Employment dynamics during the US power sector transition reveal risks and opportunities

  • Granular analysis of the US power sector job market in terms of time and occupational detail models the transition to clean energy
  • Energy transition will modest net positive impact on jobs to 2050
  • The analysis identifies likely labour spikes, coming skills shortages, and other labour friction
  • With US government oversight and targeted retraining, it is possible to smooth the frictions in the long run, research team says

While the incoming Trump administration has pledged to reduce government support for clean energy, the US will nonetheless continue to undergo a major shift towards clean energy with significant private sector investment and technological momentum behind it. Economists and scientists at the University of Oxford have alerted US policymakers to coming skills shortages in the US power sector as this transition unfolds, in one of the most detailed pieces of modelling of the impact of the energy transition on the labour market to date.

The study, published today in the Journal Joule, provides a blueprint for the government to actively manage labour markets through the ongoing energy transition, and to help maintain the smooth-functioning US power sector to 2050. It uses the US National Renewable Energy Laboratory’s fast trajectory of 95% decarbonisation of the US power sector by 2035.

The research team from the Institute for New Economic Thinking at the Oxford Martin School, a centre which previously modelled the cost of energy transition scenarios, undertook a novel, network-based approach to measure the impact on labour of a rapid decarbonisation of the electricity sector.

While previous research has made aggregate predictions about the impact of the energy transition on jobs, this research provides granular detail, breaking it down into three distinct phases: (1) scale-up, (2) scale-down, and (3) low carbon power system.

For example, it highlights that in the electricity supply chain, the job scale-up is currently happening, and is likely to peak in the early 2030s with a large spike in labour demand which declines again in the later half of the decade.

“This enables us to identify skills mismatches, which have been previously overlooked, and provide a novel framework for analysing occupation-specific skill mismatches as they evolve during the clean energy transition” Lead author Joris Bücker of the Institute for New Economic Thinking at the Oxford Martin School said.

“Our goal is to alert policymakers to possible frictions and show where targeted retraining and labour market monitoring may be useful”.

“The transition to a world powered by clean energy will involve transforming the energy labor market as well. We show that this transition has the potential to generate labor market fluctuations and skill mismatches over time. Compared to the size and normal fluctuations of the US labor market, the impact of this transition will be modest. However, impacts on particular occupations could be significant and, without proper planning, specific industries will struggle to find skilled labor and displaced workers will have difficulty finding jobs.”


Results: Stranded labour & skill shortages

The study, ‘Employment dynamics in a rapid decarbonization of the US power sector’ finds that the energy transition will both create and destroy jobs.

Mapping phases of labour demand through the transition over time, identifying industries and occupations where there will be skills shortages the study predicts:

  • A medium term spike in labour demand for manufacturing and construction industries during the phase-up and phase-down ‘transition phases’;

  • Consistent growth in demand for some occupations, including electrical power line installers and wind turbine service technicians;

  • Falling demand for others, such as power plant operators, distributors and dispatchers.

“Decarbonizing the power sector goes beyond the ‘green vs. brown jobs’ debate. It’s really about when specific skills are needed—like construction surging early to build clean energy infrastructure, but tapering off later,” co-author Maria del Rio-Chanona of University College London explains.


Skills transitions

The study’s conclusions call for targeted training to smooth out frictions, identifying areas where retraining would be beneficial; along with continued monitoring to validate.

Co-author Matthew Ives, of the Smith School of Enterprise and Environment, said he hoped their modelling efforts will alert governments to the need to plan for the impacts of the unfolding energy transition to help avoid frictions, including skills mismatches and stranded workers unemployed by declining sectors.

“With government oversight and targeted retraining, it is possible to smooth the frictions in the long run to ensure that the unfolding energy transition in the US power sector - that is now well-underway - delivers the net jobs boost for the labour market that is predicted.”

Lead author Joris Bücker concluded that he hoped the research would be useful to policymakers in the US power sector who are thinking about the long term-trajectory of the industry:

“Overall, we find a modest net positive growth of jobs in the US power sector to 2050 on this trajectory. The challenge will be to deal with modest skills mismatches, which may have a big impact in certain regional contexts in the US. Government support would be best targeted at planning to smooth these frictions.

“Nationwide, we find plenty of opportunities for fossil-fuel workers in the short term, as the transition phases unfold. Government attention should also include the later phases of the transition, where more intervention may be needed as construction activity slows."


Data visualisation of results

An accompanying data visualization project aims to illuminate and explore the intricate dynamics of labor transitions in the face of a net-zero emissions power system, offering both a visual narrative and an interactive dashboard.

In this research, the demand shock is determined by translating power sector decarbonization costs into annual changes in labor demand. Occupation data is sourced from the 2018 US data published by the Bureau of Labor Statistics. Using the 2018 data is to have an estimate of a relatively stable economic situation before the COVID pandemic. The ease of occupational transition is determined by the relatedness between occupations from O*NET data.

With thanks to Complexity Science Hub Vienna Data Visualisation Engineer Liuhuaying Yang.


Key links


Figures

Figure 1: Scenarios The US power sector scenarios we use in this study. The upper panels show the capacities in GW and the lower panels the electricity generation in TWh in yearly resolution. On the left, we show NREL’s no-new-policy reference scenario that we use as the counterfactual and on the right NREL’s fast 95% by 2035 scenario. Source: NREL, with technological categories aggregated according to SM Table S1: Gas electricity also includes gas with carbon capture and storage (CCS) technology. Up to 2020, the figures show historical data from the Electric Power Annual 2020.

Figure 2: Overall demand for workers Total additional demand change for workers in the 95% decarbonization by 2035 scenario (a) per aggregated industry and (b) per occupation category. The demand change is net of the NREL no-new-policy reference scenario. Industries are plotted at the detailed level used in the analysis (82 industries) but colored by their 2-digit aggregated categories (14 of 20 categories are minimally affected and shown in gray scale). Occupations are plotted at the detailed level used in the analysis (539 occupations) and colored by their 2-digit level aggregation (13 of 22 occupation groups are minimally affected and shown in gray scale). Different phases of the transition are demarcated with dotted vertical lines and labeled.

Figure 3: Occupational demand during transition Occupation demand change relative to employment in the 95% by 2035 scenario. On the vertical axis, the net demand change between 2021–2034 (scale-up phase), and on the horizontal axis, the change between 2034–2048 (scale-down phase). The demand change is relative to the no-new-policy reference scenario. Three occupations (Wind turbine technicians, Power plant operators, and Solar PV installers) that lie outside of the rectangular zoom-in box are labeled. The zoom-in box does not cover any data point in the main plotting area. Occupations within the gray circle shown in the zoom-in box experience less than 1% demand change are considered minimally affected; all other occupations are categorized by the labor transition typology that is formed by the four quadrants, which are labeled in purple. Occupations are colored according to their mean wage. The occupational profiles on the right show the full temporal dynamics for four selected occupations. Gray error bars are constructed via the sensitivity analysis on the trajectory calculation (See SM Section D.6).

Figure 4: Network of related occupations Nodes represent occupations, and two occupations are connected if workers can switch between them, as defined by the list of related occupations from O*NET. The layout of the networks in (a) and (b) is the same, but the occupations in (a) are colored by broad occupational categories, and in (b) by their temporal profile typology. The network layout was obtained using a force-pull algorithm.

Figure 5: Skill mismatch scale-up phase Scatter plot of demand change in the scale-up phase (2021-2034) per occupation (x-axis) and their neighbors (y-axis) in the 95% by 2035 scenario, relative to the no-new-policy reference scenario. If the occupation has a positive (negative) demand change, we average the neighbor demand change over its in- (out-) neighbors. Out-neighbors of occupation α are related occupations: they form potential career switching options for workers in α. Data points using out-neighbors are shown with squares. Vice versa, in-neighbors of α are occupations for which α is a related occupation: workers in those occupations see α as a potential career switching option. Data points using in-neighbors are shown with circles. In- and out-neighbors are not necessarily the same. The identity line is shown with a dashed line, and selected occupations are highlighted. Three occupations (Wind turbine technicians, Power plant operators, and Solar PV installers) that lie outside of the rectangular zoom-in box are labeled. The zoom-in box does not cover any data point in the main plotting area. The intensity of background shading corresponds to more occupational frictions: worker frictions for x < 0, employer frictions for x > 0. The gray scaling is a linear function of the neighborhood shock, when the sign of the demand change for individual occupations is the same as for its neighbors (i.e., top right and bottom left quadrants). On the right of the main plot, demand change profiles over time are shown for occupations highlighted in red. The four quadrants are labeled by the main effect of the occupational network faced by each occupation.

Figure 6: Skill mismatch scale-down phase Scatter plot of demand change in the scale-down phase (2034-2038) per occupation (x-axis) and their neighbors (y-axis) in the 95% by 2035 scenario, relative to the no-new-policy reference scenario. If the occupation has a positive (negative) demand change, we average the neighbor demand change over its in- (out-) neighbors. Out-neighbors of occupation α are related occupations of α: they form potential career switching options for workers in α. Data points using out-neighbors are shown with squares. Vice versa, in-neighbors of α are occupations for which α is a related occupation: workers in those occupations see α as a potential career switching option. Data points using in-neighbors are shown with circles. In- and out-neighbors are not necessarily the same. The identity line is shown with a dashed line, and selected occupations are highlighted. The zoom-in box does not cover any data point in the main plotting area. The intensity of background shading corresponds to more occupational frictions: worker frictions for x < 0, employer frictions for x > 0. The gray scaling is a linear function of the neighborhood shock, when the sign of the demand change for individual occupations is the same as for its neighbors (i.e., top right and bottom left quadrants). On the right of the main plot, demand change profiles over time are shown for occupations highlighted in red. The four quadrants are labeled by the main effect of the occupational network faced by each occupation.

Latest News

View All