Higher labor intensity in US automotive assembly plants after transitioning to electric vehicles

The U.S. automotive industry employs around 13 million workers, nearly 1 million of whom are in manufacturing, with most engaged in internal combustion engine vehicle (ICEV) production. As automakers set targets to phase out ICEVs within the next two decades, a key Just Transition question is how battery electric vehicle (BEV) production will affect overall automotive jobs. The sector’s vulnerability to employment shocks was evident during the 2008 recession when auto manufacturing employment fell 23% in a year. Existing projections for BEV-related employment are scarce and contradictory: while some analysts foresee job creation, a common narrative asserts BEV assembly requires substantially fewer workers due to fewer powertrain parts, often cited as 30% fewer workers. Policy analyses, such as from the Economic Policy Institute, suggest outcomes depend on domestic production of EV components. This study investigates the effect of the BEV transition on assembly labor intensity in the U.S. by analyzing plants that fully transitioned from ICEVs to BEVs, aiming to clarify whether BEV assembly indeed requires fewer workers per vehicle than ICEV assembly.

Public discourse since at least 2017 has repeated the claim that BEV assembly needs 30% fewer workers than ICEV assembly, primarily due to fewer powertrain components. Reports have been mixed: some argue for potential net job losses absent policies to localize BEV component supply chains, while others point to possible job gains through new BEV-related manufacturing. Prior analyses (e.g., Economic Policy Institute; Boston Consulting Group; industry and media reports) highlight policy, supply chain localization, and product strategy as key determinants. Academic work on powertrain labor demand differences has emerged, but comprehensive, plant-level, data-driven assessments of assembly labor intensity across transitions have been limited, motivating this study’s macro-level, regionally focused, empirical approach.

The study examines three U.S. “transition plants” that fully shifted from ICEV to BEV assembly: Alameda County, CA (NUMMI to Tesla); Oakland County, MI (GM Orion Assembly to Chevy Bolt); and McLean County, IL (Mitsubishi Normal to Rivian). Vehicle production data (light-duty vehicles only) were obtained from the Automotive News Research & Data Center and linked to county-level manufacturing employment under NAICS 3361 (Motor Vehicle Manufacturing) from the U.S. Bureau of Labor Statistics Quarterly Census of Employment and Wages (QCEW) and the U.S. Census Bureau Quarterly Workforce Indicators (QWI), supplemented by local news reports to address data suppression. Labor intensity is measured as workers per 1,000 vehicles per annum (WPV): WPV(k) = W(k)/V(k) × 1000, where W(k) is average annual assembly employment and V(k) is annual vehicle output for year k. This metric normalizes workforce size by production volume; conversion to hours per vehicle is possible but hours data are not available at the county level, so WPV is reported. To ensure attribution, counties were selected that contained only one automotive assembly site. Because NAICS 3361 includes both light- and heavy-duty manufacturing, the analysis assumes negligible heavy-duty presence in selected counties (light-duty-specific NAICS 33611 data were suppressed). Years with zero production and the first year after resumption were excluded due to distortion (inflated WPV) at very low volumes. National production and employment trends (Supplementary Fig. 1) provide a reference for baseline ICEV labor intensity, which ranged from ~17 to 28 WPV over two decades (lowest 17 in 2015; highest 28 in 2021). For forward-looking projections (Fig. 6), a four-parameter sigmoid function was fit to Alameda BEV WPV data, with steady-state WPV set 30% below the ICEV baseline (representing the commonly claimed long-run advantage). Least-squares fitting yielded parameters {a, b, c, WPV0} = {448.2, −0.2791, 2012, 16.8977}. Uncertainty bands were produced via 1,000 Monte Carlo samples by drawing parameters from Gaussian distributions centered at fitted values with specified standard deviations (10% for a and b, 0% for c, 30% for WPV0). All analysis was conducted in MATLAB R2023a with Mapping Toolbox for geospatial figures.

• Across all three transition plants, BEV assembly is more labor-intensive than prior ICEV assembly when measured as WPV (workers per 1,000 vehicles): • Alameda (NUMMI→Tesla): ICEV labor intensity at peak efficiency (2006) was 15–16 WPV; during BEV production, Tesla’s labor intensity remained >45 WPV for a decade, averaging 51 WPV in 2019–2022; Table 1 summarizes 16 WPV (ICEV) vs 46 WPV (BEV) at peak productivity years. Including offsite battery cell and pack workers (Sparks, NV) raises 2022 labor intensity to ~67 WPV (~50% higher than assembly-only). • Oakland (GM Orion): ICEV labor intensity reached 17 WPV post-2010 shutdown; during Bolt BEV production, labor intensity rose to ~31 WPV (Table 1). The trajectory broadly tracked the U.S. baseline, implying BEV assembly at Orion did not reduce labor intensity relative to contemporaneous ICEV norms. Battery pack assembly was offsite and excluded. • McLean (Mitsubishi→Rivian): ICEV labor intensity was 18 WPV (2014), while early BEV ramp (2022) was 316 WPV, reflecting ramp-up inefficiencies and inclusion of onsite battery pack assembly (cells offsite).
• During early BEV ramp-up, labor intensity can exceed historical ICEV assembly by an order of magnitude (e.g., McLean 316 WPV; Alameda’s first five BEV years >50 WPV).
• Even after a decade of BEV production (Alameda), labor intensity and total employment remained roughly three times higher than during ICEV peak productivity.
• National ICEV baseline labor intensity ranged ~17–28 WPV over the last two decades (min 17 in 2015; max 28 in 2021), providing context for the observed BEV WPV levels.
• Projections suggest that achieving parity with historic ICEV WPV or reaching a long-run 30% BEV labor advantage may take more than 15 years after BEV factory launch, contingent on factors such as R&D investment, vehicle complexity, production volume, automation, and process efficiency.

The findings directly challenge the widespread assumption that BEV assembly requires roughly 30% fewer workers than ICEV assembly. Empirical evidence from three fully transitioned U.S. plants shows BEV assembly currently demands equal or higher labor per vehicle than ICEV assembly, especially during extended ramp-up periods. Alameda demonstrates that even a mature BEV site can remain substantially more labor-intensive than its prior ICEV operations, while Oakland indicates that BEV assembly may mirror national ICEV labor intensity trends rather than reduce them. McLean exemplifies the extreme labor intensity during early-stage ramp at a new automaker. Taken together, these results imply that rapid, widespread displacement of assembly jobs due to BEV transitions is less likely in the near-to-medium term; assembly headcounts may need to match or exceed prior ICEV levels for years. Factors increasing BEV labor intensity include ongoing manufacturing R&D, higher product complexity (with a present bias toward premium vehicles), and vertical integration that brings formerly offsite supplier work in-house. Conversely, scaling to high volumes, advancing automation, and process improvements should lower WPV over time. Beyond assembly, integrating battery cell manufacturing and pack assembly can add significantly to total labor per BEV (up to ~50% or more), and battery cells may account for up to 75% of powertrain labor. Whether these jobs offset declines in ICEV engine/transmission component manufacturing depends on localization, co-location, and skills alignment.

This study contributes a data-driven, plant-level assessment showing that BEV assembly at transitioned U.S. plants is currently more labor-intensive than historical ICEV assembly, particularly during prolonged ramp-up phases. It counters the simplified narrative of a 30% labor reduction in BEV assembly and suggests that parity or long-run labor savings, if attainable, likely require more than 15 years post-transition. Policy and industry implications include the need to plan for sustained assembly workforce levels, invest in manufacturing R&D and process improvements, and consider vertical integration and supply chain strategies. Future research should examine the broader employment impacts across parts manufacturing, especially battery cell and pack production, analyze regional co-location and skills requirements, and expand the dataset beyond three case studies to generalize findings across plant types, automakers, and product segments.

• The analysis relies on county-level employment data (NAICS 3361) that anonymize establishments; data suppression required triangulation with QCEW, QWI, and local news sources, which may introduce measurement error.
• NAICS 3361 includes both light- and heavy-duty vehicle manufacturing; the study assumes negligible heavy-duty presence in the selected counties.
• WPV can be distorted by very low production volumes; years with zero production and the first year after resumption were excluded, but ramp-up phases still elevate WPV.
• Hours worked per vehicle could not be calculated at the county level due to lack of publicly available hours data; WPV is used as a proxy for labor intensity.
• Battery cell and pack manufacturing labor are sometimes offsite and thus excluded (e.g., Alameda Model 3/Y packs; Oakland Bolt packs), which can understate total BEV labor intensity; in McLean, onsite pack assembly elevates WPV.
• Only three transition plants were studied; findings may not generalize to all regions, automakers, or product mixes, especially as BEV models evolve toward lower-cost, higher-volume segments.
• Projections (sigmoid fit and Monte Carlo) are illustrative and depend on assumptions (e.g., eventual 30% lower steady-state WPV), introducing forecasting uncertainty.