Analysis of the Influencing Factors of Energy Carbon Emission in Shaanxi Province Based on DEMATEL Method

Jinxing Li, Jing Wang, Meizhi Wang2019 · DOI

Abstract

As a critical energy production and consumption hub in Western China, Shaanxi Province faces significant challenges in meeting national carbon reduction targets. This study applies the Decision Making Trial and Evaluation Laboratory (DEMATEL) method, enhanced by Ordinary Least Squares (OLS) regression, to analyze the complex web of factors influencing provincial carbon emissions between 2005 and 2016. By evaluating 17 indicators across four dimensions—Resource Endowment, Technology Level, Consumption Patterns, and Development Stage—the research identifies "Resource Endowment" as the most influential dimension and "Population Density" as the single most critical causal driver. These findings provide a data-driven foundation for regional low-carbon policy formulation.

Introduction

China's commitment to peaking carbon emissions by 2030 necessitates localized strategies, particularly for resource-rich regions like Shaanxi Province. Under the "13th Five-Year Plan," Shaanxi was tasked with an 18% reduction in carbon intensity. However, the drivers of emissions are multifaceted, involving demographic shifts, industrial structures, and technological advancements. Traditional decomposition methods often fail to account for the interdependencies between these factors. This paper addresses this gap by mapping the causal relationships within Shaanxi's carbon emission system to distinguish between root drivers and systemic outcomes.

Methodology: The OLS-DEMATEL Approach

The study utilizes the DEMATEL method, a structural modeling technique designed to visualize the relationships between complex variables. A common critique of standard DEMATEL is its reliance on subjective expert scoring to populate the initial influence matrix. To improve objectivity, this research employs Ordinary Least Squares (OLS) regression.

Index System Construction

The authors established a comprehensive system of 17 variables categorized into four primary dimensions:

  1. Resource Endowment (X1): Includes indicators such as the proportion of non-fossil energy (x1) and population density (x2).
  2. Technology Level (X2): Focuses on energy output efficiency (x7) and waste utilization rates (x8, x9).
  3. Consumption Patterns (X3): Covers industrial energy intensity (x12) and household consumption levels (x11).
  4. Development Stage (X4): Includes GDP per capita (x15) and urbanization rates (x17).

Mathematical Framework

Using historical data from 2005 to 2016 (sourced from the Shaanxi Statistical Yearbook), the authors calculated regression coefficients ($\beta_{ij}$) to represent the direct influence of one factor on another. Through matrix normalization and the calculation of the total relation matrix, four key metrics were derived:

  • Influence Degree ($r_i$): The sum of a factor's impact on all other variables.
  • Affected Degree ($s_j$): The extent to which a factor is influenced by the rest of the system.
  • Centrality ($r_i + s_j$): The overall importance or "prominence" of a factor.
  • Cause Degree ($r_i - s_j$): A positive value indicates a "cause" (active driver), while a negative value indicates a "result" (passive outcome).

Key Insights and Discussion

The Dominance of Resource Endowment

The analysis reveals that the Resource Endowment (X1) dimension possesses the highest centrality (0.0360). This underscores that Shaanxi’s carbon profile is primarily dictated by its physical and demographic constraints—specifically its energy mix and population distribution—rather than its current economic or technological status.

Population Density as the Primary Driver

Among the 17 specific indicators, population density (x2) emerged as the most significant variable, with the highest centrality (0.1564) and a strong positive cause degree. This identifies it as a "prime mover" in the system; shifts in where and how the population is concentrated trigger cascading effects across energy consumption and industrial demand.

Causal vs. Resultant Factors

The study effectively categorizes factors into active levers and passive symptoms:

  • Active Drivers (Causes): Population density (x2) and unit energy output efficiency (x7). These are the fundamental areas where policy intervention can yield systemic change.
  • Passive Outcomes (Results): Urbanization rate (x17), industrial energy intensity (x12), and sewage treatment rates (x9). These variables tend to react to changes in the primary drivers rather than initiating them.

The Role of Technology

Interestingly, the "Technology Level" dimension (X2) exhibited a negative cause degree. This suggests that in Shaanxi's current developmental stage, technological improvements are often reactive—deployed to mitigate the impacts of resource consumption—rather than serving as an independent driver of the low-carbon transition.

Conclusion and Policy Recommendations

The OLS-DEMATEL analysis demonstrates that Shaanxi’s carbon emissions are the result of a complex, interdependent system where demographic and resource factors play the leading roles.

To achieve its low-carbon goals, the province should:

  1. Optimize Population Distribution: Align demographic growth with the environmental carrying capacity of different sub-regions.
  2. Enhance Energy Efficiency: Focus on "unit energy output" (economic value per unit of energy) as a primary technological lever.
  3. Transition the Energy Mix: Accelerate the shift toward non-fossil fuels to decouple economic growth from carbon-intensive resource endowment.

While the study is limited by its linear assumptions and data ending in 2016, it provides a robust framework for understanding the structural roots of carbon emissions in a major industrial province.

Summary updated

BibTeX
@misc{li2019analysis,
  author = {Jinxing Li and Jing Wang and Meizhi Wang},
  title = {{Analysis of the Influencing Factors of Energy Carbon Emission in Shaanxi Province Based on DEMATEL Method}},
  year = {2019},
  doi = {10.2991/wrarm-19.2019.34},
}
Figure