## 1. Exports of Electric Vehicles

## Source details

**Canonical URL:** [1. Exports of Electric Vehicles](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023116-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023116-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023116-print-pdf.pdf.json)

---

### I. Introduction — scope and motivation
- Vehicle manufacturing undergoing its most profound transformation driven by demand for environmentally friendly vehicles and regulatory restrictions on emissions.
- EVs differ from ICEVs in component composition: EVs need batteries, more wiring, and simpler electric motors; ICEVs require more complex engines, starters, fuel tanks, exhaust pipes and gears.
- EVs will increase electricity demand and reduce demand for gasoline and diesel refining and filling stations.
- Registrations of new EVs in the EU rose from 150 thousand to nearly 2 million from 2015 to 2022.
- More than 14½ million people (or 7 percent of the workforce) depend on the automotive sector directly or indirectly in the EU.
- Germany: sector ≈ 20 percent of manufacturing, 12 percent of employment, and 10 percent of goods exports.
- Policy responses cited: EU Chips Act, US Inflation Reduction Act (IRA), Production Linked Incentives in India; IRA includes “domestic content” requirements and assembly in the United States.

### II. Transition to EVs — regulatory, demand, and infrastructure trends
- Sales of EVs and hybrids exceeded 45 percent of automotive sales in 2022, from less than 2 percent in 2016.
- European EV exports in 2020 more than quadrupled relative to 2015, amounting to 7 percent of total light-weight passenger car exports.
- Country highlights (2022 new registrations):
  - Germany: EV and hybrid share almost half of total new car registrations—five times higher than in 2019.
  - Norway: more than 94 percent of new car registrations were EVs and hybrids.
  - Sweden, the Netherlands, Finland, Iceland and Denmark: shares above 55 percent.
- EV charging infrastructure: expansion outstripped EV registration growth in the 2010s—Netherlands has the highest charging station density.
- Full life-cycle GHG emissions of EVs are around half of those of ICEV; IEA (2022) estimates further 25 percent potential reduction with low-carbon electricity.
- EU regulatory targets:
  - Current aim: fleet average of 95 grams CO2 per km for cars and 147 grams CO2 per km for light commercial vehicles (EU Regulation 2019/631).
  - By 2025: cars and vans no more than 81 and 125 grams CO2 per km, respectively.
  - By 2030: 59 and 101 grams CO2 per km, respectively.
  - Current fine: €95 for each gram per kilometer more than the requirement (example: 10 grams excess × 100,000 vehicles = €95 million fine).
- European Green Deal targets: reduce emissions by 55 percent of 1990 levels by 2030, and 100 percent by 2050.
- Fit for 55 proposal: all new cars and vans sold in the EU from 2035 onward should be zero-emission vehicles; recharging infrastructure spacing rule example: no more than 60 km apart.
- Recovery and Resilience Facility (RRF):
  - RRF total: €700 bn.
  - EU countries allocated about €71bn under the RRF to sustainable mobility related expenditures.
  - RRF expenditures include tax incentives, zero-emission buses financing, charging/refueling infrastructure, R&D for hydrogen, fuel-cell component production, and construction of battery recycling sites.

### III. Empirical strategy — two complementary difference-in-difference exercises
- Time horizon: 2015-2019 around EV take-off; baseline excludes post-COVID-19 but robust to including 2020.
- Two exercises:
  1. Sectoral-level: sectoral employment across 26 European countries to test whether ICE-focused sectors grew slower where EV exports share grew faster (2015–19).
  2. Regional-level: sub-regional (NUTS2) employment to test whether regions more exposed to car manufacturing prior to EV take-off experienced relative employment losses in countries with faster EV export growth.
- Country sample (sectoral panel): AUT, BEL, BGR, CZE, DEU, DNK, ESP, EST, FIN, FRA, GBR, GRC, HRV, HUN, IRL, ITA, LTU, LUX, LVA, NLD, POL, PRT, ROU, SVK, SVN, SWE.
- Sectors included (NACE Rev 2): C2211, C2312, C2620, C2711, C2720, C2740, C2813, C2815, C2825, C2910, C2920, C2931, C2932.
- Car exports used as proxy for production by fuel type.

### IV. Sectoral evidence — main findings and magnitudes
- Dynamic panel specification: ln L_sct = λ ln L_sct-1 + β × (Share_EV_exports_ct × D_s) + γ ln Q_ct + α_sc + ε_sct
- Estimated effects:
  - One percentage point increase in EV’s share in total car exports associated with a 1.4 percent decline in employment in ICE-focused sectors relative to other sectors in the short term (β̂ = −0.014, Table 1, column 1).
  - Long-term relative decline: 6.3 percent (β̂/(1−λ̂) = −0.064).
- Median employment growth in ICE-focused sectors was 2.8 percent in 2015.
  - Median country: median increase in EV’s share (2015–2019) → 0.3 percent reduction in employment growth of ICE-focused sectors.
  - 75th percentile increase in EV’s share → drag on growth rate by 0.9 percent.
- OEM sector employment growth not significantly hampered by EV expansion (Table 1, column 2).
- Robustness checks: additional lags, sector-specific relationships with total car exports, restricting to advanced economies, trimming extremes.
- Selected coefficients (Table 1 excerpts):
  - EV exposure for ICE focused sector: −0.014* (col 1), −0.017* (col 2), −0.016* (col 3), −0.031* (col 4), −0.029* (col 5), −0.024* (col 6), −0.008* (col 7).
  - ln(employee) t-1: 0.780** (col 1), 0.779** (col 2), 0.779** (col 3), 0.887** (col 4), 0.778** (col 5), 0.670** (col 6), 0.838*** (col 7).
  - ln(car exports): 0.103* (col 1), 0.104* (col 2), 0.104* (col 3), 0.150* (col 4), 0.111 (col 5), 0.143** (col 6), 0.072** (col 7).
  - Sample sizes N: 1112 (cols 1–3 & 5), 1032 (col 4), 916 (col 6), 1090 (col 7).
- Interpretation caveats:
  - Sectoral DID cannot alone reveal whether displaced ICE-focused workers found jobs in EV-focused sectors or other industries.
  - Cross-border spillovers possible due to international production networks.

### V. Regional evidence — main findings and magnitudes
- Regional specification: L_rct = α_rc + α_ct + β × (Share_EV_exports_ct × Share_Auto_employment_rc,2015) + controls_rct + ε_rct
- Outcomes: male employment rate and male unemployment rate (focus on male workers).
- Core results (Table 2 preferred 2015–2019 with controls):
  - EV exposure coefficient on male employment rate: −0.057*** (standard error 0.016).
  - EV exposure coefficient on male unemployment rate: 0.055*** (standard error 0.016).
- Other specification coefficients reported:
  - Employment rate columns: −0.023*** (col 1, 2015-2020), −0.046*** (col 2, 2015-2019), −0.056** (col 3, 2015,2019), −0.057*** (col 4, 2015-2019 with controls).
  - Unemployment rate columns: 0.028*** (col 5, 2015-2020), 0.045*** (col 6, 2015-2019), 0.056** (col 7, 2015,2019), 0.055*** (col 8, 2015-2019 with controls).
  - Sample sizes: employment rate regressions N = 1,435; 1,230; 492; 769 (cols 1–4). Unemployment rate regressions N = 1,391; 1,207; 478; 767 (cols 5–8).
- Economic magnitudes and examples:
  - Region at upper quartile in auto employment (2 percent of workers in 2015) in a country with median increase in EV share (2 percentage points) experienced 0.2 percentage point smaller rise in employment rate than median regional increase (2.7 percent per year vs 2.9 percent per year).
  - Region at 90th percentile in auto employment (4 percent of workers in 2015) saw 0.5 percentage points lower gain in employment rate.
- Aggregate extrapolation:
  - Using average region auto employment share 1.5 percent and β̂ = −0.057, growth in EV production between 2015 and 2019 (from 0 to 7 percent of vehicle exports) reduced the employment rate by 0.3 percentage points.
  - This reduction equates to about 600,000 fewer jobs out of 189,700,000 jobs in the EU (Table 3, column 1).
- Forward-looking extrapolation (partial):
  - If EVs account for 50 percent of vehicle exports, model (β̂ = −0.057 and average auto share 1.5 percent) suggests employment rates would be about 2 percentage points lower (text fragment ends mid-discussion).

### VI. Scenarios and aggregate estimates (Table 3)
- In-sample: EV share of exports from 0 to 7 percent
  - Employment rate (ppts): -0.3
  - Implied no. of workers (millions): 0.6
  - Upper 95% C.I. (millions): 1.0
  - Lower 95% C.I. (millions): 0.3
- Out-of-sample extrapolation: EV share of exports from 7 to 50 percent
  - Employment rate (ppts): -2.0
  - Implied no. of workers (millions): 3.8
  - Upper 95% C.I. (millions): 5.9
  - Lower 95% C.I. (millions): 1.7
- Estimation uncertainty: estimate of 3.8 million lies within a 95 percent confidence interval between 1.7 and 5.9 million jobs.
- Projections: reaching the 3.8 million fewer jobs point could occur around 2027/2028 according to EAMA projections.
- Note on extrapolation: highly uncertain because data cover only up to 7 percent.

### VII. Distributional impacts by education and age
- By education (Table 4 EV exposure coefficients):
  - All: 0.055*** (0.016)
  - Primary Education or Below: 0.032 (0.046)
  - Secondary Education: 0.066*** (0.018)
  - Tertiary Education: 0.014 (0.015)
- Interpretation: middle-skill jobs (secondary education) appear most exposed; point estimate almost 5 times larger than for tertiary education and 3 times larger than for primary or below.
- By age (Table 5 EV exposure coefficients):
  - All: 0.055*** (0.016)
  - 15-24 year old: 0.013 (0.061)
  - 25-34 year old: 0.146*** (0.032)
  - 35-44 year old: 0.041 (0.040)
  - 45-54 year old: 0.052 (0.040)
  - 55-64 year old: 0.086 (0.068)
- Interpretation: 25-34 year-olds particularly vulnerable; point estimate for 25-34 year-olds is 3 times larger than the average worker.

### VIII. Role of policies: training and active labor market policies (Table 6)
- Augmented specification interacts EV exposure with:
  - country-level spending on training as a share of GDP; and
  - percent of workforce participating in training.
- Selected coefficients (employment and unemployment rate regressions):
  - EV exposure: -0.057*** (0.016), -0.129*** (0.041), -0.185*** (0.070) for Employment Rate regressions; 0.055*** (0.016), 0.129*** (0.040), 0.185*** (0.066) for Unemployment Rate regressions.
  - EV exposure * Training Spending:
    - 0.312** (0.129) in one Employment Rate specification;
    - -0.314** (0.125) in one Unemployment Rate specification.
  - EV exposure * Training Participation:
    - 0.072** (0.032) in one Employment Rate specification;
    - -0.073** (0.031) in one Unemployment Rate specification.
- Interpretation:
  - Interaction coefficients positive and statistically significant for employment specifications, suggesting stronger training policies mitigate negative association between EV exposure and regional employment rates.
- Policy recommendations highlighted:
  - Invest in education and training to increase workforce resilience.
  - Strengthen active labor market policies (training, job search assistance) to facilitate mobility.
  - Consider broader labor market reforms (employment protection legislation, unemployment insurance, labor taxation, minimum wages, collective bargaining reforms) as part of adjustment strategy.
  - Policies to retain or attract EV manufacturers and suppliers (subsidies, industrial plans) to minimize production relocation and job losses.

### IX. Reallocation prospects and green-job opportunities
- Measurement: skill relatedness index (Neffke et al., 2017) rescaled to [-1, 1); share of green jobs computed from U.S. occupation distribution (BLS) and O*NET green-occupation taxonomy.
- Findings:
  - Workers from ICE-focused sectors have skills more related to green-sector needs than the average manufacturing worker, especially for medium green-job intensive destination sectors (20–50 percent green jobs).
  - Sectors with high skill relatedness: manufacturing of industrial gases, manufacturing of electric motors, manufacturing of electrical equipment for vehicles.
- Caveats:
  - Competition for green jobs may intensify as multiple sectors transition.
  - Technological progress (robotization, digitalization, automation) may depress demand for workers, including in green occupations.
  - Analysis uses U.S. occupation data to infer sectoral green-job shares.

### X. Conclusion and caveats
- Transition to EVs affects about 7 percent of the workforce in Europe.
- By 2035, absent policies or technological advances, full EV production implies large reallocation needs: switch jobs, change tasks, retire, emigrate, or face unemployment.
- Some mitigation from normal retirements in aging societies.
- Positive prospects:
  - Skills of ICE-supplying workers comparable to those required in green sectors, implying fair reallocation chances.
  - Evidence that training can ease reallocation and reduce adjustment costs.
- Important caveats:
  - Job implications of charging infrastructure investments not captured.
  - Increasing European battery production will add jobs and value-chain capture not included in these estimates.
  - Future work must disentangle EV transition effects from concurrent trends (automatic gearboxes, decline of diesel, automation of manufacturing and driving).

### XI. Annex I — data sources and variable construction (selected)
- Exports of total cars and EVs: Comext.
- EV CN codes:
  - Prior to 2016 January: CN code 87039010.
  - From 2016 January onwards: CN code 870380.
- Sectoral employment and regional labor market data: Eurostat European Structural Business Statistics dataset.
- Regional controls: OECD Regional Business Demography dataset.
- Share of workers in automotive sector (regional proxy): employment in NACE code 29 as a share of total workers.
- Country-level labor market policy measures: OECD labor market policy database (training spending as share of GDP; stock of participants in training as percent of labor force).
- Sectoral EV exposure categorization follows Kuhlmann et al. (2021): ICE-focused; Non-ICE focused; OEM.

*IMF WORKING PAPERS Cars and the Green Transition: Challenges and Opportunities for European Workers*

### 1. Exports of Electric Vehicles ........................................................................................

### 1. Exports of Electric Vehicles

### Section listings
- 1. Exports of Electric Vehicles ............................................................................................................................... 6
- 2. New Registrations and Sales of Electric Vehicles ............................................................................................. 7
- 3. Electric Vehicle Charging Stations .................................................................................................................... 8
- 4. Countries in Europe with Official  Targets for a 100 Percent Phase Out of ICE Vehicle Sales ........................ 9
- 5. RRF-Funded Spending Dedicated to the Automotive Sector .......................................................................... 10
- 6. Employment of Automotive-Related ................................................................................................................ 12
- 7. Number of Automotive Employees (FTE) by Region, 2015 ............................................................................ 14
- 8. Employment in the Automotive Sector in European Regions, 2015 ............................................................... 15
- 9. Change in Employment Rate, 2015–19 .......................................................................................................... 17
- 10. Prospects of Switching to a Green Job: Germany ........................................................................................ 22

### Tables
- TABLES
  - 1. Labor Market Outcomes and Exposure to Electrification: Sectoral Evidence ................................................. 13
  - 2. Labor Market Outcomes and Exposure to EV Production: Regional Evidence .............................................. 16
  - 3. Scenarios of the Numbers of EU Workers affected by the EV Transition ....................................................... 16
  - 4. Unemployment Rate and Exposure to EV Production: The Role of Workers’ Education ............................... 18
  - 5. Unemployment Rate and Exposure to EV Production: The Role of Workers’ Age ......................................... 19
  - 6. Labor Market Outcomes and Exposure to EV Production: The Role of Policies ............................................ 20

### Annex
- Annex I. Data Sources and Variable Construction ........................................................................................ 24

*Canonical source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023116-print-pdf.pdf*

### References .............................................................................................................

### wpiea2023116-print-pdf - References .............................................................................................................

### I. Introduction — scope and motivation
- Vehicle manufacturing is undergoing its most profound transformation in history driven by rising demand for environmentally friendly vehicles and regulatory restrictions on emissions.
- EVs differ from ICEVs in component composition: EVs need batteries, more wiring, and simpler electric motors; ICEVs require more complex engines, starters, fuel tanks, exhaust pipes and gears.
- EVs will increase electricity demand and reduce demand for gasoline and diesel refining and filling stations.
- Registrations of new EVs in the EU rose from 150 thousand to nearly 2 million from 2015 to 2022.
- More than 14½ million people (or 7 percent of the workforce) depend on the automotive sector directly or indirectly in the EU.
- In Germany: the sector constitutes about 20 percent of manufacturing, 12 percent of employment, and 10 percent of goods exports.
- Policy responses include the EU Chips Act, US Inflation Reduction Act (IRA), and Production Linked Incentives in India; the IRA includes “domestic content” requirements and assembly in the United States.

### II. Transition to EVs — regulatory, demand, and infrastructure trends
- Sales of EVs and hybrids exceeded 45 percent of automotive sales in 2022, from less than 2 percent in 2016.
- European EV exports in 2020 more than quadrupled relative to 2015, amounting to 7 percent of total light-weight passenger car exports.
- Country-level examples: Germany exported the most EVs in 2020; Belgium, Austria, and the Netherlands had EV contributions of more than 10 percent of total car exports in 2020.
- New registrations highlights (2022):
  - Germany: EV and hybrid share almost half of total new car registrations in 2022—five times higher than in 2019.
  - Norway: more than 94 percent of new car registrations in 2022 were EVs and hybrids.
  - Sweden, the Netherlands, Finland, Iceland and Denmark: new EV and hybrid registration shares above 55 percent in 2022.
- EV charging infrastructure: expansion of charging stations outstripped EV registration growth in the 2010s—especially in the Netherlands, France, and Germany; Netherlands has the highest charging station density.
- Full life-cycle GHG emissions of EVs are around half of those of ICEV; IEA (2022) estimates a further 25 percent potential reduction in GHG emissions from EVs with low-carbon electricity.
- EU regulatory targets:
  - Current aim: fleet average of 95 grams CO2 per km for cars and 147 grams CO2 per km for light commercial vehicles (EU Regulation 2019/631).
  - By 2025: cars and vans should emit no more than 81 and 125 grams of CO2 per kilometer, respectively.
  - By 2030: targets drop to 59 and 101 grams of CO2 per kilometer, respectively.
  - Current fine: €95 for each gram per kilometer more than the requirement (example: 10 grams excess × 100,000 vehicles = €95 million fine).
- European Green Deal targets: reduce emissions by 55 percent of 1990 levels by 2030, and 100 percent by 2050.
- Fit for 55 proposal: all new cars and vans sold in the EU from 2035 onward should be zero-emission vehicles; recharging infrastructure spacing rules noted (e.g., electric recharging infrastructures for light-duty vehicles no more than 60 km apart).
- Recovery and Resilience Facility (RRF) contributions:
  - RRF total: €700 bn.
  - EU countries allocated about €71bn under the RRF to sustainable mobility related expenditures.
  - RRF expenditures include tax incentives, purchase financing for zero-emission buses, charging/refueling infrastructure, R&D for hydrogen, production of fuel-cell components, and construction of battery recycling sites.

### III. Empirical strategy — two complementary difference-in-difference exercises
- Time horizon focused on 2015-2019 around the EV take-off; baseline excludes post-COVID-19 period but findings robust to including 2020.
- Two empirical exercises:
  1. Sectoral-level: use sectoral employment data across 26 European countries to test whether employment in ICE-focused sectors grew slower where the share of EV exports in total car exports grew faster over 2015-19.
  2. Regional-level: use sub-regional (NUTS2) employment data to test whether regions more exposed to car manufacturing prior to EV take-off experienced relative employment losses in countries with faster EV export growth.
- Key data notes:
  - Countries in sectoral panel: AUT, BEL, BGR, CZE, DEU, DNK, ESP, EST, FIN, FRA, GBR, GRC, HRV, HUN, IRL, ITA, LTU, LUX, LVA, NLD, POL, PRT, ROU, SVK, SVN, SWE.
  - Sectors included (NACE Rev 2): C2211, C2312, C2620, C2711, C2720, C2740, C2813, C2815, C2825, C2910, C2920, C2931, C2932.
  - Car exports used as proxy for production by fuel type.

### IV. Sectoral evidence — main findings and magnitudes
- Specification (dynamic panel) estimated:
  - ln L_sct = λ ln L_sct-1 + β × (Share_EV_exports_ct × D_s) + γ ln Q_ct + α_sc + ε_sct
- Estimated effects:
  - A one percentage point increase in EV’s share in total car exports is associated with a 1.4 percent decline in employment in ICE-focused sectors relative to other sectors in the short term (β̂ = −0.014, Table 1, column 1).
  - Long-term relative decline: 6.3 percent (β̂/(1−λ̂) = −0.064).
- Median employment growth in ICE-focused sectors was 2.8 percent in 2015.
  - Based on median increase in EV’s share in total car exports from 2015 to 2019, the median country experienced 0.3 percent reduction in employment growth of ICE-focused sectors.
  - A 75th percentile increase in EV’s share would drag growth rate down by 0.9 percent.
- OEM sector employment growth is not significantly hampered by EV expansion (Table 1, column 2).
- Robustness checks performed: additional lags, sector-specific relationships with total car exports, restricting to advanced economies, trimming top/bottom 1 percentile of employment growth.
- Table 1 (selected coefficients and specifications):
  - EV exposure for ICE focused sector: −0.014* (col 1), −0.017* (col 2), −0.016* (col 3), −0.031* (col 4), −0.029* (col 5), −0.024* (col 6), −0.008* (col 7).
  - ln(employee) t-1: 0.780** (col 1), 0.779** (col 2), 0.779** (col 3), 0.887** (col 4), 0.778** (col 5), 0.670** (col 6), 0.838*** (col 7).
  - ln(car exports): 0.103* (col 1), 0.104* (col 2), 0.104* (col 3), 0.150* (col 4), 0.111 (col 5), 0.143** (col 6), 0.072** (col 7).
  - Sample size N shown: 1112 (cols 1–3 & 5), 1032 (col 4), 916 (col 6), 1090 (col 7).
- Interpretation caveats:
  - Sectoral difference-in-difference cannot alone reveal whether displaced ICE-focused workers found jobs in EV-focused sectors or other expanding industries.
  - Cross-border spillovers possible due to international production networks.

### V. Regional evidence — main findings and magnitudes
- Regional specification estimated:
  - L_rct = α_rc + α_ct + β × (Share_EV_exports_ct × Share_Auto_employment_rc,2015) + controls_rct + ε_rct
- Outcomes: male employment rate and male unemployment rate (focus on male workers due to low female share in automotive employment and differing drivers of female labor supply).
- Core results (Table 2):
  - EV exposure coefficient on male employment rate (preferred specification, 2015-2019 with controls, col 4): −0.057*** (standard error 0.016).
  - EV exposure coefficient on male unemployment rate (preferred specification, 2015-2019 with controls, col 8): 0.055*** (standard error 0.016).
  - Other reported specifications:
    - Employment rate columns: −0.023*** (col 1, 2015-2020), −0.046*** (col 2, 2015-2019), −0.056** (col 3, 2015,2019), −0.057*** (col 4, 2015-2019 with controls).
    - Unemployment rate columns: 0.028*** (col 5, 2015-2020), 0.045*** (col 6, 2015-2019), 0.056** (col 7, 2015,2019), 0.055*** (col 8, 2015-2019 with controls).
  - Sample sizes: employment rate regressions N = 1,435; 1,230; 492; 769 (cols 1–4). Unemployment rate regressions N = 1,391; 1,207; 478; 767 (cols 5–8).
- Economic magnitudes and examples:
  - A region at the upper quartile in auto employment (2 percent of workers employed in the automotive sector in 2015) located in a country with median increase in EV share (2 percentage points) experienced 0.2 percentage point smaller rise in employment rate than the median regional increase over the period (2.7 percent per year vs 2.9 percent per year).
  - A region at the 90th percentile in auto employment (4 percent of workers in 2015) saw 0.5 percentage points lower gain in employment rate.
- Aggregate extrapolation:
  - Assuming average region auto employment share of 1.5 percent and β̂ = −0.057 (Table 2, col 4), the growth in EV production between 2015 and 2019 (from 0 to 7 percent of vehicle exports) reduced the employment rate by 0.3 percentage points over this period.
  - This reduction equates to about 600,000 fewer jobs out of a total of 189,700,000 jobs in the EU (Table 3, column 1).
- Forward-looking extrapolation (partial statement in provided text):
  - If EV production expands so that EVs account for 50 percent of vehicle exports, the model (β̂ = −0.057 and average auto share 1.5 percent) suggests employment rates would be about 2 percentage points lower (text fragment ends mid-discussion).

### VI. Role of labor market policies and worker reallocation prospects
- Suggestive evidence that labor market policies can offset adverse effects:
  - The link between potential exposure to EV production and regional employment is significantly smaller in countries with higher spending on workers’ training.
- Detailed evidence for Germany:
  - Given their skills, workers in the automotive sector have a fair chance to reallocate to “green” sectors—those with a high share of green occupations likely to expand during the green transition.
- Policy implications summarized from findings:
  - Strong active labor market policies, particularly those focused on workers’ training, could ease the transition and speed adjustment.
  - Policies to retain or attract EV manufacturers and suppliers (e.g., subsidies, industrial plans) are already being used or considered to minimize production relocation and job losses.

### VII. Contributions and relation to existing literature
- This paper provides the first empirical estimates of the link between EV production and employment in Europe.
- It complements existing bottom-up and top-down studies:
  - Bottom-up example: Bauer et al. (2018) estimate between 11 and 35 percent (23,000-97,000) fewer workers needed in 2030 depending on EV share in yearly sales (25-80 percent).
  - Küpper et al. (2020) find EVs require about the same labor effort as ICEVs; labor-hours would be 7 percent lower if battery cell production/assembly is outsourced.
  - Top-down examples: Barrett et al. (2021) estimate 75,000 job losses in the US under one scenario; Mönnig et al. (2019) estimate net job losses of 114,000 by 2035 in Germany under certain scenarios; European Climate Foundation (2017) predicts 145,000 more jobs in Germany by 2030 under alternative assumptions.
  - Other estimates cited: CLEPA (2021) find 275,000 jobs lost in Europe if tailpipe CO2 emissions are 100 percent lower in 2035 and batteries locally produced; Falck et al. (2021) find 613,000 jobs would be affected by the EV transition (excluding jobs needed for EV production or infrastructure expansion).
- Links to literature on technological change and labor markets: Graetz and Michaels (2015); Gregory et al. (2016); Acemoglu and Restrepo (2017); Autor and Salomons (2017, 2018); Edin et al. (2019); Grigoli, Kóczán, and Topalova (2020).
- Links to literature on real economic costs of the green transition and climate mitigation (IMF 2022a; IMF, 2022b; Bluedorn et al., 2022).

_Italic line: IMF WORKING PAPERS Cars and the Green Transition: Challenges and Opportunities for European Workers_

### 3.8 million fewer jobs in the EU (Table 3, column 1). This point could be reached around 2027/2028, according

### 3.8 million fewer jobs in the EU (Table 3, column 1). This point could be reached around 2027/2028, according

### Key findings
- The shift to electric vehicle (EV) production is already associated with smaller employment gains in ICE-supplying sectors and in vehicle-producing locations relative to other sectors and locations between 2015 and 2020.
- A one percentage point increase in the share of EVs in total car exports is associated with a one percent relative decline in employment in ICE-focused sectors.
- In the country with the median expansion of the share of EVs in total car exports:
  - employment grew by 0.3 percentage points a year less in ICE-focused sectors than in other sectors (median increase in employment in ICE-focused sectors in the sample was 2.8 percent in 2015).
  - the employment rate grew by 2.9 percentage points between 2015 and 2019 nationally, but only 2.7 percentage points in an auto-focused region (a region with 2 percent of its workers employed in the automotive industry).

### Scenarios and aggregate estimates (Table 3)
- EV share of exports goes from 0 to 7 percent (in-sample):
  - Employment rate (ppts): -0.3
  - Implied no. of workers (millions): 0.6
  - Upper 95% C.I. (millions): 1.0
  - Lower 95% C.I. (millions): 0.3
- EV share of exports goes from 7 to 50 percent (out-of-sample extrapolation):
  - Employment rate (ppts): -2.0
  - Implied no. of workers (millions): 3.8
  - Upper 95% C.I. (millions): 5.9
  - Lower 95% C.I. (millions): 1.7
- Estimation uncertainty: the estimate of 3.8 million lies within a 95 percent confidence interval of between 1.7 and 5.9 million jobs.
- Projections of reaching the scenario of 3.8 million fewer jobs could occur around 2027/2028 according to projections of EV sales from the European Automobile Manufacturers’ Association.
- Note on extrapolation: estimates are highly uncertain because they are based on extrapolating the share out to 50 percent when available data cover only up to 7 percent.

### Distributional impacts by education
- Exposure to EV exports and change in employment rates is most pronounced for workers with secondary education:
  - Point estimate on EV exposure is almost 5 times larger in magnitude for workers with secondary education compared to workers with tertiary education, and 3 times larger than for workers with primary education or below.
- Table 4 coefficients (EV exposure):
  - All: 0.055*** (standard error 0.016)
  - Primary Education or Below: 0.032 (0.046)
  - Secondary Education: 0.066*** (0.018)
  - Tertiary Education: 0.014 (0.015)
- Interpretation: middle-skill jobs appear more exposed, consistent with potential hollowing-out of the middle of the income/skill distribution and opportunities for automation when production shifts to EVs.

### Distributional impacts by age
- Younger workers (25-34 year-old) are particularly vulnerable:
  - EV exposure coefficients by age (Table 5):
    - All: 0.055*** (0.016)
    - 15-24 year old: 0.013 (0.061)
    - 25-34 year old: 0.146*** (0.032)
    - 35-44 year old: 0.041 (0.040)
    - 45-54 year old: 0.052 (0.040)
    - 55-64 year old: 0.086 (0.068)
  - The point estimate on exposure for 25-34 year-olds is 3 times larger than the point estimate for the average worker.
- Possible reasons: lower tenure, lower accumulated job-specific human capital, lower likelihood of being in a unionized job, and/or lower likelihood of having a permanent contract.

### The role of policies: training and active labor market policies
- Empirical specification augments baseline with interaction between EV exposure and:
  - country-level spending on training as a share of GDP; and
  - percent of the workforce participating in training.
- Table 6 results (employment and unemployment rate regressions):
  - EV exposure coefficients (various columns):
    - -0.057*** (0.016), -0.129*** (0.041), -0.185*** (0.070) for Employment Rate regressions; 0.055*** (0.016), 0.129*** (0.040), 0.185*** (0.066) for Unemployment Rate regressions.
  - EV exposure * Training Spending:
    - 0.312** (0.129) in one Employment Rate specification;
    - -0.314** (0.125) in one Unemployment Rate specification.
  - EV exposure * Training Participation:
    - 0.072** (0.032) in one Employment Rate specification;
    - -0.073** (0.031) in one Unemployment Rate specification.
- Interpretation:
  - The coefficients (γ) on interactions are positive and statistically significant for employment specifications, suggesting stronger training policies can mitigate the negative association between EV exposure and regional employment rates.
  - Training (both spending and participation) can play an important role in minimizing short-run adverse labor market consequences.
- Policy recommendations highlighted:
  - Further investment in education and training to increase workforce resilience.
  - Active labor market policies like job search assistance to facilitate worker mobility.
  - Broader labor market reforms including employment protection legislation, unemployment insurance, labor taxation, minimum wages and collective bargaining reforms as part of adjustment strategies.

### Reallocation prospects and green-job opportunities
- Measurement approach:
  - Skill relatedness based on Neffke et al. (2017) using German Employee History database; index rescaled to [-1, 1) where larger values indicate higher skill relatedness.
  - Share of green jobs within a sector computed from U.S. sectoral occupation distribution (BLS) and O*NET green-occupation taxonomy.
- Findings on reallocation prospects:
  - Workers from ICE-focused sectors have skills more related to those needed in green sectors than the average manufacturing worker, particularly for medium green-job intensive destination sectors (share of green jobs between 20 to 50 percent).
  - Sectors with high skill relatedness to ICE-focused sectors include manufacturing of industrial gases, manufacturing of electric motors, and manufacturing of electrical equipment for vehicles — sectors that are relatively green-job intensive.
- Caveats and challenges:
  - The green transition affects many sectors, increasing competition for green jobs (e.g., downsizing in brown energy sectors pushes more workers toward green jobs).
  - Technological progress such as robotization, digitalization and automation may depress demand for workers, including in green occupations.
  - The analysis uses U.S. occupation data to infer sectoral green-job shares due to data availability.

### Conclusion and caveats
- The transition to EVs affects about 7 percent of the workforce in Europe (workforce dependent on vehicle manufacturing).
- By 2035, absent policies or technological advances, full EV production in Europe implies a large share of workers will need to switch jobs, change tasks, retire, emigrate, or face unemployment.
- Some mitigation exists via normal retirements in aging societies.
- Positive prospects:
  - Skills of ICE-supplying sector workers are not vastly different from skills required in green sectors, implying comparable chances to reallocate as the average manufacturing worker.
  - Evidence that training can ease reallocation and reduce adjustment costs.
- Important caveats:
  - Job implications of investments in charging infrastructure are not captured here.
  - Increasing European battery production for EVs will add jobs and value-chain capture not included in these estimates.
  - Future empirical work must disentangle EV transition effects from concurrent trends such as shift from manual to automatic gearboxes, decline of diesel, and automation of manufacturing and driving.

*Italic: IMF Working Papers — Cars and the Green Transition: Challenges and Opportunities for European Workers (excerpts provided).*

### Annex I. Data Sources and Variable Construction

### Annex I. Data Sources and Variable Construction

### Key variables and trade data
- Exports of total cars and EVs: from Comext.
- EV classification changes:
  - Prior to 2016 January, EVs were registered under CN code 87039010.
  - From 2016 January onwards, EVs have been registered under CN code 870380.

### Employment and regional labor market data
- Sectoral employment and employment/unemployment rates at regional level: Eurostat European Structural Business Statistics dataset.
- Regional controls (total population, population by education categories, gender, workers by broad industrial structure): OECD Regional Business Demography dataset.
- Share of workers in the automotive sector (regional proxy): employment in NACE code 29 as a share of total workers.
- EV exposure at the regional level:
  - Constructed as the share of EV exports in total car exports at the country level, and
  - The share of automotive workers as a share of total employment in the region in 2015.

### Country-level labor market policy measures
- Source: OECD labor market policy database.
- Measures used:
  - Spending on training programs as a share of GDP.
  - Stock of participants in training programs as percent of the labor force.

### EV exposure at sectoral level
- Sectoral EV exposure categorization follows Kuhlmann et al. (2021):
  - ICE-focused
  - Non-ICE focused
  - OEM

### Notes on variable construction and proxies
- The key EV exposure regional measure relies on country-level export shares combined with 2015 regional employment shares in the automotive sector.
- Automotive sector employment is proxied using NACE code 29 employment as a share of total regional workers.

*Working Paper No. WP/2023/116 — Annex I. Data Sources and Variable Construction*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023116-print-pdf.pdf_