China’s green data-center pilot raised city inclusive green growth by around 0.9 percentage points a year, largely by mobilising government and public environmental action; gains concentrate in digital clusters and weaken with distance.
How does green digital economy policy enable inclusive green growth in cities? — from the perspective of government and public environmental participation
Using a DID on China's 2015 Green Data Center Pilot Policy, the study finds the policy increased city-level inclusive green growth by about 0.9 percentage points annually, with largest gains in high-quality pilot zones, digital economy clusters, and non-traditional industrial cities, working mainly by boosting government and public environmental participation and producing spatially decaying spillovers.
Accelerating the green and low-carbon transformation of urban development is a key pathway to achieving inclusive and sustainable development in China and around the world. This study uses the 2015 Green Data Center Pilot Policy as a quasi-natural experiment and employs the difference-in-differences (DID) method to examine the impact and mechanisms of this green digital economy policy on urban inclusive green growth. Findings indicate that the policy effectively promotes inclusive green growth in cities, increasing the average annual growth rate of inclusive green growth by 0.9 percentage points. The impact is particularly pronounced in high-quality development pilot zones, digital economy clusters, and non-traditional industrial cities. Mechanism tests reveal that green digital economy policies primarily enhance inclusive green growth by strengthening government environmental participation and public environmental participation. Concurrently, the policy exhibits distinct spatial decay characteristics, with spillover effects to neighboring regions diminishing progressively with distance. Moreover, the policy’s impact is most pronounced in cities with areas between 5,000 and 10,000 square kilometers. This study enriches research in the fields of digital economy and green development, providing empirical evidence and policy references for advancing regional low-carbon transformation and coordinated development.
Summary
Main Finding
Using the 2015 Green Data Center Pilot Policy as a quasi-natural experiment and a multi‑period DID design over 276 Chinese prefecture-level cities (2011–2023), the study finds that the green digital economy policy significantly promotes urban inclusive green growth — raising the average annual growth rate of inclusive green growth by about 0.9 percentage points. The effect operates mainly by strengthening government environmental participation and public environmental participation, shows heterogeneous strength across city types (stronger in high‑quality development pilot zones, digital‑economy clusters, and non‑traditional industrial cities), exhibits spatial spillovers that decay with distance, and is most pronounced in cities with areas between 5,000 and 10,000 km².
Key Points
- Causal identification: 2015 green data center pilot used as a quasi‑natural experiment with Treat × Post in a multi‑period DID framework and fixed effects.
- Effect size: policy increases annual inclusive green growth by ~0.9 percentage points on average.
- Mechanisms: two primary channels
- Government environmental participation: improved monitoring, green subsidies/credit, carbon accounting, digital oversight tools that guide green upgrading.
- Public environmental participation: easier access to environmental information and digital reporting, increasing public pressure and green consumption/behavior.
- Heterogeneity:
- Larger effects in high‑quality development pilot zones, digital economy clusters, and non‑traditional industrial cities.
- Strongest impacts for cities whose administrative area is 5,000–10,000 km².
- Spatial dynamics: positive spillovers to neighboring cities but magnitude declines with distance (spatial decay).
- Policy implication emphasized: aligning digital infrastructure (data centers) greening with urban planning and governance to realize both environmental and equity goals.
Data & Methods
- Sample: Panel of 276 Chinese prefecture-level cities, 2011–2023.
- Treatment/control: 66 pilot cities (green data center pilots) vs. 210 control cities, based on MIIT/NEA/National Bureau of Public Assets lists and rollout timing.
- Outcome variable (IGG): a composite Inclusive Green Growth index constructed across four dimensions — Economic Growth, Income Distribution, Welfare Inclusiveness, Pollution Reduction.
- Measurement: combined objective–subjective weighting: entropy method to derive objective indicator weights; then equal weighting across the four dimensions to compute overall IGG.
- Indicators include per-capita GDP and growth, urban/rural incomes and disparity, social insurance and public services measures, and industrial/pollution treatment statistics (SO2, effluent, waste, treatment/recycling rates).
- Empirical approach:
- Baseline: multi‑period DID (IGG_it = α0 + α1(Treat_i × Post_t) + α2 X_it + city & year fixed effects + ε_it).
- Controls: green innovation, informatization level, government regulatory intensity, industrial structure, and other standard covariates.
- Robustness and extensions reported: propensity score matching DID (PSM‑DID) for balancing, mediation/mechanism tests for government and public participation channels, spatial models to capture spillovers and distance decay, and heterogeneity tests by city type and area size.
- Data sources: National Bureau of Statistics, industry reports, local yearbooks, EPS database; missing values imputed by linear interpolation.
Implications for AI Economics
- Greening compute infrastructure matters for AI externalities: data centers are a core input for AI and cloud services. Policies that reduce their energy and carbon intensity directly lower the environmental footprint of AI workloads and models.
- Policy can shape AI’s regional distribution and inclusive impacts: green digital policies that encourage low‑carbon data center siting and energy‑efficient designs can help ensure that AI infrastructure stimulates local green growth rather than imposing environmental burdens, and can promote more equitable distribution of benefits across regions and populations.
- Government and public channels are critical levers:
- Government digital monitoring, carbon accounting, and green finance can align AI infrastructure investment incentives with low‑carbon outcomes.
- Public access to environmental information and digital reporting platforms influences demand‑side behavior and corporate responses — important when assessing social acceptability and political economy of AI infrastructure expansion.
- Spatial and planning considerations for AI infrastructure:
- Spillovers imply that green AI/data center policies produce regional externalities; centralized green pilot zones or clusters can radiate benefits but planning must account for distance decay.
- The finding about optimal city area (5,000–10,000 km²) suggests city-scale constraints/opportunities for data center siting and regional AI cluster design — relevant for infrastructure planning and cost‑benefit analysis of AI capacity expansion.
- Research and policy recommendations for AI economics:
- Incorporate green data center standards into AI infrastructure cost models and lifecycle carbon accounting for AI services.
- Evaluate subsidies, green credits, or carbon pricing targeted at AI compute to internalize environmental externalities.
- Study distributional impacts: assess how green compute policies affect labor markets, digital access, and income distribution where AI deployment occurs.
- Use spatial economic models when designing regional AI infrastructure strategies to maximize positive spillovers and avoid concentrated environmental harms.
Suggested follow-ups: quantify energy/carbon reductions attributable specifically to AI workloads under green data center policies; cost–benefit analysis of greening for AI firms; study interaction between AI automation, employment, and inclusive green growth under these policies.
Assessment
Paper Typequasi_experimental
Evidence Strengthmedium — A DID on a policy rollout provides plausible causal evidence, and the paper adds spatial spillover and mechanism tests; however, causal credibility depends on the validity of the parallel trends assumption and on whether pilot selection was endogenous (policy placement on non-random criteria). Details on pre-trends, placebo tests, robustness to time-varying confounders, and alternative specifications would be needed to elevate confidence to high.
Methods Rigormedium — The use of panel DID, heterogeneity analysis, mechanism tests, and spatial decay analysis indicates standard rigorous approaches, but the summary does not report key robustness checks (e.g., event-study pre-trends, matching/IV to address selection into pilots, sensitivity to measurement of 'inclusive green growth', clustering of standard errors, or placebo interventions), leaving room for methodological concerns.
SampleCity-level panel data for Chinese cities around the 2015 Green Data Center Pilot Policy; treated observations are pilot cities/zones and controls are non-pilot cities, with annual measures of inclusive green growth, mediators (government environmental participation, public environmental participation), and spatial/geographic information to assess spillovers (exact sample size and years not specified in the summary).
Themesgovernance innovation
IdentificationDifference-in-differences using the 2015 Green Data Center Pilot Policy as a quasi-natural experiment: treated cities (pilot zones) are compared to non-pilot cities before and after policy implementation, with tests for spatial spillovers and mechanism analysis via mediators (government and public environmental participation). Identification relies on the parallel trends assumption and on controlling for observed covariates and fixed effects.
GeneralizabilityFindings are China-specific and depend on Chinese institutional and regulatory context, Policy targets green data centers/digital infrastructure—effects may not generalize to other AI policies or technologies, City-scale outcomes may not translate to rural areas or countries with different urban structures, Heterogeneous effects by city size/industrial structure limit simple extrapolation to all cities, Results pertain to the post-2015 period and may not hold under later technological or policy changes
Claims (9)
| Claim | Direction | Confidence | Outcome | Details |
|---|---|---|---|---|
| The 2015 Green Data Center Pilot Policy effectively promotes inclusive green growth in cities, increasing the average annual growth rate of inclusive green growth by 0.9 percentage points. Fiscal And Macroeconomic | positive | high | Average annual growth rate of inclusive green growth |
0.9 percentage points
0.48
|
| The policy's positive impact on inclusive green growth is particularly pronounced in high-quality development pilot zones. Fiscal And Macroeconomic | positive | high | Inclusive green growth (relative effect size larger in high-quality pilot zones) |
0.48
|
| The policy's positive impact on inclusive green growth is particularly pronounced in digital economy clusters. Fiscal And Macroeconomic | positive | high | Inclusive green growth (relative effect size larger in digital economy clusters) |
0.48
|
| The policy's positive impact on inclusive green growth is particularly pronounced in non-traditional industrial cities. Fiscal And Macroeconomic | positive | high | Inclusive green growth (relative effect size larger in non-traditional industrial cities) |
0.48
|
| Mechanism tests indicate the policy primarily enhances inclusive green growth by strengthening government environmental participation. Governance And Regulation | positive | high | Inclusive green growth (mediated by government environmental participation) |
0.48
|
| Mechanism tests indicate the policy primarily enhances inclusive green growth by strengthening public environmental participation. Governance And Regulation | positive | high | Inclusive green growth (mediated by public environmental participation) |
0.48
|
| The policy exhibits spatial spillover effects on neighboring regions that diminish progressively with distance (spatial decay). Fiscal And Macroeconomic | positive | high | Inclusive green growth in neighboring cities/regions (spillover magnitude decreases with distance) |
0.48
|
| The policy’s impact on inclusive green growth is most pronounced in cities with areas between 5,000 and 10,000 square kilometers. Fiscal And Macroeconomic | positive | high | Inclusive green growth (heterogeneous effect by city area) |
0.48
|
| The study uses the 2015 Green Data Center Pilot Policy as a quasi-natural experiment and employs the difference-in-differences (DID) method to identify the policy's impact on urban inclusive green growth. Governance And Regulation | null_result | high | Research design / identification strategy (DID using 2015 policy) |
0.8
|