TL;DR
This study explores the complex relationship between industrial agglomeration (clustering), energy intensity, and eco-efficiency across 30 Chinese provinces. It reveals a non-linear, "inverted N-shaped" relationship where clustering initially hinders, then helps, and eventually hurts environmental efficiency, with energy intensity acting as a critical mediating factor.
What Problem Does This Paper Address?
China faces a monumental challenge: maintaining industrial growth while striving for a "Carbon Peak" by 2030 and "Carbon Neutrality" by 2060. Central to this is Industrial Eco-Efficiency (IEE)—a measure of how much economic value an industry creates relative to its environmental footprint.
While it is well-known that clustering industries together (industrial agglomeration) can create "economies of scale," it also creates "crowding effects" like localized pollution and resource competition. Previous research has been divided on whether clustering is a net positive or negative for the environment. This paper seeks to resolve these contradictions by looking at two overlooked factors: spatial correlation (how one region's efficiency affects its neighbors) and the mediating role of energy intensity (how clustering changes energy use patterns, which in turn affects eco-efficiency).
Methods
The authors employ a sophisticated three-step econometric approach:
- Super-DEA (Data Envelopment Analysis): To measure IEE, they used a "Super-SBM" model. Unlike standard DEA, which can't distinguish between multiple "perfectly efficient" units, Super-DEA allows for a more granular ranking of high-performing provinces.
- STIRPAT Model Expansion: They used the standard "Stochastic Regression on Population, Affluence, and Technology" framework but added quadratic and cubic terms for industrial agglomeration to capture non-linear effects.
- Spatial Autocorrelation Model (SAC): Recognizing that economic activity doesn't stop at provincial borders, they used an SAC model to account for spatial spillovers. They tested this using three different "weight matrices": geographic distance, economic distance, and a hybrid of both.
- Mediation Analysis: They applied the Baron and Kenny method to test if industrial clustering affects eco-efficiency directly or indirectly through its impact on energy intensity.
Datasets
The study analyzes a provincial panel dataset from mainland China (30 provinces, excluding Tibet due to data gaps) covering the decade from 2009 to 2018.
- Economic/Labor Data: Sourced from the China Statistical Yearbook.
- Energy Data: Sourced from the China Energy Statistical Yearbook.
- Environmental Data: Sourced from the China Environmental Statistics Yearbook and the Carbon Emission Accounts and Datasets (CEADs) for CO2 emissions.
Key Insights and Ideas
- The Inverted N-Curve: The most striking finding is that the relationship between clustering and eco-efficiency isn't a straight line or even a simple U-shape. It’s an inverted N.
- Phase 1: Initial clustering can actually decrease efficiency as infrastructure is built and initial pollution spikes.
- Phase 2: As the cluster matures, knowledge sharing and shared infrastructure boost efficiency.
- Phase 3: If a region becomes "over-agglomerated," crowding effects (high costs, pollution, competition) cause efficiency to drop again.
- Spatial Spillovers are Real: The paper proves that IEE has a positive spatial spillover effect. If a province improves its eco-efficiency, it tends to pull its neighbors up with it through technology transfer and policy imitation.
- Energy Intensity as the "Middleman": Industrial clustering doesn't just magically change eco-efficiency; it does so largely by changing how much energy is required to produce a unit of GDP.
Key Results
- Regional Disparities: Five regions—Beijing, Tianjin, Shanghai, Jiangsu, and Guangdong—consistently showed IEE scores above 1.0, meaning they are at the frontier of "green" industrial production. Western provinces generally lagged behind.
- Agglomeration Trends: Interestingly, the national average for industrial agglomeration showed a downward trend from 2009 to 2018, likely due to government policies shifting heavy industry away from oversaturated eastern hubs toward central and western regions.
- Mediation Confirmed: The statistical tests confirmed that energy intensity is a significant mediator. Specifically, clustering has a U-shaped impact on energy intensity, which then negatively impacts IEE.
Limitations and Future Work
The authors acknowledge a few hurdles:
- Data Granularity: The study relies on provincial-level data. The authors suggest that future research should dive into city-level or even county-level data to provide more precise policy guidance.
- Indicator Breadth: While the study includes CO2, SO2, and wastewater, the authors note that a lack of unified, long-term data for other pollutants (like soot or ammonia nitrogen) limited the scope of the "environmental load" variable.
- The COVID Factor: Since the data ends in 2018, the massive industrial shifts caused by the global pandemic are not captured. Future studies will need to account for this "new normal" in industrial sustainable development.
Summary updated