Industrial robots strengthen Chinese cities' ability to withstand energy shocks, chiefly by modernising industry and boosting green innovation; the resilience gains are larger in cities with stricter environmental rules and higher science budgets.

Does the Application of Industrial Robots Enhance Urban Energy Resilience? Evidence from China

Bingnan Guo, Mengyu Li · March 21, 2026 · Energies

openalex quasi_experimental medium evidence 7/10 relevance DOI Source PDF

Using a city-level panel (2009–2023) and Double Machine Learning, the paper finds that greater industrial robot adoption significantly improves urban energy resilience in Chinese cities, primarily through industrial structure upgrading and green technology innovation, with stronger effects where environmental regulation and science spending are higher.

Against the backdrop of the in-depth adjustment of the global energy pattern and the accelerated advancement of the energy transition, coupled with the frequent occurrence of extreme climate events and the continuous intensification of risks such as supply fluctuations and external shocks faced by urban energy systems, improving urban energy resilience has become a core measure for all countries to address the vulnerability of energy systems and promote urban sustainable development. As a core technical carrier of intelligent manufacturing, the enabling role of industrial robots (IRs) in enhancing urban energy resilience (UER) has also become an important research topic in the field of the energy economy. This paper takes 280 prefecture-level and above cities in China from 2009 to 2023 as research samples and empirically examines their impact effects by constructing a Double Machine Learning (DML) model, transmission mechanism, and moderating effect of IRs on UER and ensures the reliability of conclusions through various robustness tests. The research findings indicate that IRs significantly promote the improvement of UER; industrial structure upgrading and green technology innovation are the main mediating paths, verifying how IRs affect UER from two different aspects and both environmental regulation (ER) and science expenditure (SE) positively moderate the promoting effect of IRs on UER, with the coefficients of the interaction terms being significantly positive. Robustness tests show that the core conclusions are highly reliable. This study fills the research gap in the transmission mechanism between IRs and UER and provides empirical evidence for the formulation of relevant policies. Accordingly, it is proposed that governments should strengthen the policy support for the application of industrial robots in high-energy-consuming industries, optimize the synergy mechanism between environmental regulation and scientific and technological expenditure, guide the deep integration of industrial robots with industrial structure upgrading and green technology innovation, and formulate differentiated promotion strategies based on regional energy resilience characteristics and industrial development foundations, so as to fully release the energy-resilience-improvement effect of industrial robots.

Summary

Main Finding

Industrial robots (IRs) significantly improve urban energy resilience (UER) in China (280 prefecture-level+ cities, 2009–2023). The effect operates mainly through industrial-structure upgrading and green technology innovation, and it is strengthened by stronger environmental regulation (ER) and higher science expenditure (SE). Results are estimated with a Double Machine Learning (DML) approach and hold under multiple robustness checks.

Key Points

Scope: Panel of 280 Chinese cities, 2009–2023.
Estimation: Causal effect of IR deployment on UER estimated using Double Machine Learning.
Main mediators:
- Industrial structure upgrading — automation induces higher-value, less energy-intense industry composition.
- Green technology innovation — IR adoption stimulates or complements green tech development that raises system resilience.
Moderation:
- Environmental regulation (ER) and science expenditure (SE) both positively moderate the IR → UER effect (interaction coefficients significantly positive).
Robustness: Findings remain highly reliable across alternative specifications and robustness tests reported by the authors.
Policy recommendations from the paper:
- Support IR adoption in high-energy-consuming sectors.
- Align environmental regulation and R&D spending to amplify IR benefits.
- Promote coordinated integration of IRs with industrial upgrading and green innovation.
- Use regionally differentiated strategies based on local resilience and industrial bases.

Data & Methods

Data: City-level panel data for 280 prefecture-level and above Chinese cities, covering 2009–2023.
Outcome: Urban energy resilience (UER) — city-level composite indicator (paper describes UER as the dependent variable).
Treatment: Industrial robot deployment/intensity at the city level.
Identification strategy: Double Machine Learning (DML) to flexibly control high-dimensional confounders and estimate treatment effects with reduced bias.
Mechanism analysis: Mediation models testing industrial-structure upgrading and green technology innovation as transmission channels.
Moderation analysis: Interaction terms with environmental regulation and science expenditure to test policy/contextual amplification.
Reliability: Multiple robustness checks reported (alternative specifications and tests) that support the main conclusions.

Implications for AI Economics

Automation as resilience-enhancing capital: The study provides empirical evidence that automation (industrial robots) can raise macro-level energy resilience through structural change and technological innovation, shifting the narrative from solely productivity/labor effects to system-level resilience benefits.
Complementarity of policy and AI investment: Positive moderation by ER and SE shows that regulatory and R&D environments shape the returns to automation—implying that industrial robot impacts depend on institutional context and public spending decisions.
Distributional and structural considerations: While IRs foster industrial upgrading and green innovation, economists should assess labor-market dislocation, re-skilling needs, and regional inequality when evaluating net welfare effects.
Methodological takeaway: DML is a useful causal-inference tool in AI-economics research for estimating heterogeneous and mediated effects of technology adoption with many confounders.
Directions for future research:
- Firm- and sector-level causal channels connecting robot adoption to energy use and resilience.
- Long-run dynamics: persistent vs. transient effects of automation on resilience and growth.
- Quantification of welfare trade-offs (productivity vs. employment) and optimal policy mixes (ER + SE + robot subsidies).
- Cross-country comparisons to test external validity and institutional interactions.

Assessment

Paper Typequasi_experimental Evidence Strengthmedium — Large panel, modern DML methods, mediation and moderation analysis, and multiple robustness checks increase credibility; however, the lack of plausibly exogenous variation (e.g., an instrument or natural experiment) leaves open risks of reverse causality and omitted time-varying confounders, so causal claims remain somewhat tentative. Methods Rigormedium — Use of Double Machine Learning is methodologically appropriate for high-dimensional confounding and improves over simple OLS; the inclusion of mediation/moderation analysis and robustness tests indicates careful work, but rigor is limited by reliance on observational variation, potential measurement issues for both industrial robots and urban energy resilience, and (as reported) no explicit exogenous identification strategy. SampleAnnual panel of 280 prefecture-level and above Chinese cities from 2009 to 2023 (up to ~15 years per city); dependent variable is an index or measure of urban energy resilience (UER); key treatment is city-level industrial robot (IR) adoption/intensity; controls include variables for industrial structure, green innovation, environmental regulation, science expenditure and other city-level socio-economic covariates (data sources not fully specified in the summary). Themesinnovation adoption IdentificationEstimates causal effects using Double Machine Learning (DML) on a panel of 280 Chinese cities (2009–2023), flexibly controlling for high-dimensional observable covariates (likely city and year fixed effects and other controls) via machine‑learning nuisance estimators; identification therefore relies on conditional independence (selection on observables) and functional form robustness rather than exogenous variation or instrumental variables. GeneralizabilityChina-specific institutional, industrial and energy-policy context may not generalize to other countries, Prefecture-level city sample excludes rural areas and small towns, limiting applicability to non-urban or smaller jurisdictions, Findings pertain to industrial robots (manufacturing automation) and may not transfer to other forms of AI (e.g., generative models, service robots), Time period (2009–2023) includes structural changes; effects may differ as technology, energy systems and policy evolve, Possible measurement differences in UER or IR adoption across cities could limit comparability

Claims (6)

Claim	Direction	Confidence	Outcome	Details
Industrial robots (IRs) significantly promote the improvement of urban energy resilience (UER). Organizational Efficiency	positive	high	urban energy resilience	n=280 0.48
Industrial structure upgrading is a main mediating path through which IRs improve urban energy resilience. Organizational Efficiency	positive	high	urban energy resilience (mediated by industrial structure upgrading)	n=280 0.48
Green technology innovation is a main mediating path through which IRs improve urban energy resilience. Organizational Efficiency	positive	high	urban energy resilience (mediated by green technology innovation)	n=280 0.48
Environmental regulation (ER) positively moderates the promoting effect of IRs on urban energy resilience; the interaction term coefficient is significantly positive. Organizational Efficiency	positive	high	urban energy resilience (moderation by environmental regulation)	n=280 0.48
Science expenditure (SE) positively moderates the promoting effect of IRs on urban energy resilience; the interaction term coefficient is significantly positive. Organizational Efficiency	positive	high	urban energy resilience (moderation by science expenditure)	n=280 0.48
Robustness tests confirm that the core conclusions about IRs improving urban energy resilience and the identified mechanisms/moderators are highly reliable. Organizational Efficiency	positive	high	robustness of estimated effects on urban energy resilience	n=280 0.48