China’s green data‑centre pilots lower firms’ energy intensity by boosting ESG performance and green spending, with the biggest gains in resource‑based cities, competitive industries and mature firms led by managers combining green and digital vision. As successive pilot rounds roll out the aggregate payoff first rises and then fades, implying early concentrated benefits and later dilution.

How Does Urban Green Data Center Policy Empower Corporate Energy Utilization Efficiency? Evidence from China

Guoliang Wang, Haoxuan Sheng, Yue Lu · March 12, 2026

openalex quasi_experimental medium evidence 8/10 relevance DOI Source PDF

China's urban green data center pilot policies increased listed firms' energy utilization efficiency primarily by raising ESG performance and green investment, with effects varying by city type, industry competition, firm lifecycle, and executives' green/digital cognition, and with aggregate impacts that rise then fall as pilot batches expand.

<title>Abstract</title> Against the backdrop of China’s “dual-carbon” goals and the deepening national digital transformation strategy, urban green data center development has emerged as a key policy instrument for improving firms’ energy utilization efficiency and advancing the green and low-carbon transition. Drawing on unbalanced firm-year panel that links Chinese A-share listed firms to prefecture-level cities from 2012 to 2024, this study employs a staggered-adoption difference-in-differences (DID) framework to systematically estimate the effect of urban green data center policy on corporate energy utilization efficiency and to clarify the mechanisms at play. The results show that the policy significantly enhances firms’ energy utilization efficiency. Mechanism tests suggest that the effect operates primarily through two channels: improved ESG performance and policy-induced increases in firms’ green investment. Heterogeneity analyses indicate that the policy impact is stronger in resource-based cities, in more competitive industries, and among mature-stage firms. Moderation analyses further show that executives’ green cognition significantly strengthens the policy’s marginal effect on energy utilization efficiency. By contrast, executives’ digital cognition tends to weaken the policy effect when considered alone; however, when executives simultaneously exhibit high levels of both green and digital cognition, their synergistic alignment yields a significantly positive enabling effect. Specifically, an integrated cognitive profile helps alleviate the adjustment costs of early-stage digital transformation and fosters integrated innovation between green and digital technologies, thereby further reinforcing the policy’s enabling role. In addition, as successive pilot batches expand, the policy’s diffusion effect exhibits a nonlinear trajectory, increasing initially and then declining thereafter. Building on these findings, this study derives policy implications and provides China-specific empirical evidence and managerial insights for developing countries seeking to achieve energy transition and sustainable development while unlocking digital dividends.

Summary

Main Finding

The national urban green data center policy in China significantly improves corporate energy utilization efficiency. The effect operates mainly through improved firm ESG performance and increased firm-level green investment. Policy impacts vary by context and managerial cognition: stronger effects occur in resource-based cities, more competitive industries, and mature firms; executives’ green cognition amplifies the policy benefit, while executive digital cognition alone can weaken it (via a potential computing-rebound tendency). However, when executives are high in both green and digital cognition, the two interact synergistically to further strengthen the policy’s positive effect. The diffusion effect of successive pilot batches is nonlinear—rising at first and then declining.

Key Points

Policy target and instruments: The green data center policy integrates green/low‑carbon goals with digital infrastructure upgrading (e.g., energy-efficiency thresholds, PUE benchmarks such as ≤1.30, and a National Green Data Center selection program). It combines supply-side cleaner computing power and normative benchmarking to induce firm-level change.
Primary result: Coverage by the green data center policy is associated with measurable improvements in firms’ energy utilization efficiency.
Mechanisms:
- Technological spillovers and access to efficient cloud computing reduce firms’ energy inputs per unit of computing/service.
- Policy-induced improvements in ESG performance encourage managerial and operational changes that lift energy efficiency.
- Increased firm green investment (CAPEX directed at energy-saving/green tech) transmits the policy into operational outcomes.
Heterogeneity:
- Larger effects in resource-based cities (where energy constraints and transition needs are acute).
- Stronger in more competitive industries (pressure to optimize costs and efficiency).
- Stronger for mature-stage firms (better capacity to invest/absorb change).
Managerial cognition:
- Executive green cognition strengthens policy transmission—executives who prioritize environmental value are more likely to convert infrastructure availability into energy-efficiency actions.
- Executive digital cognition shows a mixed role: it can help absorb digital resources but, by itself, may encourage expansion of computing capacity and produce rebound effects that reduce energy-utilization gains.
- Jointly high green + digital cognition produces a positive synergistic effect—aligning purpose (green) with capability (digital) lowers adjustment costs and encourages integrated green-digital innovation.
Policy diffusion: As pilot batches expand, marginal benefits first increase (learning and demonstration) and later decline (diminishing returns, dispersion of best-practice effects).

Data & Methods

Data: Unbalanced firm-year panel linking Chinese A-share listed firms to prefecture-level cities, covering 2012–2024.
Empirical approach: Staggered-adoption difference-in-differences (DID) design exploiting variation in timing and location of green data center policy rollout to identify causal effects on firm energy utilization efficiency.
Identification strategy: Firms in cities exposed to green data center pilots (or later expansion batches) are compared to firms in non-exposed cities before/after adoption to estimate average treatment effects; mechanism tests and heterogeneity analyses probe pathways and conditional effects.
Mechanism and moderation tests: Examined mediators (ESG performance, firm green investment) and moderators (executive green cognition, executive digital cognition). (The paper grounds these choices in upper‑echelons theory; specific operationalizations of cognition and efficiency measures are not detailed in the provided excerpt.)
Robustness: The study reports heterogeneity, moderation, and diffusion dynamics to strengthen causal interpretation; typical robustness checks for staggered DID (e.g., event-study patterns, pre-trends) are implied though not enumerated in the excerpt.

Implications for AI Economics

For AI deployments and cloud/edge compute policy:
- Green-certified data centers and PUE standards can materially affect firm-level energy efficiency, so green infrastructure procurement should be central to AI and cloud strategy.
- Policymakers should combine technical standards (e.g., PUE thresholds, energy-efficiency grades) with benchmarking and demonstration programs to accelerate diffusion of green compute.
Rebound risks for AI-driven expansion:
- High digital capability without green alignment risks a computing-rebound effect: firms expand compute and can increase total energy use, eroding energy-utilization gains. AI policy must account for demand-side effects of cheaper/greener compute.
- Complementary measures (energy-aware AI pricing, caps, incentives for model/algorithmic efficiency, or accounting rules for compute-related emissions) can mitigate rebound.
Managerial and organizational interventions:
- Firms should promote combined green + digital leadership (training, incentive alignment, hiring) to unlock both the technological and strategic channels that translate green compute into real energy-efficiency gains.
- Investors and regulators can use executive cognitive indicators as signals of a firm’s ability to convert digital infrastructure into sustainable outcomes.
Designing green-digital policies for developing economies:
- The Chinese experience suggests integrated policy (green + digital) is more effective than fragmented approaches. Developing countries should bundle digital infrastructure support with clear energy-efficiency standards and demonstration programs.
- Targeted support for resource-constrained or early-stage firms—whose capacity to absorb and align green-digital change is limited—can avoid uneven effects.
Research directions in AI economics:
- Quantify compute-intensity and energy impacts of firm-level AI adoption under green infrastructure regimes.
- Study policy mixes that internalize rebound effects (e.g., compute taxes, energy-based usage pricing, or algorithmic-efficiency incentives).
- Examine measurement and operationalization of executive green/digital cognition and how managerial traits interact with firm AI strategies and carbon outcomes.

Limitations / notes - The excerpt does not report exact measures used for energy utilization efficiency, executive cognition, or control variables—interpretation should consider those measurement details when applying results. - Results are China-specific; mechanisms are broadly informative but local institutional features (market structure, regulatory capacity, data-center market) matter when generalizing to other countries.

Assessment

Paper Typequasi_experimental Evidence Strengthmedium — The staggered DID design provides plausibly causal variation and the paper tests channels and heterogeneity, but identification depends on parallel trends and on the assumption that pilot placement/timing is exogenous; potential concerns include selection into pilots, policy spillovers across jurisdictions, measurement/endogeneity of ESG and green investment, and limited detail on whether modern estimators were used to address heterogeneous treatment effects in staggered designs. Methods Rigormedium — The study uses a standard and appropriate quasi-experimental approach, mediator and moderator analysis, and multiple robustness checks, but the description lacks mention of key diagnostics and state-of-the-art corrections (e.g., event-study plots, staggered-DID estimators robust to heterogeneous effects, placebo tests, matching or IV for pilot assignment), and executive-cognition measures may be subject to measurement error. SampleUnbalanced firm–year panel of Chinese A-share listed firms matched to prefecture-level cities, 2012–2024, with corporate energy utilization efficiency as the main outcome and firm-level mediators (ESG scores, green investment); firm- and industry-level covariates, measures of industry competition, resource-based city indicator, firm life-cycle stage, and executive green/digital cognition used for heterogeneity and moderation tests (sample size not reported here). Themesadoption governance productivity IdentificationStaggered-adoption difference-in-differences exploiting variation in timing and location of prefecture-level urban green data center pilot policies, with firm and year controls and tests for heterogeneity, mediation (ESG, green investment), and moderation by firm/executive characteristics (parallel-trends assumption implied). GeneralizabilityLimited to publicly listed Chinese firms (A-share) — excludes private firms and SMEs, Findings reflect China's institutional and regulatory context; may not generalize to other countries, Policy is prefecture-level urban pilot — results may differ for national programs or purely private data-center investments, Time window (2012–2024) may capture idiosyncratic technological or policy regimes not persistent into the future, Effects on AI-specific outcomes (e.g., model costs, AI deployment rates, labor/productivity from AI) are indirect and not directly measured

Claims (12)

Claim	Direction	Confidence	Outcome	Details
Implementation of urban green data center pilot policies leads to measurable improvements in firms' energy utilization efficiency. Organizational Efficiency	positive	high	corporate energy utilization efficiency	0.48
Improved firm ESG performance mediates part of the positive effect of the green data center pilot policy on corporate energy utilization efficiency. Organizational Efficiency	positive	medium	corporate energy utilization efficiency (mediated by ESG performance)	0.29
Policy-induced increases in firms' green investment constitute another primary channel through which the pilot policy improves energy utilization efficiency. Organizational Efficiency	positive	medium	corporate energy utilization efficiency (mediated by green investment flows)	0.29
The policy's positive impact on energy utilization efficiency is stronger in resource-based cities than in non-resource-based cities. Organizational Efficiency	positive	medium	corporate energy utilization efficiency	0.29
Firms operating in more competitive industries experience larger energy-efficiency gains from the green data center pilot policy. Organizational Efficiency	positive	medium	corporate energy utilization efficiency	0.29
The policy effect on energy utilization efficiency is more pronounced for mature-stage firms than for early-stage firms. Organizational Efficiency	positive	medium	corporate energy utilization efficiency	0.29
High executive green cognition strengthens the marginal positive effect of the green data center pilot policy on firms' energy utilization efficiency. Organizational Efficiency	positive	medium	corporate energy utilization efficiency	0.29
High executive digital cognition on its own tends to weaken the policy's positive effect on energy utilization efficiency (interpreted as short-run adjustment costs from digital transformation). Organizational Efficiency	negative	medium	corporate energy utilization efficiency	0.29
When executives have both high green cognition and high digital cognition, the two cognitions reinforce each other, producing a significantly positive enabling effect on the policy's impact (facilitating integrated green+digital innovation and reducing adjustment frictions). Organizational Efficiency	positive	medium	corporate energy utilization efficiency	0.29
As successive pilot batches of urban green data center policies are rolled out, the aggregate policy impact follows a nonlinear rise-then-fall (increase followed by decline) diffusion trajectory. Organizational Efficiency	mixed	medium	aggregate policy impact on corporate energy utilization efficiency over pilot-batch rollout	0.29
The main results are robust to inclusion of controls and a range of heterogeneity and moderation checks, supporting that findings are not driven by simple time trends or obvious confounders. Organizational Efficiency	positive	high	corporate energy utilization efficiency (stability of estimated policy effect across specifications)	0.48
The pilot policy is associated with increases in firm-level ESG scores and green-investment flows (direct effects of policy on the mediators). Organizational Efficiency	positive	medium	ESG performance; green-investment flows	0.29