated as the electricity consumption in region j divided by the carbon emissions from electricity generation in region j.

The change of carbon dioxide emissions from the electricity sector between the base year 0 and the year t can be divided into eleven parts, i.e., the effects of eleven factors as Equation 2: △ C O 2 j = C O 2 j t ‐ C O 2 j 0 = ∑ i △ U F i j + △ P F i j + △ F I j + △ P S j + △ E S I j + △ G D P p C j + △ C O 2 p C j + △ E S P j + △ E S D j + △ E C C j + △ P j (2)

Four Categories of power carbon emission features include emission intensity effect, energy structure effect, electricity supply effect, and macroeconomic effect. Five categories by the stages are power generation sector effect, Power CTD (Conversion, Transmission, Distribution) sector effects, emission scale effect, economic output effect, and population scale effect.

This dual-framework design serves two purposes. First, the characteristic-based categorization enables a policy-relevant understanding of emission drivers, reflecting sectoral dynamics such as technological efficiency and structural transition. Second, the stage-based decomposition aligns with the physical and institutional stages of electricity flow—from generation to end-use—allowing clearer attribution of emissions to each operational segment of the electricity system. Such categorization approaches have been applied in previous empirical studies (Zhang et al., 2013; Yan et al., 2016; Zhang et al., 2022; Wang and Yang, 2023; Wei et al., 2023) to enhance interpretability and to facilitate more actionable policy recommendations.

Table 1 lists the variables used in the LMDI decomposition analysis, including their symbols and definitions, cover technical, structural, and socioeconomic factors affecting carbon emissions in the power sector. These factors are categorized according to electricity carbon emission features as follows:

• Emission intensity effects: UF and ECC.

They are also categorized according to electricity generation stages as follows:

• Power generation sector effects: UF, PF, FI.

Equations 3–13 are logarithmic mean Divisia index (LMDI)-based decomposition formulas that quantify the contributions of various driving factors to changes in carbon emissions from the power sector. Where, ΔCO_2j refers to the total change of carbon dioxide emissions from the electricity sector during the period from year 0 to year t; ΔUF _ij is the contributions of Fossil Fuel Carbon Emission Factor; ΔPF_j is the contributions of Fossil Fuel Power Generation Efficiency; ΔFI_j is the contributions of Fossil Fuel Power Generation Ratio; ΔPS_j is the contributions of Electricity Self–Sufficiency Rate; Δ ESI_j is the contributions of Electricity Supply Intensity; GDPpC_j GDP per capita in region j; CO₂pC_j is Carbon Dioxide Emission per capita in region j; ESP_j is the contributions of Available Population Size of Unit Electricity Supply, can be represented as Unit Electricity Supply in terms of contribution; ESD_j is the contributions of Electricity Supply–Demand Ratio; ECC_j is the contributions of Electricity Consumption Carbon Emission Factor.

The data used in this study were obtained from internal sources provided directly by the China Southern Power Grid Company Limited, covering provincial-level electricity generation, transmission, distribution, and consumption data for the five provinces (Guangdong, Guangxi, Yunnan, Guizhou, and Hainan) from 2013 to 2021. Due to confidentiality agreements, the raw data are not publicly accessible, but the aggregated statistics are consistent with those reported by provincial energy yearbooks and grid bulletins.

To ensure consistency across datasets, all variables were standardized into uniform measurement units—for example, electricity quantities were converted into terawatt-hours (TWh), and carbon emissions into metric tons of CO₂. The conversion of electricity-related activity data into CO₂ emissions followed the IPCC Guidelines (2006) and China’s regional emission factor specifications, with adjustment for fuel mix and combustion efficiency. Economic indicators such as GDP and population were derived from official provincial statistical yearbooks and the National Bureau of Statistics of China.

Changes in the ratio of fossil fuel power generation contributed to the decrease in power sector carbon emissions (Figure 1), likely due to a decline in coal-fired power generation that aligns with policy Figure 2 directives aimed at carbon emission reduction. The provinces of Hainan and Yunnan, with relatively smaller economic scales, demonstrated higher contributions to emission reduction, while provinces with larger economic scales, such as Guangdong and Guangxi, showed lower contributions. This suggests that changes in the ratio of fossil fuel power generation have a relatively greater impact on power sector carbon emissions in economically developing regions.

During the period from 2013 to 2021, changes in fossil fuel carbon emission factors and fossil fuel power generation efficiency had relatively minor impacts on changes in power sector carbon emissions.

Changes in the power self–sufficiency rate and power supply–demand ratio had a somewhat negligible impact on power sector carbon emissions. Conversely, changes in power supply intensity and per capita power supply had a relatively significant impact on emission reduction. While all four driving factors are directly related to power demand and supply, factors involving changes in power demand, such as power self–sufficiency rate and power supply-demand ratio, had relatively minor contributions to emission reduction, whereas factors centered on changes in power supply had relatively significant impacts on power sector carbon emission changes.

Among the driving factors in Guizhou Province, the contribution of power supply intensity to the decrease in power sector carbon emissions was most prominent. This indicates that, among the southern provinces, Guizhou has optimized power supply volume and economic scale for carbon emission reduction between 2013 and 2021.

Changes in per capita electricity supply had a significant impact on emission reduction across all southern provinces, particularly in Guangdong, Guangxi, and Hainan. Notably, while the difference in the contribution of driving factors to emission reduction was relatively small in Guangdong, changes in per capita power supply had a substantial impact on emission reduction in Guangxi. This suggests that, during the period from 2013 to 2021, the per capita power supply in Guangxi was optimized for emission reduction, while adjustments to other driving factors could maximize emission reduction in the region.

Power consumption and carbon emission factors contributed to increased power sector carbon emissions in all southern provinces, with relatively larger contributions from Hainan and Yunnan due to differences in economic scale, development environment, and power demand.

Changes in per capita carbon emissions had a significant impact on emission reduction in the southern provinces, with relatively larger contributions from Hainan and Yunnan due to their smaller economic scales. This indicates that, as regional economic development increases, per capita carbon emissions tend to rise proportionally with population size, resulting in an increased impact on carbon emission changes, highlighting the direct relationship between carbon emission changes and economic development.

Changes in per capita GDP and population had an impact on increased power sector carbon emissions, with per capita GDP having a substantial impact on emission increase. Conversely, population changes had relatively low impacts on increased power sector carbon emissions across all southern provinces. Particularly noteworthy is the relatively larger impact of per capita GDP compared to other driving factors, indicating a significant influence of regional economic development on increased power sector carbon emissions. Despite being a macro–social indicator like per capita GDP, population size had relatively minor impacts, suggesting that population changes in the southern provinces during the period from 2013 to 2021 had somewhat negligible effects on carbon emission changes.

In Guangdong Province, the absolute values of each effect show a relatively consistent pattern, except for the macroeconomic effect which has no change from 2013 to 2021 (Figure 4). Particularly noteworthy is the steady increase in the contribution of the macroeconomic effect to carbon emission reduction. This suggests that Guangdong’s economic development is contributing to carbon emission reduction efforts.

For Guangxi and Guizhou, the contribution of each effect was minimal during the period from 2013 to 2019 (Figure 4). However, from 2013 to 2021, there was a significant increase in both the electricity supply and macroeconomic effects. This indicates that despite a rapid increase in the contribution of carbon emissions due to increased energy demand in these provinces, the contribution of the electricity supply effect offset the contribution of the macroeconomic effect.

In Hainan Province, apart from the sudden increase in all effects, other effects exhibit irregular trends except for the macroeconomic effect, where the contribution of the macroeconomic effect increased in 2019–2021 (Figure 4).

Yunnan Province experienced a steady increase in the contribution of each effect but maintained low cumulative contribution rates relatively (Figure 4). Particularly, the emission scale, energy structure, and macroeconomic effects increased significantly, suggesting a balance in emission scale, energy structure, population scale, and economic output concerning power sector carbon emissions. However, despite the small increase, the cumulative contribution rate of the electricity supply effect rarely has changed, and no effect transitioned to other states. These findings underscore the need for Yunnan Province to adequately consider electricity supply planning.

In Guangdong Province, the absolute values of each effect show a relatively consistent pattern. However, from 2013 to 2021, there was an overall decrease, particularly for the power CTD effect (Figure 5). This suggests that despite the stimulated Guangdong’s economic development, should progress with systematical emission reduction, particularly through the power CTD effect.

For Guangxi and Guizhou, similar to the categorized analysis, from 2013 to 2021, the contribution of each effect increased steadily (Figure 5). From 2019 to 2021, noteworthy that is significant in the high absolute values of power CTD effect. This analysis indicates that while Guangxi and Guizhou relied solely on the power CTD effect to offset carbon emissions from the other effects, where policies should balance the contributions of economic output and emission scale effects to carbon emission reduction. This underscores the need for customized sector-specific carbon emission reduction policies in both Guangxi and Guizhou.

In Hainan Province, apart from the sudden increase in the power generation effect, power CTD effect and economic output effect, other effects have negligible change (Figure 5). This is consistent with the results from the categorized analysis.

For Yunnan Province, shortly after 2015, the absolute values of each effect experienced a sharp increase, whereas the economic output effect and population effect maintained low cumulative contribution rates (Figure 5). This suggests that as of 2021, the effects of each stage in Yunnan Province are in balance in terms of carbon emissions or reduction, except for the economic output effect and population effect. Particularly, considering that the power generation effect and the emission scale effect are in the emission state, current carbon emission reduction policies in Yunnan province should primarily focus on the generation sector.

Figure 6 visualizes the relative contribution rates of power sector carbon emission driving factors for each province by category (Figure 6). Looking at the relative contribution rate of the emission intensity effect, we observe a gradual decrease in the area over time, indicating a decrease in the contribution of the emission intensity effect to power sector carbon emissions in all southern provinces. The shape of the area from 2015 to 2017 appears different, suggesting a transitional period for the emission intensity effect during that time.

The relative contribution rate of the energy structure effect shows varying shapes of the areas for each period. This indirectly implies that the energy structures of each province changed over time. The highest relative contribution rates were observed in Yunnan, Guangxi, and Guizhou from 2013 to 2015, and in Guangdong from 2017 to 2019. These periods likely represent transitional phases driven by relevant policies in each province. Although the absolute size of the areas remained relatively stable from 2013 to 2019, there was a significant decrease in area size from 2019 to 2021, indicating a reduction in the absolute contribution of the energy structure effect to power sector carbon emissions.

The relative contribution rate of the power supply effect maintains relatively consistent area sizes regardless of the period, suggesting that the differences in the relative contribution rates of power supply effects among provinces are not significant. However, the relative contribution rates of Guangxi and Guizhou show a stark contrast in terms of electricity consumption but remain almost unchanged compared to previous years. This anomaly may be attributed to the reduced electricity supply during the COVID–19 period.

The relative contribution rate of the macroeconomic effect maintains a relatively stable shape over time. However, the area size steadily increased from approximately 25% in 2013–2015 to around 50% in 2019–2021, showing higher contribution rates compared to other effects. Notably, Guangdong’s relative contribution rate in 2017–2019 was in the 0% range, while it surpassed 50% in 2019–2021. This suggests that the increase in carbon emissions from 2019 to 2021 was significantly higher than in the previous period, leading to this result.

Figure 7, visualizes the relative contribution rates of power sector carbon emission driving factors for each province by stage. The shape of the power generation sector effect was not consistent, but the results from 2019 to 2021 were concentrated towards the center, indicating a decrease in the overall relative contribution rate. This trend resembles the change in shape observed for the energy structure effect, suggesting that these results are also likely due to different policies related to the power generation sector in each province.

The shape of the transmission and distribution sector effect was also inconsistent, and the results from 2019 to 2021 were widely dispersed, indicating an overall increase in the relative contribution rate. Among these, Guangxi showed relatively consistent relative contribution rates over the periods, suggesting that the carbon emissions from the transmission and distribution sector in Guangxi are relatively stable.

The shape of the emission scale effect remained relatively stable except for Hainan. However, from 2019 to 2021, the emission scale effect in Guangdong increased dramatically from around 25%–15%, reflecting the significant increase in carbon emissions in Guangdong.

The economic output effect exhibited extreme shapes from 2013 to 2015, with Guangxi and Yunnan showing relative contribution rates in the 0% range, indicating a relatively low contribution of economic output to carbon emissions. Apart from these extremes, the economic output effects were distributed more evenly, indicating that the contribution of economic output to power sector carbon emissions in southern provinces has remained relatively stable from 2013 to 2021. The population size effect exhibited a shape similar to the economic output effect, with a more even distribution. However, Guangdong showed a clear decreasing trend in relative contribution rates, indicating a relatively stable overall contribution to power sector carbon emissions. The relative contribution rate of the population size effect from 2019 to 2021 was approximately 5%, indicating a very low contribution of the population size effect to the macroeconomic effect of southern provinces as a whole.

The total electricity carbon emissions of the five provinces in southern China have been steadily increasing from 2013 to 2021. Regarding this increase in electricity carbon emissions, previous research interpreted that the contribution of macro-social indicators was the highest. (Liu, 2022). This study corroborates this perspective, highlighting the paramount contribution of macro-social indicators while noting minimal differentiations attributable to temporal and regional factors. Though nuances in the contributions of other effects such as energy intensity, energy structure, power generation, and power CTD were observed across regions, their significance remained modest. Thus, it appears that these results stem from the increased energy demand due to population growth and economic development in each province, even as the Chinese government maintains a policy emphasis on electricity supply security. (Yao and Chang, 2014). It should be noted, however, that frequent extreme climate events (e.g., the 2021 Yunnan drought) may episodically alter carbon emission intensity. For instance, short-term emissions increase from coal-fired power compensation during drought years could partially offset structural decarbonization efforts, further underscoring the complexity of ensuring energy supply security amid macro socioeconomic drivers.

Analyzing individual effect-related factors in detail, it was found that Guangdong, Guangxi, and Hainan have wielded considerable contributions to carbon emission reduction through factors such as per capita electricity supply, electricity supply intensity, and the ratio of fossil fuel power generation. This underscore differing approaches to addressing electricity demand contingencies across different provinces. For instance, Guangdong’s reliance on external power procurement manifests in a proportional relationship between per capita electricity supply and carbon emission reduction. (Cheng et al., 2016). Conversely, Guangxi, Guizhou, and Hainan generate enough power domestically, resulting in relatively lower carbon emissions burdens from electricity supply, thus allowing per capita electricity supply to contribute to carbon emission reduction. Yunnan, with its high proportion of renewable energy generation and forest carbon sequestration efforts, also showed the significant contribution of carbon emission reduction from the factor of fossil fuel power generation (Chen et al., 2009). However, its emission reduction effectiveness exhibits significant climate sensitivity. For instance, the 2021 drought caused a 30% decline in hydropower output, necessitating an 18% temporary increase in coal-fired power generation (resulting in approximately 20 million tons of additional carbon emissions). This highlights the climate vulnerability of provinces predominantly reliant on renewable energy.

Provinces with larger economies like Guangdong, Guangxi, and Guizhou are making greater efforts to reduce carbon emissions through various strategies. For instance, Guangdong implemented measures related to carbon emission management, trading, establishment of carbon funds, and integration of carbon emission trading before the nationwide carbon trading system was introduced in 2016. (Comission, 2015). In 2017, through the 32nd session of the Standing Committee of the Twelfth Guangdong Provincial People’s Congress, Guangdong Province promulgated the Regulations on Electricity Supply and Utilization. Emphasizing principles of safety, reliability, efficiency, orderliness, livelihood assurance, energy conservation, and emission reduction, it officially encourages and supports the development of smart grids, the construction of distributed power sources such as photovoltaic power generation, and the promotion and application of new technologies, materials, and devices for energy conservation and environmental protection, while highlighting the avoidance or reduction of harm to the nation, collective, and individuals (Province, 2019). Despite these concerted efforts, however, Guangdong’s electricity carbon emissions escalated, and cumulative contributions to carbon emission reduction from various factors remained relatively stable, except for the power CTD effect, which decreased due to the transfer of carbon emissions through power purchases from other provinces.

Guangxi, cumulative contributions to carbon emissions reduction due to macroeconomic effects and scale effects, as well as cumulative contributions to carbon emissions reduction due to the power CTD effect, are increasing. However, the contributions from other effects show minimal changes. While Guangxi Province’s economic development continues steadily, and its future energy system heavily relies on clean energy sources (Li et al., 2023; Zhou et al., 2023). Therefore, necessitates exploring carbon emission reduction measures and policies related to the intensity of development and power generation effects. Thorough management and supervision are essential to ensure the continuous sustainability of carbon emission reduction contributions from the power CTD effect.

Similar to Guangxi, Guizhou faces critical challenges in simultaneously ensuring energy security and advancing energy structure transition. In implementation, particular attention must be paid to the impacts of extreme climate events on achieving these dual objectives. The freezing disaster that struck the Yunnan-Guizhou Plateau in December 2023 serves as a telling example: it not only caused significant daily reductions in regional wind power output but also disrupted several transmission lines. The emergency diesel generators deployed to address power shortages substantially increased electricity carbon intensity in the short term. This case underscores that while maintaining power system stability, it is imperative to concurrently implement carbon reduction measures and establish continuous monitoring mechanisms for power supply effectiveness - all of which are crucial for Guizhou to meet its carbon control targets.

Guizhou, akin to Guangxi, needs to coordinate energy security and energy transition efforts. (Yan et al., 2022). This requires seeking carbon emission reduction measures while ensuring a stable electricity supply, emphasizing the importance of continuously monitoring the carbon emission reduction contribution of power supply effects.

Hainan, designated as a special economic zone, has experienced rapid economic growth, and sudden increases in the cumulative contribution of all effects since 2017 indicate the need for proactive policies to minimize carbon emissions during economic development (Meng et al., 2018). Specifically, policies focusing on minimizing carbon sequestration capacity reduction in urban planning and construction due to increased tourism and real estate development could be considered. (Meng et al., 2018; Duan and Su, 2021).

Yunnan showed a different trend in relative contribution, with high contributions from emission intensity, energy structure, and power generation effects, but low contributions from macroeconomic, economic output, and population size effects. This can be attributed to Yunnan’s relatively small economy and unique power generation and trading structure (Cheng et al., 2018; Liu et al., 2023). The decreasing trend in the cumulative contribution of the power CTD effect suggests a need for strict supervision and incentives to maintain Yunnan’s high proportion of renewable energy usage, to reserve high potential of carbon sequestration.

Since the five southern provinces of China share a single power grid, the specific electricity-related situations and policies of a particular province have overall effects on southern China. Each region has different economic scales, electricity supply structures, and climate risk exposure, consequently leading to varying contributions of electricity carbon emission influencing factors. Therefore, the establishment of cogent policies and frameworks necessitates dynamic coordination considering the differences between provinces, to engender decarbonizing electricity supply and energy structure.

To effectively reduce electricity carbon emissions, China Southern Power Grid released its Carbon Peak and Carbon Neutrality Plan in 2021, driving the expansion of non-fossil energy installed capacity share and power generation volume (China Southern Power Grid Co, 2021). Additionally, Guangdong Province issued the 2024–2025 Energy Conservation and Carbon Reduction Action Plan in 2024, advancing large-scale offshore wind power development and high-quality distributed photovoltaic deployment (Government, 2024). These policies will directly reduce electricity carbon intensity and alleviate fossil fuel dependence, accelerating the carbon peaking trajectory. However, the demand surge driven by economic growth may partially offset structural emission reductions, highlighting the dual challenge for industrial powerhouses like Guangdong, Guangxi, and Guizhou in balancing development and low-carbon transition.

The decomposition attributes approximately 82% of the cumulative growth in electricity-related carbon emissions to macro-social drivers—namely population expansion and economic growth. This proportion is congruent with the region’s compound annual GDP growth rate of 6.4% and the 68% rise in residential electricity sales recorded in the China Energy Statistical Yearbooks (2014–2022). Notably, the marginal contribution of macro-social factors declines after 2017, coinciding with the commissioning of ultra-high-voltage transmission lines that enabled large-scale imports of low-carbon electricity, and with the accelerated retirement of sub-critical coal units in Guangdong following the enlargement of its pilot emissions-trading scheme in 2017. The temporal alignment between model output and these exogenous interventions lends credibility to the decomposition.

Guangdong demonstrates an 11% emission-reduction contribution from the cross-territorial dispatch (CTD) effect. Net electricity imports increased from 14 TWh in 2013 to 52 TWh in 2021, a volume that equates to roughly 32 Mt CO₂ when benchmarked against Guangdong’s average coal intensity (China Southern Power Grid Co, 2021). Yunnan records a 19% mitigation contribution from the energy-structure effect, reflecting the rise of hydropower’s generation share. The 2021 drought, which precipitated a 30% short-term reduction in hydro output and an 18% compensatory increase in coal generation, illustrates the climate sensitivity of hydro-dominant systems and explains the volatility observed in the annual series (Yunnan Provincial Energy Bureau, 2022). Guangxi and Guizhou show modest efficiency-related abatements of 2%–3%, in line with National Energy Administration audits reporting heat-rate improvements of 2.9% and 3.6% respectively over the study period (National Energy Administration South China Regulatory Bureau, 2023; Guizhou Provincial Bureau of Statistics, 2015). Hainan exhibits a pronounced rise in macro-economic contribution after 2017 that is traceable to the Free-Trade Port initiative, during which real GDP grew by 39% between 2017 and 2021.

Provincial average coal-fired emission factors may over-state Guangdong’s intensity—owing to its fleet of ultra-super-critical units—and under-state Guizhou’s, where smaller captive plants remain prevalent. The resulting bias is estimated to be within ±5 Mt CO₂ per annum and does not alter the qualitative hierarchy of effects. Future research would benefit from unit-level dispatch data.

Annual aggregation obscures the short-term impact of hydrological shocks such as the 2021 Yunnan drought. Integrating runoff indices to construct a counter-factual hydro baseline would allow the separation of climate-induced volatility from structural change, thereby refining the energy-structure effect in hydro-rich provinces.

The omission of approximately 7 GW of rooftop photovoltaic capacity in Guangdong leads to a slight over-statement of consumption-based emissions. Feeder-level metering suggests that correcting for this hidden generation would reduce Guangdong’s 2021 electricity-related emissions by about 3%, marginally increasing the contribution of the energy-structure effect.

First, the asymmetric CTD effect corroborates findings by Zhou et al. (2024) that uncoordinated provincial targets can displace rather than abolish emissions. A grid-level balancing mechanism analogous to the carbon-adjusted dispatch used in the PJM interconnection could internalize such spill-overs. Second, the temporary reversal of emission reductions during extreme events—exemplified by the 2021 Yunnan drought and the 2023 Guizhou ice storm—indicates that climate resilience must be treated as a precondition for sustained decarbonization. Embedding resilience criteria in provincial power-sector action plans would mitigate the risk of rebound emissions. Third, the dominance of macro-social drivers implies substantial mitigation potential on the demand side.

The LMDI results are broadly consistent with documented economic growth patterns, generation-mix shifts, and policy milestones in the Southern Power Grid. Nevertheless, the numerical findings must be interpreted alongside climate-related supply volatility, cross-border leakage through electricity trade, and untapped demand-side potential. By aligning the decomposition output with empirical developments, this thesis advances from a purely quantitative attribution toward a decision-relevant diagnosis of provincial pathways to a low-carbon electricity future.

First, by treating the five southern provinces as components of a single operational grid, the study shows that electricity-related emissions are shaped as much by network interactions as by provincial policies. The prominence of the cross-territorial dispatch term reveals emission transfers that conventional province-by-province inventories tend to mask. Recognising these transfers can help planners understand why identical macro-social pressures yield contrasting trajectories across the region.

Second, the year-to-year movement of factor contributions during droughts and ice storms illustrates how climatic volatility can erode structural gains. This empirical link between extreme weather and the stability of decomposition effects enriches an emerging conversation about power-system resilience, yet it does so through the lens of routine statistical reporting rather than bespoke climate models.

Third, the two-stage decomposition framework produces replicable, low-cost indicators—for example, the elasticity of emissions with respect to imported electricity or the share of growth attributable to macro-social drivers. These metrics require only standard Yearbook data and therefore travel easily to other multi-province grids, offering practitioners a quick way to rank mitigation levers before committing to more sophisticated modelling exercises.

Finally, by anchoring numerical shifts in the chronology of infrastructure upgrades, market reforms, and educational attainment, the analysis links technical attribution to institutional narratives. This fusion ensures the results read as reflections of ongoing socio-economic transformation rather than as abstract statistical artefacts.

While this study offers valuable insights into regional carbon emission dynamics within the Southern Power Grid, several limitations warrant attention. First, the analysis relies primarily on aggregated statistical data at the provincial level, which may obscure intra-provincial disparities and prevent identification of sector-specific emission sources. Second, the study does not explicitly quantify the impact of extreme weather events—such as droughts or heatwaves—which may disrupt renewable energy production and undermine the stability of decarbonization in hydropower- or wind-dependent provinces. Third, due to the time lag in data availability, the study period ends in 2021, thus excluding the effects of post-2021 climate policies following China’s “Dual Carbon” pledge. These recent developments may have significantly altered regional electricity structures and emission patterns.

To address these limitations, future research could integrate higher-resolution datasets, including hourly electricity dispatch data, facility-level emissions, and spatially disaggregated demand statistics. Quantitative modeling of climate-policy co-evolution—particularly the interaction between extreme weather events and dynamic policy responses—could further improve forecasting accuracy. In addition, expanding the scope beyond electricity-related emissions to include industrial, transportation, and building sectors would allow for a more comprehensive assessment of decarbonization trajectories. Despite these limitations, this study provides a replicable analytical framework and region-specific findings that can support more informed, localized policymaking in the context of complex multi-provincial electricity systems.

To further enhance the robustness and applicability of the findings, it is important to evaluate the reasonableness of the results against real-world developments. For instance, the dominance of economic growth and electricity supply in driving emissions aligns with the rapid industrialization and growing electricity demand observed in Guangdong and Guangxi during the study period. Meanwhile, the role of energy structure improvements in mitigating emissions corresponds to Yunnan’s significant hydropower investments.

Moreover, the analysis is constrained by annual temporal resolution, preventing the disentanglement of structural trends from short-lived shocks. End-use granularity is limited; linking factor decomposition to sectoral load curves could identify the specific demand segments that dominate the macro-social effect. Integrating abatement cost curves would further allow the ranking of provincial mitigation priorities under fiscal constraints. These findings support the validity of the decomposition results within the context of regional development trajectories.