Impacts of workplace automation on energy poverty: The new challenge of achieving SDG 7 in the context of technological revolution

Xiaoru Niu, Chao Li, Xiang Li, Yuhan ZhangHeliyon 2024 · DOI

TL;DR

This study reveals that workplace automation, driven by the current technological revolution, significantly increases household energy poverty. This effect is primarily mediated by reduced income and social capital, with disproportionate impacts on vulnerable populations, highlighting a new challenge for achieving Sustainable Development Goal 7.

What Problem Does This Paper Address?

The paper tackles a critical, yet under-explored, challenge at the intersection of technological advancement and global development goals: the impact of workplace automation on energy poverty. While automation is often discussed in terms of economic growth and productivity, its demand-side effects on household energy access and affordability remain largely unexamined. Energy poverty, defined as the lack of access to affordable, reliable, and sustainable energy, affects billions worldwide and hinders economic development and well-being. Sustainable Development Goal 7 (SDG 7) aims to ensure universal access to modern energy services by 2030. However, the rapid rise of automation, particularly with AI, presents a potential new obstacle to achieving this goal, especially in developing economies like China, which is a global leader in automation adoption. The authors aim to bridge this knowledge gap by systematically investigating how automation influences energy poverty and identifying the underlying mechanisms and mitigating factors.

Methods

The researchers employ a multi-faceted econometric approach to analyze the relationship between workplace automation and energy poverty.

  • Benchmark Regressions: Ordinary Least Squares (OLS), Probit, Logit, Ordered Probit, and Ordered Logit models are used to establish an initial link between automation and energy poverty.
  • Instrumental Variable (IV) Approach: To address potential endogeneity issues (where automation and energy poverty might influence each other or be jointly determined by unobserved factors), a Two-Stage Least Squares (2SLS) model is utilized. The instrumental variable used is the "routine operational task-intensity" of occupations from a 1977 US Department of Labor dataset, which is theoretically uncorrelated with current worker characteristics but predictive of current automation susceptibility.
  • Robustness Checks: To ensure the reliability of their findings, the authors conduct several robustness tests:
  • Different Automation Measures: They use various indices to quantify automation's impact on occupations, including weighted, unweighted, median, and maximum values, as well as indices from other scholars (RTI and RII).
  • Placebo Tests: The automation indicator is randomly reallocated thousands of times to ensure the observed effect is not due to chance or omitted random factors.
  • Penalized Machine Learning: Lasso, Ridge, and Elastic Net models are employed to assess automation's explanatory power relative to other factors.
  • Varying Energy Poverty Measures: The study uses different thresholds for energy affordability (10%, 15%, 20% of income/expenditure) and considers energy subsidies to define energy poverty, as well as relative poverty measures.
  • Mechanism Analysis: Mediation analysis is performed using a system of equations to test how income and work-related social capital act as pathways through which automation affects energy poverty.
  • Heterogeneity Analysis: Moderation models are used to examine how factors like urban-rural background, human capital (education, migration), labor protection (contracts, union membership), and energy supply quality (price, stability, security, accessibility) influence the automation-energy poverty relationship.

Datasets

The primary dataset used is the Chinese General Social Survey (CGSS) 2018 wave. This dataset is chosen for its comprehensive coverage of household expenditure, income, energy consumption details (electricity, various fuels), and detailed occupational information (using ISCO2008 codes). The authors also reference the Dictionary of Occupational Titles (DOT) from 1977 and the US Department of Labor's Occupational Information Network (O*NET) for constructing automation measures.

Key Insights and Ideas

  • Automation as an Energy Poverty Driver: The most striking insight is that, contrary to potential assumptions about efficiency gains, workplace automation significantly increases household energy poverty. This challenges the notion that technological progress automatically translates to improved living standards for all.
  • The Income and Social Capital Nexus: The study pinpoints reduced income and diminished work-related social capital as the primary mechanisms through which automation exacerbates energy poverty. This highlights the dual impact on individuals' financial resources and their social support networks.
  • Vulnerability Amplification: Automation's negative effects are not evenly distributed. Rural households, less educated individuals, non-migrants, and those without formal labor protections are disproportionately affected, underscoring the need for targeted interventions.
  • Energy Supply Quality as a Buffer: Improving the quality of energy supply—making it more reasonable in price, stable, secure, accessible, and convenient—can help mitigate the adverse impacts of automation on household energy consumption.
  • Human Capital and Labor Protection as Mitigators: Investing in human capital (education, training) and strengthening labor protections (contracts, union membership) can significantly weaken the negative shock of automation on energy poverty.

Key Results

  • Significant Positive Impact: Across various models and robustness checks, workplace automation consistently shows a statistically significant positive coefficient on energy poverty, indicating that increased automation leads to higher energy poverty.
  • Mechanism Confirmation: Income is found to mediate automation's effect, with automation reducing income, which in turn increases energy poverty. Similarly, automation negatively impacts work-related social capital, which then contributes to higher energy poverty.
  • Heterogeneity Findings:
    • Urban households experience a milder impact of automation on energy poverty compared to rural households.
  • Higher education levels and migrant status reduce the negative impact of automation.
  • Labor protection measures (labor contracts, trade union membership, in-system employment) significantly buffer the adverse effects of automation.
  • Energy Supply Improvements: Enhancing price reasonability, stability, security, accessibility, and convenience of energy supply all significantly reduce the adverse impact of automation on energy poverty.

Limitations and Future Work

  • Data Limitations: While the CGSS is comprehensive, the study relies on the 2018 wave, and the occupational codes used for automation measures are based on older classifications (ISCO2008, O*NET 2010). Future work could explore more recent data and finer-grained automation metrics.
  • Causality Nuances: While the IV approach addresses endogeneity, the complex interplay of factors in the energy and labor markets means that establishing definitive causality remains challenging.
  • Generalizability: The study focuses on China. While China's experience with automation is significant, further research could explore these dynamics in other developed and developing economies to assess global generalizability.
  • Policy Design: The authors suggest that future work could delve deeper into designing specific policy interventions based on the identified mechanisms and heterogeneity factors, such as tailored social security policies, enhanced education and training programs, and improved energy infrastructure.

Summary updated

Abstract

This paper systematically examines how workplace automation impacts energy poverty from a demand-side perspective, revealing a new challenge for Sustainable Development Goal 7 (SDG 7) in the context of technological revolution. Our research demonstrates that workplace automation significantly increases household energy poverty. This finding is robust when using the instrumental variable approach to tackle endogeneity, as well as employing different automation and energy poverty measures, placebo tests, and machine learning methods for robustness checks. Automation's impact mechanism is that it reduces people's income and work-related social capital, thus exposing households to higher risks of energy poverty. Moreover, its consequences are more prominent for rural households, less educated people, non-migrants, those without labor contracts, non trade-union members, and out-of-system workers. Thus enhancing human capital, promoting free movement of workers, and providing better labor protection contribute to weakening the adverse impact of the technological shock. Meanwhile, we find that improving the price reasonability, stability, security and accessibility of energy supply can also mitigate the negative effects of workplace automation on household energy consumption. In the dual context of the fourth technological revolution promoting industrial automation as well as the increasing urgency to achieve SDG 7, findings of this paper have important policy implications.
BibTeX
@inproceedings{niu2024impacts,
  author = {Xiaoru Niu and Chao Li and Xiang Li and Yuhan Zhang},
  title = {{Impacts of workplace automation on energy poverty: The new challenge of achieving SDG 7 in the context of technological revolution}},
  booktitle = {Heliyon},
  year = {2024},
  doi = {10.1016/j.heliyon.2024.e25087},
}
Figure