## Transitioning to a Greener Labor Market: Cross-Country Evidence from Microdata

## Source details

**Canonical URL:** [Transitioning to a Greener Labor Market: Cross-Country Evidence from Microdata](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022146-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022146-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022146-print-pdf.pdf.json)

---

### Introduction — Scope, objectives, and core questions
- Investigates environmental properties of jobs, worker mobility into greener employment, and how policies can help green the labor market.
- Creates a new cross-country, harmonized set of indicators of the environmental properties of jobs for a sample of 34 countries (mainly the United States and advanced economies in Europe), covering 2005–19.
- Examines environmental properties both by occupation and by sector, considering three dimensions: green intensity, pollution intensity, and emissions intensity.
- Core research questions:
  - How green is the labor market? Incidence and variation across economies, sectors, demographic characteristics, and earnings.
  - How easily do workers transition into greener jobs? Worker characteristics and employment histories associated with greener employment.
  - How are environmental policies related to the reallocation of workers into greener jobs? How labor market policies and structural features affect policy effectiveness.

### Introduction — Data, definitions, and main high-level findings
- Sample: 34 countries, 2005–19.
- Sector emissions data: IMF Climate Change Indicators Dashboard (December 2021 vintage), available for the years 2005–15, based on gross CO2 emissions.
- Occupational task classification: Dierdorff and others (2009); O*NET (2010, 2021); cross‑walked to ISCO-08.
- Definitions:
  - Green intensity (occupation-based): employment-weighted share of green tasks in total tasks for an occupation, expressed as a percent (scaled 0–100).
  - Pollution intensity (occupation-based): employment-weighted share of polluting activities in an occupation, expressed as a percent (scaled 0–100). Polluting occupations identified via Vona and others (2018) methodology (top 5 percent emissions per worker across substances; occupations with share in these sub-sectors at least 7 times larger than across all occupations).
  - Emissions intensity (sector-based): carbon emissions (in CO2 tons) per worker for a sector, covering direct and indirect emissions based on input-output demand.
- High-level findings:
  - Green- and pollution-intensive jobs are concentrated among a subset of the workforce; average green and pollution intensities are low; many jobs are neutral.
  - Higher-skilled and urban workers tend to hold more green-intensive occupations.
  - Green-intensive occupations exhibit an average earnings premium of almost 7 percent compared with pollution-intensive occupations.
  - Environmental properties of jobs are sticky in transitions; moving from pollution-intensive or neutral jobs into greener work is comparatively difficult.
  - Higher skills facilitate transitions into more green-intensive work.
  - More stringent environmental policies are associated with employment that is more green- and less pollution-intensive.
  - Labor market policies and structural features can hinder or enhance environmental policy effectiveness; reducing job retention support may help incentivize reallocation during recovery phases.
- Caveats:
  - Green and pollution intensities assigned to occupations are invariant over time in the analysis.
  - Sample largely advanced economies; results less applicable to many emerging market or developing economies with large informal employment.
  - Policy-related empirical results are associational rather than causal.

### Evolution and prevalence (key statistics from Introduction)
- Average employment-weighted green intensity of occupations: ranges from around 2 to 3 percent for most economies in the sample.
- Average employment-weighted pollution intensity of occupations: goes from about 2 to 6 percent across economies in the sample.
- Emissions intensity varies by sector, year, and country; median individual-level emissions intensity for the average country within the sample stood at about 8 CO2 tons per worker in 2015.
- Average share of employment in higher-emissions-intensive sectors (mining, manufacturing, utilities): from around 18 percent in 2005 to 15 percent in 2015.
- Employment distribution of emissions intensity is substantially right skewed.

### Job transitions and policy implications (from Introduction)
- Environmental properties of jobs tend to be sticky in transitions; probability of moving into greener work from pollution-intensive work is comparatively low.
- Higher skills increase likelihood of transitioning into more green-intensive work; human capital accumulation could improve greener employment prospects.
- Policy implication: realign labor market policies to avoid inhibiting incentives for reallocation triggered by environmental policies; reduce job retention support as recovery proceeds, in line with country-specific circumstances.

### Conclusion summary (from Introduction excerpt)
- The three measures (green, pollution, emissions intensity) are distinct but sensibly associated with each other: green intensity negatively related to pollution intensity; pollution intensity positively related to sector emissions intensity.

---

### 1. Green Intensity — Trends, heterogeneity, and worker characteristics
- Trends:
  - Average green intensity change and fall in average pollution intensity over the last decade have been only very incremental.
  - Emissions intensity of employment has fallen noticeably over the same period for sample economies, partly reflecting labor reallocation toward lower-emissions-intensive sectors as services expand.
- Sectoral and within-sector heterogeneity:
  - Mining, manufacturing, and utilities tend to have occupations with substantially higher pollution intensity scores; construction follows.
  - Within-sector examples:
    - Mining: mining plant operator pollution intensity = 1; accountant in same sector pollution intensity = 0.
    - Manufacturing: steel factory worker high pollution intensity; aerospace firm engineer lower.
    - Water supply: water resource specialist high green intensity; motor vehicle driver low green intensity.
  - Sector-level emissions intensity treats all workers equally; mining, manufacturing, and utilities highest in emissions intensity, followed by transportation, construction, and real estate (analysis uses log scale).
  - Whiskers in sectoral emissions intensity reflect cross-country differences in energy-use efficiency, country energy mix, and production technologies.
- Implications of within-sector variation:
  - Large within-sector variation implies scope for greening both via labor reallocation across sectors and greening within sectors (occupation switching or task reorientation).
- Environmental properties by worker/job characteristics:
  - Skill and geography:
    - Higher-skilled workers (post secondary or tertiary education) have higher green and lower pollution and emissions intensities relative to lower-skilled workers.
    - Urban workers have higher green and lower pollution intensities than rural workers; urban workers’ average emissions intensity is higher.
  - Contract type and hours:
    - Permanent and full-time workers tend to have higher green, pollution and emissions intensities.
  - Job tenure and firm size:
    - Green intensity does not change with years in job; pollution and emissions intensities increase with tenure.
    - Larger firms have higher average green intensity and higher emissions intensity; pollution intensity shows no clear firm-size pattern.
  - Demographics:
    - Youth (15–29) have higher green intensities than prime-age (30–54) and old (55–64); youth tend to have lower emissions intensity but higher pollution intensities occupationally.
    - Female workers have lower green intensities and distinctly lower pollution and emissions intensities.
- Routinizability:
  - Routine jobs have systematically higher green, pollution, and emissions intensities.
  - Relative gap: pollution intensity relative gap ~6 times that for green intensity; emissions intensity shows similar pattern with smaller gap.
- Green skills and reorientation potential:
  - “General green skills” (engineering/technical, operations management, monitoring/surveillance, science) are relatively evenly distributed across sectors and rising marginally since 2015.
  - Dispersion suggests potential for (re)training to repurpose skills toward greener opportunities.

### 1. Green Intensity — Quantitative highlights (sector and distribution)
- Average share of employment in mining, manufacturing, utilities: 18 percent in 2005 → 15 percent in 2015.
- Median individual-level emissions intensity: about 8 CO2 tons per worker in 2015.

---

### 2. General Green Skills Index in 2015 — Earnings, transitions, persistence, and policy associations

- Earnings premium estimates (pooled Mincer-type regression):
  - Green-intensity coefficient: 4.625*** (0.397)
  - Pollution-intensity coefficient: 0.335*** (0.0218)
  - Prime Age (30-54) Dummy: 0.434*** (0.0104)
  - Elderly (55-64) Dummy: 0.462*** (0.0145)
  - Male Dummy: 0.310*** (0.00581)
  - High-Skill Dummy: 0.410*** (0.00729)
  - Urban Dummy: 0.111*** (0.00497)
  - Constant: 8.178*** (0.00921)
  - Observations: 2,243,990
  - R-squared: 0.858
  - Implied Earnings Premium (average green-intensive job vis-à-vis average pollution-intensive job): 0.067197*** (0.006776)
- Time pattern:
  - Figure 8 (not reproduced here) plots evolution of the earnings premium between 2005 and 2018 (90 percent confidence interval).

- Transition definitions and baseline rates:
  - Job-to-job transition (EEcj): employed in two consecutive years and changed job between t and t-1.
  - Out-of-work job transition (ENE): employed in t-2, not employed in t-1, found a job in t.
  - Job separation (EN): employed in t-1, non-employed in t.
  - Baseline constants (Table 3):
    - EEcj constant: 0.0765*** (0.00219)
    - EN constant: 0.0604*** (0.00146)
    - ENE constant: 0.517*** (0.0151)
    - Observations: 1,393,240 (EEcj), 1,690,588 (EN), 19,829 (ENE)

- How past job environmental properties affect transitions (raw coefficients, Table 4):
  - Past Green Int:
    - EEcj: -0.128*** (0.0283)
    - EN: -0.145*** (0.0228)
    - ENE: 0.496 (0.412)
  - Past Pollution Int:
    - EEcj: -0.0142* (0.00723)
    - EN: -0.0352*** (0.00672)
    - ENE: 0.123 (0.0778)
  - Observations: 1,319,907 (EEcj), 1,527,289 (EN), 18,785 (ENE)

- Rescaled coefficients (percent change of baseline transition rates, Table 5):
  - Past Green Int:
    - EEcj: -.0291*** (.00642)
    - EN: -.0417*** (.00656)
    - ENE: .01469 (.0122)
  - Past Pollution Int:
    - EEcj: -.00755* (.00384)
    - EN: -.0234*** (.00447)
    - ENE: .00888 (.0056)
  - Interpretation:
    - Workers previously in green-intensive jobs are 2.9 percent less likely to experience on-the-job switch relative to those previously in neutral jobs (rescaled -.0291***).
    - Workers previously in pollution-intensive jobs are 0.75 percent less likely to experience on-the-job switch relative to neutral jobs (rescaled -.00755*).
    - Neither group shows statistically significant differences in job finding rates (ENE) relative to neutral jobs.

- Persistence of environmental properties in destination jobs (Table 6):
  - On-the-job switchers (EEcj) — Green-int. dependent variable:
    - Past Green Int: 0.477*** (0.00894) — Observations: 101,988; R-squared: 0.302
    - Past Pollution Int: 0.00210 (0.00128)
  - On-the-job switchers (EEcj) — Pollution-int. dependent variable:
    - Past Green Int: 0.0374** (0.0171)
    - Past Pollution Int: 0.480*** (0.0107) — Observations: 101,988; R-squared: 0.311
  - Out-of-work job finders (ENE) — Green-int. dependent variable:
    - Past Green Int: 0.516*** (0.0183) — Observations: 8,807; R-squared: 0.343
    - Past Pollution Int: -0.00217 (0.00285)
  - Out-of-work job finders (ENE) — Pollution-int. dependent variable:
    - Past Green Int: 0.152*** (0.0512)
    - Past Pollution Int: 0.468*** (0.0197) — Observations: 8,807; R-squared: 0.317
  - Interpretation:
    - Environmental properties of origin jobs are persistent: past green intensity strongly predicts destination green intensity (e.g., 0.477*** for EEcj); past pollution intensity strongly predicts destination pollution intensity (e.g., 0.480*** for EEcj).

- Policies and associations with environmental policy stringency (EPSI):
  - Regression framework: controls for country-year fixed effects, individual controls, country-specific structural controls or labor market policy indicators; effects rescaled by 25th→75th percentile difference of EPSI (.9083) and divided by mean of environmental property for percent-change interpretation.
  - Core Table 7 results:
    - Green-int. (Column 1):
      - EPSI: 0.000349** (0.000169)
      - Constant: 0.00692*** (0.000519)
      - Observations: 1,754,565; R-squared: 0.0662
      - Interpretation: EPSI change from 25th→75th percentile (.9083) associated with a 1.83 percent higher green intensity.
    - Pollution-int. (Column 2):
      - EPSI: -0.00170*** (0.000602)
      - Constant: 0.0371*** (0.00183)
      - Observations: 1,754,565; R-squared: 0.0757
      - Interpretation: EPSI change from 25th→75th percentile associated with a 4.22 lower pollution-intensity.
    - Emissions-int. (Column 3, log scale):
      - EPSI: -0.0635*** (0.0202)
      - Constant: 2.402*** (0.0624)
      - Observations: 7,312,510; R-squared: 0.242
      - Interpretation: EPSI change from 25th→75th percentile associated with a 5.76 percent reduction in emissions intensity.
  - Interactions with labor market policies / features (selected):
    - Interaction with job retention policies (Column 4):
      - Retention: 0.000830*** (0.000239)
      - Retention x EPSI: -0.000370*** (0.0000799)
      - Text interpretation: For a country with average retention policy, total impact of EPSI associated with a -1.66 percent lower green intensity among the employed (Column 4).
    - Interaction with labor market institutions (Column 5):
      - RepRate: -0.000563*** (0.000161)
      - RepRate x EPSI: 0.000271*** (0.0000450)
      - ColBar x EPSI: -0.0000606* (0.0000360)
      - Text interpretation: Pollution intensities among employed workers are 22.6 percent lower for a country with the average level of collective bargaining index and 9.2 percent higher for a country with the average level of unemployment insurance spending between countries of 25th and 75th percentile of the EPSI (Column 5).
    - Additional notes:
      - Worker reallocation support (spending on training programs) showed no statistically significant relationship with EPSI effectiveness.
      - Labor market policies/features that reduce incentives for reallocation (higher job retention spending, greater unemployment insurance generosity) tend to dampen EPSI effectiveness.
      - Greater prevalence of coordinated labor market and collective bargaining arrangements is associated with increased EPSI effectiveness in reducing pollution intensity.
  - On-the-job switchers (Table 8) interactions:
    - Past Green Int x EPSI: 0.0570*** (0.0188)
    - Past Pollution Int x EPSI: -0.00744** (0.00293)
    - Interpretation: A country shifting from the 25th→75th percentile of EPSI yields destination jobs for on-the-job switchers with about 4 percent higher average green intensity and about 2 percent lower average emissions intensity.

- Summary empirical takeaways (from Section 2):
  - Implied earnings premium of green- versus pollution-intensive jobs: 0.067197*** (0.006776).
  - Workers from green-intensive or pollution-intensive jobs are less likely to experience on-the-job switches and separations than those from neutral jobs (rescaled: past green intensity -.0291*** on EEcj; past pollution intensity -.00755* on EEcj).
  - Environmental properties are persistent across transitions (past green/pollution intensities ~0.48–0.52 predictive of destination properties).
  - Stronger EPSI associated with higher green intensity (+1.83 percent for 25th→75th EPSI), lower pollution intensity (-4.22 percent), and lower emissions intensity (-5.76 percent).
  - Labor market institutions and policies moderate EPSI effectiveness: job retention spending and generous unemployment insurance reduce effectiveness; coordinated labor market arrangements/collective bargaining enhance it.
  - Analysis is associational; caution due to potential endogeneity, reverse causality, and limits to extrapolation.

---

### Conclusion — Policy recommendations and final insights
- Metrics and dispersion:
  - The paper quantified job-level environmental metrics across three dimensions; economy-wide average green and pollution intensities are relatively low with wide dispersion across and within sectors.
  - Industrial sectors tend to be simultaneously more green-, pollution-, and emissions-intensive than services.
- Worker characteristics and earnings:
  - Green-intensive occupations are more likely to have higher-skilled and urban workers; these occupations command an earnings premium of almost 7 percent relative to pollution-intensive jobs.
- Reallocation challenges and role of human capital:
  - Workers with histories of pollution-intensive or neutral work are less likely to move into greener jobs.
  - Higher skills facilitate matching to green-intensive jobs.
  - Policy recommendation: targeted and effective training programs to boost human capital of lower-skilled workers in pollution-intensive or neutral occupations to improve mobility into green jobs.
- Environmental policy and labor market dynamics:
  - Environmental policies are associated with shifting employment toward greener jobs, but work best where incentives for reallocation are not inhibited.
  - Policy implication: move from job retention measures to measures that support worker reallocation as COVID-19 shifts from pandemic to endemic; greener employment was relatively more resilient during the COVID-19 recession.

*IMF WORKING PAPERS Transitioning to a Greener Labor Market: Cross-Country Evidence from Microdata — Excerpted sections (Introduction; 1. Green Intensity; 2. General Green Skills Index in 2015; Conclusion).*

### Introduction ...........................................................................................................

### Introduction

### Scope and objectives
- Investigates environmental properties of jobs, worker mobility into greener employment, and how policies can help green the labor market.
- Creates a new cross-country, harmonized set of indicators of the environmental properties of jobs for a sample of 34 countries (mainly the United States and advanced economies in Europe), covering 2005–19.
- Examines environmental properties both by occupation (what workers do) and by sector (where workers work), considering three dimensions: green intensity, pollution intensity, and emissions intensity.

### Core research questions
- How green is the labor market? Incidence and variation of environmental properties across economies, sectors, demographic characteristics, and earnings.
- How easily do workers transition into greener jobs? Worker characteristics and employment histories associated with greener employment.
- How are environmental policies related to the reallocation of workers into greener jobs? How labor market policies and structural features affect policy effectiveness.

### Key empirical sample and data
- Sample: 34 countries, 2005–19.
- Sector emissions data source: IMF Climate Change Indicators Dashboard (December 2021 vintage), available for the years 2005–15, based on gross CO2 emissions.
- Occupational task classification sources: Dierdorff and others (2009); Occupational Information Network (O*NET) (2010, 2021); cross-walked to ISCO-08.

### Main findings (high-level)
- Green- and pollution-intensive jobs are concentrated among a subset of the workforce, producing low average green and pollution intensities and a large share of neutral jobs.
- Higher-skilled and urban workers tend to hold more green-intensive occupations.
- Green-intensive occupations exhibit an average earnings premium of almost 7 percent compared with pollution-intensive occupations.
- Environmental properties of jobs are sticky in transitions; moving from pollution-intensive or neutral jobs into greener work is comparatively difficult.
- Higher skills facilitate transitions into more green-intensive work.
- More stringent environmental policies are associated with employment that is more green- and less pollution-intensive.
- Labor market policies and structural features can hinder or enhance the effectiveness of environmental policies; reducing job retention support may help incentivize reallocation during recovery phases.

---

### Definition of environmental properties of jobs
- Green intensity (occupation-based): the employment-weighted share of green tasks in total tasks for an occupation, expressed as a percent (scaled 0–100).
  - Green tasks: tasks directly related to improving environmental sustainability and reducing greenhouse gas emissions (example: “install vapor barriers or layers of insulation” for roofers).
- Pollution intensity (occupation-based): the employment-weighted share of polluting activities in an occupation, expressed as a percent (scaled 0–100).
  - Polluting occupations (Vona and others 2018 methodology): identified in two steps — (i) polluting sub-sectors are those where emissions per worker of at least three substances are in the top 5 percent; (ii) polluting occupations are those where the share of employees in these polluting sub-sectors is at least 7 times larger than the share across all occupations.
  - Note: because of crosswalk and employment weights, pollution intensity and green intensity under ISCO-08 may both be positive in some cases.
- Emissions intensity (sector-based): carbon emissions (in CO2 tons) per worker for a sector, covering both direct emissions and indirect emissions based on input-output derived demand.

---

### Relationships among the three environmental properties
- Green intensity and pollution intensity show a negative relationship within the employed-worker sample (more green-intensive occupations tend to be less polluting).
- Pollution intensity is positively related to emissions intensity of the sector (pollution-intensive occupations are more likely to be in sectors with higher CO2 tons per worker).
- The three measures are distinct but sensibly associated with each other.

---

### Evolution and prevalence of environmental properties (key statistics)
- Average employment-weighted green intensity of occupations: ranges from around 2 to 3 percent for most economies in the sample.
- Average employment-weighted pollution intensity of occupations: goes from about 2 to 6 percent across economies in the sample.
- Most occupations have very low green and/or pollution intensities; the bulk of jobs are neutral (green and pollution intensity scores of zero).
- Emissions intensity varies by sector, year, and country; carbon emissions per worker are presented on a log scale in underlying analysis.

---

### Job transitions and mobility
- Environmental properties of jobs tend to be sticky in transitions: the probability of moving into greener work from pollution-intensive work is comparatively low.
- Moving into greener occupations from neutral occupations is not significantly easier than from pollution-intensive occupations, reflecting the difficulty of changing occupations.
- Higher skills increase the likelihood of transitioning into more green-intensive work, implying human capital accumulation could improve prospects for greener employment.

---

### Policy findings and implications
- More stringent environmental policies are associated with employment that is more green- and less pollution-intensive, contributing to a greener labor market.
- Labor market policies and structural features can affect the reallocation incentives needed for greener employment:
  - Policymakers should consider realigning labor market policies to avoid inhibiting incentives for reallocation triggered by environmental policies.
  - With a strong recovery from the COVID-19 pandemic recession underway, it will be important to reduce job retention support measures to help incentivize reallocation (in line with country-specific circumstances).

---

### Caveats and limitations
- Green and pollution intensities assigned to occupations are invariant over time in the analysis; technological changes could alter intensities within occupations without occupational reallocation.
- Sample composition: largely advanced economies, making results less applicable to many emerging market or developing economies, particularly those with large shares of informal employment.
- Policy-related empirical results are associational rather than causal due to potential omitted variables and reliance on historical patterns, which may not represent the size and mix of policy changes needed to achieve net zero emissions.

---

*IMF Working Paper: "Transitioning to a Greener Labor Market: Cross-Country Evidence from Microdata" — Introduction section (excerpt).*

### 1. Green Intensity

### 1. Green Intensity

### Trends in Green, Pollution, and Emissions Intensities
- The change in average green intensity and the fall in average pollution intensity over the last decade have been only very incremental.
- Emissions intensity of employment has fallen noticeably over the same period for the economies in the sample, partly reflecting labor reallocation toward lower-emissions-intensive sectors as the services sector expands.
- The average share of employment in the higher-emissions-intensive sectors of mining, manufacturing, and utilities went from around 18 percent in 2005 to 15 percent in 2015.
- The median individual-level emissions intensity for the average country within the sample stood at about 8 CO2 tons per worker in 2015.
- The employment distribution of emissions intensity is substantially right skewed, indicating only a small share of workers are involved in activities generating high carbon emissions.

### Sectoral Heterogeneity in Environmental Properties of Employment
- Mining, manufacturing, and utilities tend to have occupations with substantially higher pollution intensity scores, followed by the construction sector.
- Within-sector heterogeneity examples:
  - Mining: a mining plant operator has a pollution intensity score equal to 1, while an accountant in the same sector has a pollution intensity score of 0.
  - Manufacturing: a steel factory worker has a high pollution intensity score, while an aerospace firm engineer has a lower score.
  - Water supply sector: a water resource specialist has a high green intensity, while a motor vehicle driver in the same sector has a low green intensity.
- Sector-level emissions intensity treats all workers in a sector equally and exhibits significant within- and across-sector heterogeneity; mining, manufacturing and utilities are highest in emissions intensity, followed by transportation, construction and real estate (chart depicted on a logarithmic scale).
- Whiskers in sectoral emissions intensity reflect cross-country differences in energy-use efficiency, country energy mix, and production technologies.

### Implications of Within-Sector Variation
- Large within-sector variation in green, pollution, and emissions intensities suggests substantial room for further greening within each sector.
- Two margins for adjustment during the green transition are highlighted:
  - Labor reallocation across sectors.
  - Further greening within sectors (occupation switching or task reorientation).

### Environmental Properties by Worker and Job Characteristics
- Skill and geography:
  - Higher-skilled workers (post secondary or tertiary education) tend to have higher green and lower pollution and emissions intensities relative to lower-skilled workers (at most secondary and nontertiary education or below).
  - Urban workers tend to have higher green and lower pollution intensities than rural workers; however, urban workers’ average emissions intensity is higher.
- Contract type and hours:
  - Workers with permanent contracts and those in full-time employment tend to have higher green, pollution and emissions intensities.
- Job tenure and firm size:
  - Green intensity does not seem to change with years in the job, while pollution and emissions intensities increase with job tenure.
  - Larger firms have occupations with higher green intensity on average; emissions intensity of employment is distinctly higher in larger firms; pollution intensity shows no discernible difference across firm sizes.
- Demographics:
  - Youth (15–29 years old) have higher green intensities than prime-age (30–54 years old) and old (55–64 years old) workers, and they tend to exhibit lower emissions intensity but higher pollution intensities from an occupational perspective.
  - Female workers have lower green intensities coupled with distinctly lower pollution and emissions intensities, reflecting lower shares of employment in highly polluting sectors such as mining, manufacturing and utilities.

### Routinizability and Environmental Properties
- Jobs more vulnerable to automation (routine jobs) have systematically higher green, pollution, and emissions intensities on average.
- The relative gap (intensity for routine occupations divided by that for non-routine) varies by environmental property:
  - The relative gap for pollution intensity is about 6 times the size of that for green intensity.
  - Emissions intensity shows a similar pattern but with a smaller gap.
- This indicates that jobs more vulnerable to automation are more likely to have higher pollution and emissions intensity.

### Green Skills and Potential for Reorientation
- “General green skills” (associated with engineering and technical skills, operations management, monitoring/surveillance, and science) are relatively evenly distributed across sectors and have been rising marginally since 2015.
- The wide dispersion within sectors and similar levels across sectors suggest potential for further greening of the economy.
- The prevalence of general green skills among workers suggests that appropriate (re)training could help workers repurpose and reorient their skills toward greener job opportunities.

*Source: IMF Working Papers — Transitioning to a Greener Labor Market: Cross-Country Evidence from Microdata (content unit: 1. Green Intensity).*

### 2. General Green Skills Index in 2015

### 2. General Green Skills Index in 2015

### Earnings premium of green-intensive versus pollution-intensive jobs
- Regression specification (Mincer-type): Y_i,s,c,t = α + α_ct + β GreenInt_i,s,c,t + γ PolInt_i,s,c,t + θ′ X_i,s,c,t + ε_i,s,c,t, where Y is log earnings (real in US 2015 dollars); X contains age (youth, prime, old), educational attainment (low/high), gender (female/male), location (urban/rural); baseline: young, female, low educational attainment, rural, in a neutral job. Country-year fixed effects included; standard errors clustered at country-year.
- Table 2 estimated coefficients:
  - Green-intensity: 4.625***
    - (0.397)
  - Pollution-intensity: 0.335***
    - (0.0218)
  - Prime Age (30-54 yrs old) Dummy: 0.434***
    - (0.0104)
  - Elderly (55-64 yrs old) Dummy: 0.462***
    - (0.0145)
  - Male Dummy: 0.310***
    - (0.00581)
  - High-Skill Dummy: 0.410***
    - (0.00729)
  - Urban Dummy: 0.111***
    - (0.00497)
  - Constant: 8.178***
    - (0.00921)
  - Observations: 2,243,990
  - R-squared: 0.858
  - Adjusted R-squared: 0.858
  - Implied Earnings Premium (average green-intensive job vis-à-vis average pollution-intensive job): 0.067197***
    - (0.006776)
- Implied interpretation:
  - The estimated average earnings premium of green- versus pollution-intensive jobs is 6.7197 log points (reported as 0.067197), after controlling for individual characteristics and country-year fixed effects.
- Time pattern:
  - Figure 8 plots the evolution of the earnings premium between 2005 and 2018 (with 90 percent confidence interval).

### Transition across environmental properties of jobs — definitions and baseline rates
- Transition dummies Z_i,s,c,t:
  - Job-to-job transition (EEcj): employed in two consecutive years and changed job between t and t-1.
  - Out-of-work job transition (ENE): employed in t-2, not employed in t-1, found a job in t.
  - Job separation (EN): employed in t-1, non-employed in t.
- Baseline (unconditional constant-only regressions; Table 3):
  - Job-to-job transition (EEcj) constant: 0.0765***
    - (0.00219)
  - Job separation (EN) constant: 0.0604***
    - (0.00146)
  - Job finding (ENE) constant: 0.517***
    - (0.0151)
  - Observations: 1,393,240 (EEcj), 1,690,588 (EN), 19,829 (ENE)
  - R-squared equals each constant (0.0765, 0.0604, 0.517 respectively).

### How past job environmental properties affect transitions (raw and rescaled results)
- Regression: Z_i,s,c,t = α + α_c + α_t + θ′ X_i,s,c,t + γ′ Y_i,s,c,t−k + ε_i,s,c,t (Y contains lagged job environmental properties; k=1 for on-the-job and job separation, k=2 for out-of-work job finding).
- Raw coefficients (Table 4; country and year fixed effects, individual controls included):
  - Past Green Int:
    - EEcj: -0.128*** (0.0283)
    - EN: -0.145*** (0.0228)
    - ENE: 0.496 (0.412)
  - Past Pollution Int:
    - EEcj: -0.0142* (0.00723)
    - EN: -0.0352*** (0.00672)
    - ENE: 0.123 (0.0778)
  - Observations: 1,319,907 (EEcj), 1,527,289 (EN), 18,785 (ENE)
- Rescaled coefficients expressed as percent change of baseline transition rates (Table 5):
  - Past Green Int:
    - EEcj: -.0291*** (.00642)
    - EN: -.0417*** (.00656)
    - ENE: .01469 (.0122)
  - Past Pollution Int:
    - EEcj: -.00755* (.00384)
    - EN: -.0234*** (.00447)
    - ENE: .00888 (.0056)
- Interpretation (from text):
  - Workers who previously held green-intensive jobs are 2.9 percent less likely to experience on-the-job switch relative to those who previously held neutral jobs (rescaled coefficient -.0291***).
  - Workers who previously held pollution-intensive jobs are 0.75 percent less likely to experience on-the-job switch relative to those who previously held neutral jobs (rescaled coefficient -.00755*).
  - Neither group was statistically significantly different in job finding rates (ENE) relative to neutral jobs.

### Persistence of environmental properties of jobs and implications for transitions
- Regression for destination job property V (restricted to transitions): V_i,s,c,t = α + α_c + α_t + θ′ X_i,s,c,t + γ′ Y_i,s,c,t−1 + ε_i,s,c,t, estimated on samples of on-the-job switchers and out-of-work job finders (restricting to Z=1).
- Persistence estimates (Table 6):
  - On-the-job switchers (EEcj) — Green-int. dependent variable:
    - Past Green Int: 0.477***
      - (0.00894)
    - Past Pollution Int: 0.00210
      - (0.00128)
    - Observations: 101,988; R-squared: 0.302
  - On-the-job switchers (EEcj) — Pollution-int. dependent variable:
    - Past Green Int: 0.0374**
      - (0.0171)
    - Past Pollution Int: 0.480***
      - (0.0107)
    - Observations: 101,988; R-squared: 0.311
  - Out-of-work job finders (ENE) — Green-int. dependent variable:
    - Past Green Int: 0.516***
      - (0.0183)
    - Past Pollution Int: -0.00217
      - (0.00285)
    - Observations: 8,807; R-squared: 0.343
  - Out-of-work job finders (ENE) — Pollution-int. dependent variable:
    - Past Green Int: 0.152***
      - (0.0512)
    - Past Pollution Int: 0.468***
      - (0.0197)
    - Observations: 8,807; R-squared: 0.317
- Interpretation:
  - Environmental properties of origin jobs are persistent: past green intensity strongly predicts destination green intensity (e.g., 0.477*** for EEcj), and past pollution intensity strongly predicts destination pollution intensity (e.g., 0.480*** for EEcj).

### Policies and the green transition — associations between environmental policy stringency and job environmental properties
- Regression: V_i,s,c,t = α_c + α_t + β′ X_i,s,c,t + θ′ W_c,t + δ P_c,t + μ′ (W_c,t ∙ P_c,t) + ε_i,s,c,t, where P_c,t is EPSI (OECD environmental policy stringency index), W_c,t are country-specific structural controls or labor market policy indicators; coefficients rescaled by difference between 25th and 75th percentile of EPSI (.9083) and divided by mean of environmental property to express percent changes.
- Core Table 7 coefficients (columns reported):
  - Column (1) Green-int.:
    - EPSI: 0.000349** (0.000169)
    - Constant: 0.00692*** (0.000519)
    - Observations: 1,754,565; R-squared: 0.0662
    - Interpretation: An EPSI change from 25th to 75th percentile (.9083) is associated with a 1.83 percent higher green intensity (raw coefficient .000349).
  - Column (2) Pollution-int.:
    - EPSI: -0.00170*** (0.000602)
    - Constant: 0.0371*** (0.00183)
    - Observations: 1,754,565; R-squared: 0.0757
    - Interpretation: An EPSI change from 25th to 75th percentile is associated with a 4.22 lower pollution-intensity (raw coefficient -0.00170).
  - Column (3) Emissions-int.:
    - EPSI: -0.0635*** (0.0202)
    - Constant: 2.402*** (0.0624)
    - Observations: 7,312,510; R-squared: 0.242
    - Interpretation: An EPSI change from 25th to 75th percentile is associated with a 5.76 percent reduction in emissions intensity (raw coefficient .0635 noting emissions intensity is measured in logarithm).
- Interactions with labor market policies and structural features (selected findings):
  - Interaction with job retention policies (Column 4):
    - Retention: 0.000830*** (0.000239)
    - Retention x EPSI: -0.000370*** (0.0000799)
    - Text interpretation: For a country with average retention policy, the total impact of EPSI is associated with a -1.66 percent lower green intensity among the employed (Column 4).
  - Interaction with labor market institutions (Column 5):
    - RepRate: -0.000563*** (0.000161)
    - RepRate x EPSI: 0.000271*** (0.0000450)
    - ColBar x EPSI: -0.0000606* (0.0000360)
    - Text interpretation: Pollution intensities among employed workers are 22.6 percent lower for a country with the average level of collective bargaining index and 9.2 percent higher for a country with the average level of unemployment insurance spending between countries of 25th and 75th percentile of the EPSI (Column 5).
  - Additional notes:
    - Worker reallocation support (spending on training programs) showed no statistically significant relationship with the effectiveness of EPSI in greening employment.
    - In general, labor market policies and features associated with reduced incentives for worker reallocation (higher job retention spending, greater unemployment insurance generosity) tend to dampen the effectiveness of environmental policies in greening the labor market.
    - By contrast, greater prevalence of coordinated labor market and collective bargaining arrangements is associated with increased effectiveness of EPSI in reducing pollution intensity of employment.
- On-the-job switchers (Table 8):
  - For on-the-job switchers, interaction results:
    - Past Green Int x EPSI: 0.0570*** (0.0188)
    - Past Pollution Int x EPSI: -0.00744** (0.00293)
  - Interpretation in text:
    - For a country shifting from the 25th to the 75th percentile of EPSI, destination jobs for on-the-job job switchers have about 4 percent higher average green intensity and about 2 percent lower average emissions intensity.

### Summary findings and caveats
- Main empirical takeaways:
  - Green-intensive jobs command an earnings premium relative to pollution-intensive jobs: implied earnings premium 0.067197*** in the pooled sample.
  - Workers from green-intensive or pollution-intensive jobs are less likely to experience on-the-job switches and separations relative to those from neutral jobs (rescaled effects: past green intensity -.0291*** on EEcj; past pollution intensity -.00755* on EEcj).
  - Environmental properties of jobs are persistent across job-to-job and out-of-work transitions (e.g., past green intensity strongly predicts destination green intensity with coefficients ~0.48–0.52).
  - Stronger environmental policy stringency (EPSI) is associated with higher green intensity (+1.83 percent for 25th→75th EPSI), lower pollution intensity (-4.22 percent), and lower emissions intensity (-5.76 percent).
  - The greening impact of environmental policies is moderated by labor market institutions and policies: job retention spending and generous unemployment insurance can reduce the effectiveness of EPSI in promoting green employment, while coordinated labor market arrangements/collective bargaining can enhance it.
- Caution:
  - The analysis is associational; the authors note limitations including potential endogeneity, reverse causality, lack of granularity on alternative policy instruments, and the unprecedented nature of the climate mitigation challenge, arguing for caution in extrapolating results.

*IMF Working Paper — Transitioning to a Greener Labor Market: Cross-Country Evidence from Microdata (excerpt: 2. General Green Skills Index in 2015).*

### Conclusion

### Conclusion

### Quantification and job-level environmental metrics
- The paper quantified the environmental properties of individual workers’ jobs through three different metrics, reflecting how green, polluting, and carbon-emitting each job is.
- Economy-wide average green and pollution intensities are relatively low.
- There is a wide dispersion of these environmental properties across and within sectors, suggesting capacity for labor reallocation along both dimensions.
- Industrial sectors tend to be simultaneously more green-, pollution-, and emissions-intensive than services.

### Worker characteristics and earnings
- More green-intensive occupations tend to have higher-skilled and more urban workers.
- The reverse is true for more pollution-intensive jobs.
- Even after controlling for skills, green-intensive jobs exhibit an earnings premium—almost 7 percent—compared with pollution-intensive jobs on average.

### Reallocation challenges and human capital
- A worker with a history of more pollution-intensive or neutral work is less likely to move into a more green-intensive job.
- Higher skills facilitate matching to more green-intensive jobs, highlighting the importance of human capital in easing transitions.
- Policy recommendation: targeted and effective training programs to boost the human capital of lower-skilled workers in pollution-intensive or neutral occupations could improve their ability to move into more green-intensive occupations.

### Environmental policy and labor market dynamics
- Environmental policies are effective in shifting employment towards greener jobs, but they work best in economies where incentives for reallocation are not inhibited.
- This underscores the importance of moving from job retention measures to measures that support worker reallocation as COVID-19 shifts from pandemic to endemic.
- Recent labor market dynamics indicate that greener employment was relatively more resilient during the COVID-19 recession.

*IMF WORKING PAPERS Transitioning to a Greener Labor Market: Cross-Country Evidence from Microdata — Conclusion*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022146-print-pdf.pdf_