Policy Research Working Paper 10779 Profiling Green Jobs and Workers in South Africa An Occupational Tasks Approach Jacqueline Mosomi Wendy Cunningham Social Protection and Jobs Global Practice A verified reproducibility package for this paper is May 2024 available at http://reproducibility.worldbank.org, click here for direct access. Policy Research Working Paper 10779 Abstract To adequately prepare the labor force for the green economy, green occupations. Strictly green occupations tend to be policy makers and workers require a detailed understanding male-dominated and held by prime-age (25–44) workers of the nature of green jobs. This study profiles green jobs in with post-secondary school. However, the profile of those the South African labor market. It uses labor force survey in the greenest of the green occupations shows that they are data and applies an occupational task-based approach to older (age 45–65) workers and Black Africans with lower identify current green occupations and associated jobs, than completed high school education. Policies to prepare count them, and profile their workers and wages. The South Africans to engage in the green economy include findings show that 5.5 to 32 percent of South Africa’s jobs developing a strategy to teach new and existing workers can be labeled as “green,” where the former estimate uses a to use green technologies; targeting green occupations in strict definition and the latter uses a broad definition. The youth development programs; making a concerted effort share of strictly green jobs has not changed over the past to support women in science, technology, engineering, and eight years. While 65 percent of strictly green occupations mathematics; helping low-skilled green workers to orga- can be classified as high (skill) occupations, only 55 percent nize and improve their work conditions; and continuing to of workers are in these occupations, reflecting numerous collect and analyze data for better tracking South Africa’s employment opportunities in mid-level and elementary progress in becoming a green labor force. This paper is a product of the Social Protection and Jobs Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at Wcunningham@worldbank.org and jackiemosomi@gmail.com. A verified reproducibility package for this paper is available at http://reproducibility.worldbank.org, click here for direct access. RESEA CY LI R CH PO TRANSPARENT ANALYSIS S W R R E O KI P NG PA The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Profiling Green Jobs and Workers in South Africa: An Occupational Tasks Approach Jacqueline Mosomi1 and Wendy Cunningham2 3 Keywords: green jobs, green occupations, South Africa, task-based approach, labor market structure JEL Classification: J21, 055, Q01, Q50 1 Southern Africa Labour and Development Research Unit (SALDRU), UCT 2 The World Bank 3 We are grateful to Josefina Posadas, Abla Safir, Samantha De Martino, Gibson Mudiriza, Javier E. Baez, German Caruso and Elizabeth Ninan for detailed comments and to Debbie Budlender and the participants at a SALDRU-UCT seminar and the Inequality, Work and Nature (IWN) conference in Cape Town. 1 Introduction Like many parts of the world, South Africa is experiencing deteriorating climate conditions. Rising temperatures, droughts and water shortages have become the new normal (Republic of South Africa 2021; Arndt, Gabriel, Hartley, Strzepek, and Thomas 2021; World Bank Group 2022; Engelbrecht et al. 2015). These weather events are destroying and disrupting communities and livelihoods. 4 At the same time, South Africa’s high reliance on coal for its energy needs may be contributing to the climate crisis. Global pressure to move away from carbon-based energy is a challenge in a labor market where the mining industry directly and indirectly creates thousands of jobs (Maseko 2021; Makgetla and Patel 2021; Makgetla 2021) even as aggregate employment rates have been on the decline and unemployment has been on the rise since 2008 (StatsSA 2022; Maseko 2021; World Bank Group 2018). 5 The South African government is committed to growing its economy and creating jobs through environmentally friendly strategies (Inglesi-Lotz 2021, DEA 2011, NPC 2011). The National Development Plan of 2011 envisions the green economy – which can be characterized as industries and outputs that do not create significant environmental risks 6 – as an avenue for job creation and a pathway to alleviate the challenges of high levels of poverty and inequality by promoting an “environmentally sustainable, climate change resilient, low-carbon and just society” (NPC 2011). The education and skills sectors are preparing to train workers for green jobs, which can be defined as jobs that aim to reduce the human-induced negative impact on the environment and climate change. 7 Research supports the view that the transition to a green economy could be a job-creator in South Africa (World Bank Group 2022; PCC 2021; Merven et al. 2019, Maia et al. 2011, Rutovitz 2010), though these studies do not directly measure green jobs. Instead, they use an “output approach” (Granata and Posadas 2024), where they select industrial sectors that are directly linked to energy, natural resources, or waste pollution and estimate job growth scenarios based on how the industries are expected to grow. The estimates range from 149,000 (Rutovitz 2010) to nearly 500,000 (Maia et al. 2011) new jobs over the next few decades. While these studies give rough orders of magnitudes of the future number of jobs 4 For example, during the severe 2018 water crisis in the Western Cape, two-thirds of tourism businesses reported being adversely affected (World Bank Group 2022). 5 Employment rates in South Africa have been on the decline and aggregate unemployment has been on the rise since the 2007-2008 global financial crisis, exacerbated by the recent Coronavirus (COVID- 19) pandemic. In the first quarter of 2022, the unemployment rate stood at 34.5 percent (StatsSA 2022) for the working age population but was much higher for young people (63.9 percent for those aged 15- 24 years). Of 10.2 million youth aged 15-24 years, 37 percent were not in employment, education nor training (NEET) in the first quarter of 2022 (StatsSA 2022). 6 Specifically, the NPC adopts the UNEP definition of the green economy which is “...a system of economic activities related to the production, distribution and consumption of goods and services that result in improved human well-being over the long term, while not exposing future generations to significant environmental risks and ecological scarcities”. 7 There is no consensus in the literature as to the concise definition of green jobs, making the process of measuring green jobs complex. There is however a convergence on the concept that green jobs are those that aim to reduce the human-induced negative impact on the environment and climate change, as defined by UNEP and the ILO (Stanef-Puică et al. 2022). 2 in climate-related industries, 8 they do not tell us about the nature of green jobs today, wages earned, or who works in those jobs. A recent international literature uses a more systematic approach to identify and characterize green jobs. These studies use an occupational task approach, where they label an occupation – and the jobs within – as green based on how green the tasks within an occupation are, rather than the industry that houses a job (Granata and Posadas 2024; Vona et al. 2019; Consoli et al. 2016; Vona et al. 2018). The occupational tasks approach has been widely applied in labor economics to analyze how technological change has affected skills, wages and job polarization over time (Autor et al. 2003; Acemoglu and Autor 2011; Goos et al. 2014; Cunningham et al. 2022). 9 We can use the same logic to understand how a shift to a green economy may affect jobs. Using the occupational task approach allows for greenness of an occupation to be defined according to the number of tasks relating to environmental sustainability that are typically carried out in that occupation (Vona et al. 2019). The advantage of the occupational task approach is that one can identify green jobs in any industry, even those not directly engaged in environmentally sensitive industries. This paper aims to fill the gap in the South African literature by using South African household survey data and applying an occupational task-based approach to characterize green jobs and workers in South Africa. Inspired by Granata and Posadas (2024), we define two sets of green jobs: strict and broad. Recognizing that occupations are groupings of jobs with similar tasks, we first identify “strictly green jobs,” as those in occupations that require at least one task that is directly linked to the environment. We then broaden our definition to include occupations with job tasks that could reduce human-induced environmental risks, if they used green technology or processes 10 and refer to the larger set of jobs as “broadly green jobs”. Notably, strictly green jobs are a subset of broadly green jobs. We then take stock of occupations that can be classified as strictly and broadly green in the current South African economy and, using regression analysis, profile the related jobs and workers. The paper does not forecast the number of new jobs or occupations that will arise from the transition to a 8 These studies may over-estimate the number of green jobs since not all workers in these sectors are involved in green activities. On the other hand, they underestimate green jobs because they do not consider other sectors where workers might be carrying out tasks that reduce the human-induced negative impact on the environment and climate change. 9 In this literature, it is useful to distinguish between tasks and skills. A task is defined as a unit of work activity that produces output (goods and services). A skill is defined as a worker’s endowment of capabilities for performing various tasks (Acemoglu and Autor 2011). Workers with similar skills can perform different tasks depending on the technology available to them and the production function.10 These jobs could be carried out using green or non-green technologies or processes. For example, workers in agriculture could use traditional methods (non-green technologies) or could practice sustainable farming (green technology). Since we do not have information about the take-up of green technologies within occupations, we assume that those occupations where green technologies or processes could be used do indeed use them. This gives us an upper-bound of broadly green jobs in South Africa. 10 These jobs could be carried out using green or non-green technologies or processes. For example, workers in agriculture could use traditional methods (non-green technologies) or could practice sustainable farming (green technology). Since we do not have information about the take-up of green technologies within occupations, we assume that those occupations where green technologies or processes could be used do indeed use them. This gives us an upper-bound of broadly green jobs in South Africa. 3 green economy. 11 We find that a substantial number of South African jobs and a broad range of workers can be classified as green. We estimate that 5.5 percent of South Africa’s jobs can be strictly defined as green while 32 percent can be defined as broadly green. 12 This maps to 46 strictly green occupations and 137 broadly green occupations, from a list of 428 occupations. The green jobs estimates are within the range of other countries. While 65 percent of strictly green occupations in the South African labor market require highly skilled workers, resulting in a wage premium for those employed in these occupations, they employ only 55 percent of workers in strictly green jobs. Instead, a smaller share of mid-level and elementary occupations are responsible for 45 percent of total (strictly) green employment. While the share of strictly green jobs has barely changed over the past eight years, there was an increase in the share of broadly green jobs between 2013 and 2014. Consistent with international literature, relative to the general labor force, the average workers in strictly green occupations tend to be male, of prime working age (25-44) and have some type of tertiary qualification. This exercise contributes to the literature in several ways. First, it provides a benchmark against which to measure South Africa’s evolution of green jobs on average and among traditionally excluded groups. Second, it offers guidance on how to prepare today’s South Africans for today’s green jobs. Third, it joins a small literature using the occupational task approach (Granata and Posadas 2024; Vona et al. 2019; Consoli et al. 2016; Vona et al. 2018) and illustrates how to utilize the methodology for a country-specific analysis. The rest of this paper is organized as follows. Section 2 describes the data and methodology while Section 3 shares a descriptive analysis. In Section 4 we present and discuss regression results. Section 5 concludes and offers five policy messages for a more inclusive and successful green labor force and economy. 2 Data We use the Post-Apartheid Labour Market Series (PALMS) dataset (Kerr et al. 2019). PALMS is constructed from several South African Labour Force Surveys for the years 1993-2019, 13 curated and harmonized by DataFirst at the University of Cape Town. We extract the Quarterly Labour Force Survey (QLFS) 2012-2019 data from the PALMS dataset for our analysis. The QLFS is a nationally representative household survey collected by the national statistical agency Statistics South Africa (Stats SA) since 2008. It includes data on the labor market activities of individuals aged 15 years and older who live in South Africa. Each survey samples approximately 33,000 dwelling units based on about 3,324 Primary Sampling Units (PSUs). For each quarter of the QLFS, a quarter of the sampled dwellings are rotated out of 11 Examples of studies that make projections about employment due to the transition to a greener South African economy include Rutovitz (2010), Maia et al. (2011) and Merven et al. (2019). 12 As of 2019, there were approximately 16.4 million jobs in South Africa. 13 The PALMS dataset includes the 1993 Project for Statistics on Living Standards and Development (PSLSD), the annual October Household Surveys (1994-1999), the biannual Labour Force Surveys (2000- 2007) and the Quarterly Labour Force Surveys (2008-2019). We only use the QLFS 2012-2019 in this analysis. 4 the sample and replaced by new dwellings from the same PSU or the next PSU on the list. PALMS is well suited for this paper’s analysis. It includes occupational information up to the four-digit level and uses the South Africa Standard Classification of Occupations (SASCO) from 2003 (Statistics South Africa 2003). SASCO 2003 is based on the International Standard Classification of Occupations of 1988 (ISCO-88). 14 To ensure comparability, DataFirst has gone to great lengths to harmonize the PALMS dataset in terms of variable names over time. Additionally, the data set comes with re-calibrated weights using a cross entropy (CE) approach (Branson and Wittenberg 2014) to ensure continuity and comparability between surveys over time. For this study, we restrict our analysis to the period 2012-2019. Data are available for more recent years but to avoid contaminating the results with the effects of the COVID-19 pandemic, we exclude data collected in 2020 or after. We pool data from the four quarters of each year resulting in 32 quarters (waves) of data. 15 The pooling of the data over the 8 years gives a sample size that will allow analysis at the 4-digit SASCO level. Using these data, we profile the labor force working in green occupations by age, gender, location, skill, and education level. 2.1 Methodology to Generate the List of Green Occupations We adopt the occupational task approach and use Dierdorff et al. (2009) to adapt our definition of green jobs to reference occupational tasks. Specifically, we define strictly green jobs as occupations that require tasks that have environmental impacts. We define broadly green jobs as occupations that are strictly green or occupations with tasks that could have environmental impacts if green technology is used. Some of these jobs benefit from the green revolution, such as miners of lithium for electric car batteries. Others are negatively affected, such as coal miners and those working in coal-fired plants. We include both in our analysis. While many studies in various countries use tasks information from the US Occupational Information Network (O*NET) inventory and its associated Green Economy Program (GEP) (Consoli et al. 2016; Vona et al. 2019, 2018) to measure green jobs, we find these data are not appropriate for our analysis for several reasons. The O*NET is a US cross-sectional database with detailed information about work context, tasks, activities and skills at the occupational level. 16 The O*NET Green Economy program (GEP) identifies 12 industries that should be impacted by the greening of the economy and an associated list of occupations and 14 International Standard Classification of Occupations 1988 (ISCO-88) is a four-level hierarchically structured system that allows all jobs in the world to be classified into unit groups based on their similarity in terms of the skill level and skill specialization required for the jobs. The most aggregated level is the one-digit code (10 major groups), followed by the two-digit code (28 sub-major groups), the three-digit code (116 minor groups) and the most detailed level of the classification is the four-digit code (390 groups). The SASCO includes an additional 39 occupations, reflecting uniquely South African labor markets. 15 To ensure there is no double counting of individuals, we divide the survey weights by the number of quarters. 16 The O*NET is the primary source of occupational information in the United States. The data are collected through industry expert interviews and surveys of representative samples of workers. https://www.onetonline.org/ 5 green tasks used therein. 17 This data set, while quite rich, faces three primary shortcomings in our context. First, it may not be applicable to countries outside the US, especially middle- income and developing nations since production methods and technology may differ across economies or the required task profiles to realize those occupations may differ. Second, the GEP data were collected before 2009 while green technologies have evolved rapidly in the past 14 years. Additionally, it only considers jobs in 12 industries. Third, researchers report significant measurement error (Vona et al. 2019; Granata and Posadas 2024) when transforming 8-digit O*NET SOC codes to 4-digit ISCO codes. Other studies have developed a methodology using text analysis of occupational tasks data to identify green jobs. Janser (2018) applied text analysis to a German occupational tasks database (BERUFENET) to identify green jobs in Germany. Granata and Posadas (2024) applied this method to ISCO occupational descriptions to profile green jobs in Indonesia. Given that currently there is no occupational tasks database for South Africa, and the limitations of the O*NET GEP, we follow Granata and Posadas (2024) who apply a text analysis methodology to the ISCO-08 (developed in 2008) to identify occupations that perform green tasks. Granata and Posadas (2024) create a dictionary of green terms that are used to build a green occupation database. The dictionary of green terms is basically a collection of terms (words, roots, or expressions) that are commonly found in environmental economics literature. To make their list as comprehensive as possible, they draw green terms from various databases. 18 They reviewed more than 70 references to fine-tune the list. The final green terms dictionary includes 347 green terms, of which 308 are considered strict green because they are directly linked to the green economy. Another 39 terms are added to generate the broad green list of terms. These terms are not explicitly linked to environmental sustainability and climate change, but they could become directly linked if green technology or practices are adopted. These are added to the database to capture tasks that could be green. For example, the statement “land use” might not be directly linked to the green economy, but if greener technologies and practices are adopted in land use then this statement would be linked to the green economy. Any green terms clusters (a grouping of similar terms) can include only strict green terms, only terms that could be green that are used for the broad green definition, both, or neither. That said, the strict green list of terms is a subset of the broad green list of terms. Table 1 shows a sample of terms in the dictionary. 17 To build the GEP, the O*NET team reviewed the literature and identified 12 green sectors. Work activities identified as green included those that reduce the use of fossil fuels, decrease pollution and greenhouse gas emissions, increase the efficiency of energy usage, have to do with recycling materials and developing and adapting renewable sources of energy (Dierdorff et al. 2009). To identify green tasks within occupations and develop a definition of green jobs, the United States Department of Labor conducted representative surveys of job incumbents and interviewed industry experts. This information has been integrated into the United States standard occupations classification (SOC) and can be matched with labor force surveys. 18 Including the O*NET GEP, Burning Glass Technologies (BG) green list, US Bureau of Labor Statistics (BLS), GTP survey, IAB Janser (2018) list, UN Environmental GS, and the European Skills, Competences, Qualifications and Occupations (ESCO) skills taxonomy. 6 Table 1: An excerpt from the dictionary of green terms Additional terms on the Green terms clusters Strict green terms broad green terms list Agriculture, forestry, and fish Biochar, precision irrigation Agri, Crop production Environmental Knowledge Biotechno, geophysics energy engineer, wood science Environmental Regulations and environmental law, environmental Compliance liabil Greenhouse Gas Reduction emissions analyz, emissions inspect Low-carbon mobility electric vehicles, transportation motor buses, passenger efficiency train Low-polluting construction green building, sustainable lighter materials, materials lightweight construction Natural Resource Conservation Deforestation, drought land use, reforest Recycling and reuse of waste and waste management, waste Disposal materials reduction Clean Energy alternative energy, alternative fuel Common terms climate change, global warming Energy efficiency detect hot spots, efficiency heat Dictionary total (451 terms) 308 39 Source: Granata and Posadas (2024) After compiling the green dictionary, Granata and Posadas (2024) apply text analysis to all occupations in the four-digit ISCO-08 occupational taxonomy to calculate a green task intensity score for each occupation. The ISCO-08 database provides a description of each occupation at the four-digital level, including tasks carried out in the occupation. For each occupation (i), the text analysis identifies the number of tasks in the ISCO-08 occupation description that match green terms in the dictionary. Following Vona (2021), the count is used to calculate a greenness score (Green Task Intensity - GTI) for each occupation, which is the number of green tasks divided by the total number of tasks in each occupation: (1) Granata and Posadas (2024) calculate two measures of greenness for each occupation: GTI- strict and GTI-broad. The strict measure includes tasks that are matched to strict green terms in the dictionary. The broad measure includes tasks that are matched to either strict or the broad green terms, as defined in the green dictionary. An occupation may include only strict green terms, only the terms added when the green terms definition is broadened, both, or neither. The GTI-strict and GTI-broad are equal for occupations that do not include at least one of the 39 broad green terms. The GTI-strict will have a smaller value than the GTI-broad for occupations with tasks that match at least one of the 39 broad green terms. The GTI-strict will equal 0 if there are no tasks from the strict green terms list. The GTI-broad will equal 0 if there are no tasks from the GTI-broad list (which includes the GTI-strict list). 19 19 In other words, GTI-strict ≥ 0 and GTI-broad ≥ 0 and GTI-strict ≤ GTI-broad. 7 We further adjust the green dictionary so that it reflects the South African context. We reviewed the South African literature and policy reports for occupations that are expected to reduce the human-induced negative impact on the environment and climate change. We found discussion of industries that will adapt due to the greening of the economy - energy generation, resource efficiency, emissions control, and natural resource management – and industries that are most vulnerable to job loss –mining, petrochemical, electricity, agriculture, and tourism sectors (PCC 2021; ILO 2018; McLean 2018; Montmasson-Clair 2012; Maia et al. 2011; Rutovitz 2010; Nhamo 2010). None of these studies, however, takes an occupational approach, i.e., looks at the actual activities or tasks individuals do in jobs that could affect the human-induced impact on the environment. The only list of green occupations that we found was compiled by the Department of Higher Education and Training (DHET) through their organizing framework for occupations (OFO). 20 We use the DHET list to adjust the Granata and Posadas (2024) green dictionary. We follow a three-step process to adapt the green dictionary for the South African context and connect it to the PALMS database. First, we adjusted a few terms in the green dictionary based on the DHET list of green occupations: heat pump, biotechno, chemistry, geograph and geophysics. These were in the Granata and Posadas (2024) green dictionary but were not identified as strict green terms. However, they should be classified as strict in the South African context since the OFO list includes occupations with those root terms: geophysicist, biotechnologist, geophysical technician, and heat pump installer (DHET 2013). After this adjustment, we ended up with 321 strict green terms and 37 additional terms for the broad green list. Second, we use the South Africa-adjusted green dictionary and follow the method used by Vona (2019) to calculate the GTI scores for each occupation as per Equation 1. Third, to merge the GTI scores to PALMS, we used crosswalks to transform the ISCO-08 occupation codes to the ISCO-88 codes used in the PALMS. Finally, we define an occupation – and the jobs within it – as green based on its calculated GTI. An occupation is defined as green if it has a GTI greater than zero. All the jobs corresponding to that occupation are assigned the same GTI and the same greenness status. We generate a list of strictly green occupations, which are those occupations with a GTI-strict value that is a positive number. The broadly green occupation list are those occupations with a GTI-broad value that is greater than zero. We primarily use the strictly green occupations for our analysis, as the list of GTI-strict terms more concisely identifies jobs with tasks that are currently engaged in the green economy. But we present and discuss the results for both strictly and broadly green occupations, understanding that the results emerging from the analysis of strictly green jobs can be taken as the lower bound of green employment and broadly green jobs taken as the upper bound of green employment. 20 The OFO is DHET’s tool for identifying, reporting, and monitoring scarce and critical skills. This list cannot be easily mapped to household survey data because the structure of the occupation code differs from that of ISCO-88 (DHET 2013). For example, the OFO only has 8 one-digit ISCO occupations because major code 7 and 8 have been combined, making it hard to disentangle crafts and related trades and machine operators. 8 2.2 Limitations to the Methodology to Generate the List of Green Jobs There are some shortcomings to this method. First, Granata and Posadas (2024) observe that the GTI may underestimate the level of greenness of an occupation as the ISCO-88 does not present a comprehensive list of tasks for every job in an occupational category, but rather a description of what the occupation involves for the majority of jobs. Second, emerging occupations, some of which are associated to the greening of the economy, may be under the residual occupation code “Not elsewhere classified” and may not be captured since the task description of these residual occupations tend to be quite general (Granata and Posadas 2024). Third, the ISCO descriptions of occupations are static and therefore we might fail to capture new and emerging green tasks overtime. Lastly, it is possible that the text analysis methodology underestimates the share of green tasks in managerial occupations because administrative tasks such as policy makers who form clean energy and environmental policies are not emphasized. 21 2.3 Analytical Methodology We use standard econometric techniques to understand the factors associated with being in a green job, being in a greener job, and earnings in green jobs. 2.3.1 Factors associated with being in a green job To understand the factors related to being in a green job, we estimate a logit model of the probability of being in a green job as specified in equation 2. Pr( = 1 ∣ ) = �0 + + α + η + δ + θ + ε � (2) Y is a binary dependent variable equal to one if the respondent is in a green job (GTI >0) and zero otherwise. We use the strictly green job definition for the first round of estimates and the broadly green job definition for a second round of estimates. F is the logit function link while the individual characteristics (age, race, gender, education level) are represented by . α is a location dummy for urban residence, η is a vector of province dummies and δ are quarter of year dummies. θ is a vector of industry dummies. is the error term. Our sample includes working-age individuals (age 15-65) who are employed. To create the variables, we adapt some variables in the PALMS data. We convert the continuous education variable into a four-category variable: 22 1=less than grade 12 (those without a high school qualification), 2=matric (complete high school), 3=other tertiary (those with more than a high school qualification but less than university), and 4=degree (those with a university degree). For the age variable, we construct 5 categories from a continuous age variable as follows: age 15-24, age 25-34, age 35-44, age 45-54 and age 55-65. Gender is defined as a binary variable (1=female, 0=male). For ethnicity group or race, 23 we use the 21 This final caveat underlies one of the key differences between the text analysis and O*NET tags a much larger share of managerial jobs as green, as compared to the text analysis. 22 As we used categorical variables to show green employment by education and age categories in the descriptives sector, for consistency, we maintain the same categories for the regressions. The South African literature (Branson and Leibbrandt 2013) has also found that the relationship between education and labor market outcomes is convex and for this reason it is more informative to use education categories. 23 Race is included in the regression analysis because it remains an important determinant of labor 9 categories currently used by Statistics South Africa i.e., African, Coloured, Indian/Asian and White while the location variable is a dummy with urban=1 and rural=0. The province vector is a 9-category variable of the nine provinces in South Africa namely, Western Cape, Eastern Cape, Northern Cape, Free State, KwaZulu-Natal, Northwest, Gauteng, Limpopo, Mpumalanga. Industry dummies represent each of the nine one-digit groups i.e., agriculture, mining, manufacturing, utilities, construction, trade, transport, finance, and community services or other private household enterprises. 2.3.2 Factors associated with being in a greener occupation Restricting the sample to only those in a strictly green job, we estimate an ordinary least square (OLS) model of the factors associated with being in a greener occupation, as captured by the GTI, as per equation (3). ′ ln = X +α + η + δ + θ + (3) is the log of the continuous measure of strictly green task intensity (GTI-strict) in and are individual characteristics gender, age, race, and level of education. Location, province, location, industry of employment and time dummies are represented by α , η , θ and δ respectively. 2.3.3 Do green jobs pay higher wages? To understand whether there is a wage premium as a result of being in a green job, we estimate the commonly used wage regression model (Mincer 1974). We augment this regression model with industry dummies and our variable of interest, a binary variable of whether an individual is in a green job, as specified in equation 4. ′ ln = X + D′ +α + η + δ + θ + ⍴ + (4) lnWi is the log monthly wage of individual i, are individual characteristics (age, race, gender, education level). The binary variable Di takes a value of 1 if the individual i is in an occupation that has a positive share of green tasks (GTI-strict), and 0 otherwise. We estimate equation (4) again using the GTI-broad to indicate green jobs. The location, province, industry and time dummies are denoted by α , η , θ and δ respectively, and is the error term. We estimate two models, where the control variables differ across models. The first model only includes the variable of interest Di and controls for province and time dummies. In the second model, we add controls for gender, age, education, and race ( ), as well as α (location) and ⍴ (vector of nine occupation dummies). Our sample includes working-age individuals (age 15-65) who are employed and report positive earnings. We create a variable to proxy for how skill intensive an occupation is, which we use in the descriptives section. The three skill categories follow SASCO 2003 and ISCO-88 classification as market outcomes and people’s livelihoods even decades after the demise of apartheid. During apartheid, discrimination was institutionalized with policies such as the Bantu education act and the job reservation act ensuring that Black (Coloured, Indian and African) South Africans received inferior education and could not access professional occupations. 10 detailed in Table 2. High (skill) level occupations include managers, professionals and technicians. Mid (skill) level occupations include clerks, services, crafts and machine operators. Elementary (skill) level are elementary occupations, including domestic workers. Table 2: Skill levels in SASCO 2003 and ISCO-88 major code (one digit) occupations Code Major group Skill level Equivalent education 1 Legislators, senior officials, and managers * * 2 Professionals 4 Postgraduate 3 Technicians and associate professionals 3 College/tech degree 4 Clerks 2 Complete high school 5 Service workers and shop/market sales workers 2 Complete high school 6 Skilled agricultural and fishery workers 2 Complete high school 7 Craft and related trades workers 2 Complete high school 8 Plant and machinery operators and assemblers 2 Complete high school 9 Elementary occupations 1 Primary/no schooling Note: Major code 0 (Armed forces with 20 four-digit occupations) is not included in this table and is also excluded from our sample. The *asterisks mean that there is no skill reference for this occupation. In this paper we combine this occupation with professionals and technicians therefore assuming a skill level of 3 and 4. Source: SASCO 2003 3 Description of Green Jobs in the South African Labor Market 3.1 Which Occupations Are Green? In this section, we provide a description of green jobs using both the strict and broad definitions of green tasks. Using the strict 24 definition of green tasks, we find that 46 out of 428 occupations in ISCO-88 can be classified as strictly green, meaning that their strict-GTI is greater than zero. 25 When using the broad definition of green tasks, we count 137 occupations can be classified as broadly green; the additional 91 occupations are those that do not include any strictly green tasks but do include tasks on the broad green tasks list. See Tables A1 and A2 in the Appendix for the full lists of strictly green and broadly green occupations. 24 The strict definition of green tasks is measured by the total number of strictly defined green tasks in a 4-digit occupation divided by the total number of tasks in that occupation. A job is classified as green if it has a green task intensity (GTI) greater than zero, meaning it has at least one strictly defined green task. 25 Our original list had 47 strictly green occupations, including ISCO code 5169 (protective services not elsewhere classified) with a GTI of 10 percent. We chose to shift this occupation to the list of broadly defined green occupations. While this occupation contains jobs such as park rangers and wardens that would be classified as green (park rangers are classified as green in the DHET (2013) OFO skills list), the majority of the workers in this occupation are security officers whose tasks cannot be classified as strictly green. Because this occupation accounts for a very large share of jobs (over 500,000 employees), removing it from the strictly green occupations list reduced the share of strictly green employment in South Africa from 9 percent to 5.5 percent. 11 Table 3: An extract of the 20 4-digit strictly green occupations with the highest green task intensity (GTI) and the total employment in each GTI- ISCO Occupational Title Strict n 9161 Garbage collectors 79.2 61023 2112 Meteorologists 77.8 463 2211 Biologists, botanists, zoologists and related professionals 75.0 3603 2114 Geologists and geophysicists 58.3 3048 2210 Scientist 50.0 1581 8163 Incinerator, water-treatment and related plant operators 50.0 22455 3222 Sanitarians 45.0 5058 9321 Assembling laborers 41.7 2803 3119 Physical and engineering science technicians not elsewhere classified 40.0 3112 2149 Architects, engineers and related professionals not elsewhere classified 36.7 6971 2213 Agronomists and related professionals 33.3 6101 3213 Farming and forestry advisers 33.3 3831 5161 Fire-fighters 33.3 19234 7134 Insulation workers 33.3 275 2142 Civil engineers 28.6 15455 2113 Chemists 25.0 1109 7132 Floor layers and tile setters 25.0 34631 20129 7231 Motor vehicle mechanics and fitters 22.9 2 3112 Civil engineering technicians 22.2 10641 6141 Forestry workers and loggers 20.0 12121 Notes: Own calculations using PALMSv3.3. Note: the 1-digit code for each occupation is simply the first digit in the 4-digit span. The 2-digit code is simply the first 2-digits in the 4-digit span. Strictly green (4-digit) occupations are a mix of high-level, mid-level, and elementary jobs. The occupation with the highest green task intensity is the elementary occupation of garbage collectors (code 9161) with a GTI of 79.2 (Table 3). This means that 79 percent of tasks in this occupation’s description include terms from the strictly green list. The high GTI for garbage collectors is attributable to the terms “recycling” and “collecting recyclable materials or items” in the occupational tasks description, all derivations of the green dictionary term “recycl12”. This root term appears in most tasks for this occupation. 26 The next two greenest occupations require high skills. Meteorologists (code 2112) and biologists and related professionals (code 2211) have GTIs of 77.8 and 75.0 respectively; both are classified as high-level occupations. The majority of strictly green occupations are high-level, though a disproportionate share of 26 Recall that the GTI score is calculated using ISCO-08 which is a newer version of occupation classification to ISCO-88. Our data PALMS is based on ISCO-88 and the occupation Garbage collectors, ISCO-88 code 9161 maps to two ISCO-08 occupations. Garbage recyclers and collectors (code 9611) where 3 out of 4 tasks in this occupation are green with a GTI score of 75 and Refuse sorters (code 9612) where 5 out 6 tasks are green with a GTI score of 83. The GTI score for ISCO-88 code 9161 (Garbage collectors) is then 79, the average of the two which is what we report in the paper. 12 strictly green jobs are mid-level or elementary occupations. High-level strictly green occupations – defined as those with a 1-digit ISCO code of 1-3 – are 65 percent of the sample but high-skilled green jobs (jobs are the number of people working in an occupation) are only 55 percent of the sample (Figure 1). This mismatch is due to a larger number of South Africans working in mid-level (ISCO 1-digit code of 4-8) or elementary (ISCO 1-digit code of 9) strictly green occupations. For example, an estimated 61,000 work as garbage collectors (elementary occupation) and 201,000 are employed as motor vehicle and mechanics fitters (mid-level occupation). This is in contrast with only about 3,600 employed as biologists and related professionals and about 3,000 geologists and geophysicists, all classified as high-level occupations. That said, a much higher proportion of strictly green jobs are in high-level occupations (55 percent) as compared to the share of high-level occupations in the general labor market (24 percent) (Figure 1). The greater share of high-level green occupations reflects findings from similar studies in the global north (Consoli et al. 2016; Vona et al. 2019; Bowen, Kuralbayeva, and Tipoe 2018). The greenest South African occupations are disproportionately classified as elementary or mid-level. Of the 20 strictly greenest occupations, 85 percent of the jobs are in mid-level or elementary occupations (1-digit codes 5,6,7,8, and 9 in Table 3), though they only represent 8 of the top 20 occupations with the highest strictly green GTI scores (Table 3). Figure 1: Green jobs and all jobs, by skill intensity of the occupation Green strict Green broad All employed 100 100 100 55 12 24 62 80 80 80 47 60 60 60 Percent 40 37 40 40 29 20 20 25 20 9 0 0 0 Elementary Mid-level High-level Source: authors’ calculations using PALMS. Note: strictly green jobs are those with at least one task on the strictly green task list. Broadly green jobs are those with at least one task that is on the broad green task lists (which includes the strictly green tasks). Broadly green jobs are more prevalent in mid-level and elementary occupations. Once the definition of green tasks is expanded to include broad green tasks, 88 percent of jobs are in mid-level and elementary occupations. For example, if we allow that occupations under code 7 – such as extraction and building trades (code 71), metal machinery and related trades (code 72) and precision, handicraft and craft trades (code 73) – may adopt green technologies, they 13 would be considered green. A much lower share of broadly green jobs (12 percent) can be classified as high-level occupations, as compared to the overall labor market (24 percent). Thus, while strictly green occupations are highly skilled and concentrated in high-level occupations, there is potential for more employment in mid-level and elementary occupations if green technology and processes are more widely adopted to stem the impact of humans on the environment. 3.2 How Prevalent Are Green Jobs in South Africa? The 46 occupations that can be classified as strictly green account for about 5.5 percent of total employment in the South African labor market that employs 16.4 million people. This share has barely moved over the period 2012-2019 with only a small increase in 2014 (Figure 2). The lack of increase in the strictly green jobs trend could be an indication of a slow implementation of green economy practices. Or it may be a casualty of the data since the task description of occupations is static. In other words, jobs may have adopted technology or processes that require tasks in the strictly green list of the green dictionary, but these tasks may not be reflected in the ISCO-88 occupational task descriptions. Further, strictly green jobs may have added more green tasks over time, but greening within a job is not captured in our data. Figure 2: Green employment over time (2012-2019) 6 33 5.5 5 Share green broad (%) 32 Share green strict (%) 4.5 4 31 3.5 3 30 1 1 1 2 1 3 1 4 1 1 1 2 1 3 1 4 1 1 1 2 1 3 1 4 1 1 1 2 1 3 1 4 1 1 1 2 1 3 1 4 1 1 1 2 1 3 1 4 1 1 1 2 1 3 1 4 1 1 1 2 19 3 q4 20 2q 20 2q 20 2q 20 2q 20 3q 20 3q 20 3q 20 3q 20 4q 20 4q 20 4q 20 4q 20 5q 20 5q 20 5q 20 5q 20 6q 20 6q 20 6q 20 6q 20 7q 20 7q 20 7q 20 7q 20 8q 20 8q 20 8q 20 8q 20 9q 20 9q 20 9q 1 20 Year Green strict Green broad Source: authors’ calculations using PALMS. Expanding the definition to capture broadly green jobs, the share of broadly green employment is nearly 33 percent in 2019, up from 31 percent in 2012. A look at employment trends over time shows that there was a visible upward shift in 2013-2014 as a result of an increase in employment in a few mid-level and elementary occupations, specifically extraction and building trades (ISCO code 71), metal machinery and related metal trades (code 72), laborers in agriculture (code 92) and laborers in mining and construction (codes 93). These occupations, especially extraction and building trades and metal machinery and related metal trades, include many broadly green 4-digit ISCO occupations. The increase in broadly green 14 jobs is temporally aligned with the Renewable Energy Independent Power Producer Programme (REIP4) that offered incentives to firms to invest in green energy production. The program was introduced in 2011, active in 2013 and 2014, and slowed down after 2014, mirroring the potential green jobs trend in Figure 2. 27 The share of green jobs in the South African labor market greatly differs by gender with only 3 percent of working South African women being in strictly green occupations, compared to 8 percent of men. When using the broad definition of green jobs, 14 percent of women are in broadly green jobs compared to 47 percent of men. As we discuss below, the gender difference in the share of green jobs is linked to the persistent gender occupational segregation in the labor market. The South Africa estimates are within the range of estimates from other countries. Granata and Posadas (2024), whose methodology we use in this paper, report 2.3 percent green employment for Indonesia. 28 Similarly, Doan et al (2022) find that 3.6 percent of Vietnamese jobs can be classified as strictly green while 41 percent are broadly green. Using the O*NET definition of green occupations and a continuous measure of green jobs, Vona et al. (2019) estimate the share of green jobs in the US to be between 2-3 percent over the period 2006- 2014. Using a binary definition of green jobs which allowed for a broader set of green occupations, Consoli et al. (2016) report that between 9.8 and 12.3 percent of U.S. workers were in green employment in the period 2011-2012. In another study from the U.S, Bowen et al. (2018) estimate that 10.3 percent of all employment in the US were in “directly green” occupations while "indirectly green" occupations made up another 9.1 percent of all employment. Valero et al. (2021) estimate the share of green employment in the UK to be between 17 percent and 39 percent in 2019. The higher share of green jobs for the UK as compared to other countries is partly due to data challenges when cross-walking between the O*NET which uses the US 8-digit standard occupational classification and the UK labor force surveys which is based on the 4-digit ISCO-08 classification. 27 Introduced in 2011, the REIP4 used a competitive procurement program to increase electricity generation through renewable energy sources. Investing companies needed 40 percent local ownership, leading to increased partnerships between foreign and South African companies, including knowledge and skill sharing and green jobs (ILO 2018). The increase in the demand for solar PVs may have increased the demand for solar panel installers and assemblers, explaining the increase in potentially green jobs between 2013 and 2014. However, the stalling of the REIP4 initiative after 2014 meant that investment in renewable energy technology also stalled corresponding to the flattening of the trend. 28 Although we use the same methodology as Posadas and Granata (2024), there are a few reasons that could explain the differences in the estimated share of green jobs in South Africa and Indonesia. First, the Granata and Posadas (2024) analysis excludes agricultural occupations while we include the agricultural occupations in the sample. Second, it is possible that the South African economy differs from the Indonesian economy both in structure and in the profile of workers. For example, the mining sector, which will be substantially affected by the green transition, contributes to a significant share of employment in the South African economy and for this reason, our green occupations list includes occupations such as mining managers and mining engineers, occupations which are not included in Granata and Posadas’ (2024) green occupations list. 15 3.4 Who Works in Green Occupations? Consistent with international literature (Vona 2021; Valero et al. 2021, Doan 2022), we find that regardless of the definition (strict or broad), men are over-represented in green jobs. While women account for 44 percent of all workers, only 21 percent of strictly green jobs are held by women compared to 79 percent held by men (Figure 3). Women account for an even smaller share of broadly green jobs (19 percent). The underrepresentation of women in green jobs stems from gendered occupational segregation in the labor market, such that men and women sort into different occupations. Of the 20 occupations in Table 2, only three have a female share of more than 50 percent, 29 all of which employ a relatively small share of workers. The two-digit ISCO level occupations with the greatest number of four-digit green occupations are physical, mathematical, and engineering professionals (code 21) where the female share is only 22 percent. In contrast, office clerks (code 41) have zero GTI (strict or broad) though 70 percent of people working in the occupation are women. The green occupations where women are well represented, such as life science and health professionals (code 22), employ a relatively small proportion of South Africans. 30 Figure 3: Share employed in green occupations by gender Green strict Green broad All employed 100 100 100 21 19 80 80 80 44 Green employment by gender 60 60 60 40 79 40 81 40 56 20 20 20 0 0 0 Male Female Source: authors’ calculations using PALMS A disproportionate share of white South Africans hold strictly green jobs as compared to the general employed population. As we have shown above, strictly green occupations are concentrated among highly skilled occupations. The continued gaps in educational attainment inherited from the apartheid era mean that white South Africans are better prepared for strictly green occupations. While Black South Africans dominate green and all jobs due to their high population share in South Africa, they are 60 percent of strictly green jobs but 74 percent of the working population. Instead, white workers hold 24 percent of strictly green jobs as compared to 13 percent of all jobs in the general labor market. There is little difference in representation between strictly green and non-green of the other racial or ethnic groups. The gaps by race between broadly green jobs and the general labor market are much narrower. White South Africans hold 9 percent of broadly green jobs while Black South Africans hold 78 29 Meteorologists (2112), biologists (2211) and chemists (2113). 30 For example, while the female share in the life science and health professionals is above 60 percent, this occupation only employs 3 percent of all women and 0.9 percent of all men (Mosomi et al. 2020). 16 percent of the jobs, as compared to 13 percent and 74 percent, respectively, for the general labor market (Figure 4). Figure 4: Share employed in green occupations by race Green strict Green broad All employed 100 100 100 9 13 2 24 3 11 80 80 80 10 5 11 60 60 60 Percent 40 40 78 40 74 60 20 20 20 0 0 0 African/Black Coloured Indian/Asian White Source: authors’ calculations using PALMS The distribution of green jobs by age reflects the overall labor market. Strictly green jobs are most prevalent within the 25-34 (35 percent) and 35-44 (30 percent) age groups (Figure 5), as compared to 62 percent of all jobs belonging to the prime working age group (25-44). When considering both the strict and broad definitions of green jobs, the 25-34 age group has a slightly higher propensity for green jobs than for all jobs, by a few percentage points (Figure 5). The older (55-65) and youngest (15-24) age groups hold 9 percent and 8 percent of all jobs, strictly green jobs, and broadly green jobs. Figure 5: Share employed in green occupations by age Green strict Green broad All employed 100 100 100 9 9 9 18 19 20 80 80 80 30 29 29 60 60 60 Percent 40 40 35 40 35 33 20 20 20 8 8 8 0 0 0 15-24 25-34 35-44 45-54 55-65 Source: authors’ calculations using PALMS 17 Compared to the general labor market, individuals with some tertiary level qualifications are overrepresented in strictly green jobs while individuals with high school or less are overrepresented in broadly green jobs. Of all workers in strictly green jobs, an estimated 27 percent are university graduates and 9 percent have a post high school qualification (principally a technical or vocational school certificate). This contrasts with only 17 percent of workers with university degrees and 6 percent with post-secondary qualifications in the general labor market. Still, the largest share of strictly green jobs is held by those with a high school or lower qualification (64 percent). This is because, as discussed above, while the majority of strictly green occupations are high skilled, more people are employed in mid-level and low skill elementary jobs in the South African labor market. A high share (59 percent) of the workers in broadly green jobs have less than a high school qualification, while only 8 percent of broadly green jobs belong to university graduates. This is consistent with the above discussion that broadly green occupations include many lower skilled occupations (Figure 6). 31 Figure 6: Share employed in green occupations by education level Green strict Green broad All employed 100 100 100 27 8 17 6 27 80 80 80 6 32 9 60 30 60 60 Percent 59 40 40 46 40 34 20 20 20 0 0 0 Less than gr12 Grade 12 Other tertiary Graduates Source: authors’ calculations using PALMS The industries with the highest shares of green jobs are manufacturing, trade, and finance but compared to the general employed population, manufacturing, utilities, and finance sector jobs are over-represented among green jobs (Figure 7). The green jobs in the utility industry are those related to electricity, gas and waste collection and purification. The finance sector enables others to develop and use green technologies, including research and development, renting of machinery and other business activities, computer and related services and financial intermediaries. While finance might not seem like a directly green occupation, other studies have found that it is a sector that records a significant share of green jobs (Valero et al. 2021). This reinforces the idea that quantifying green jobs through the occupational lens allows one to capture green jobs in industries that might not be thought 31 Doan et al. (2022) find that, on average, green jobs in Viet Nam require higher qualifications than non-green jobs. 18 of as green. 32 Figure 7: Share employed in green occupations by Industry Green strict Green broad All employed 100 100 100 Distribution of employment by main industry 15 16 31 80 80 80 16 17 5 9 14 60 60 60 24 9 6 40 15 40 40 21 7 2 4 13 8 20 20 20 20 1 5 11 4 14 3 5 5 0 0 0 Agriculture Mining Manufacturing Utilities Construction Trade Transport Finance Services CSP Source: authors’ calculations using PALMS 4 Regression Results Conditional correlations allow us to understand the nuance in the patterns observed in the comparison of means. 4.1 Factors Associated with Being in a Green Job Regardless of the definition used (strictly or broadly green jobs) women are less likely to be in green jobs (Table 4). Women are 5 percent less likely than men to be in strictly green jobs and 25 percent to 31 percent less likely than men to hold broadly green jobs. The conditional results reflect the clustering of broadly green jobs in typically male dominated occupations. Workers aged 25-44 are the most likely to be in strictly and broadly green jobs. The conditional correlates for the age dummy variables are not stable for strictly green jobs, with statistical significance changing based on the set of control variables. The estimated correlation for the two oldest categories is negative, though with a very small value, reflecting the unconditional correlations. The estimates for the broadly green jobs are consistent across age groups, with all age cohorts being more likely to work in a green job as compared to the youngest age group. This result is consistent with the unconditional estimates in the descriptive section. More educated workers are more likely to be in strictly green jobs while high school dropouts are most likely to hold broadly green jobs. This reflects the unconditional correlations, where workers with more than a high school (matric) education were over-represented among strictly green workers. The results for broadly green jobs also follow the conditional correlations since the category includes several populous mid-skilled and elementary 32 Doan et al. (2022) find that green jobs in Viet Nam are concentrated in utilities (electricity, gas, water), mining and quarrying, and market services. 19 occupations (Table 2), that are aligned with lower levels of education. Table 4: The probability of being in a green job in 2019 Model Pr(strictly green job) Pr(broadly green job) Marginal effects dy/dx dy/dx dy/dx dy/dx Female -0.060*** -0.053*** -0.314*** -0.250*** (0.001) (0.001) (0.001) (0.001) base=age 15-24 25-34 0.005*** 0.005*** 0.042*** 0.033*** (0.001) (0.001) (0.002) (0.002) 35-44 0.001 0.004*** 0.037*** 0.031*** (0.001) (0.001) (0.002) (0.002) 45-54 -0.005*** -0.001 0.010*** 0.013*** (0.001) (0.001) (0.003) (0.002) 55-65 -0.003** 0.000 0.008*** 0.013*** (0.002) (0.002) (0.003) (0.003) base=less than matric Matric 0.007*** 0.007*** -0.105*** -0.074*** (0.001) (0.001) (0.001) (0.001) Other tertiary 0.036*** 0.036*** -0.054*** -0.039*** (0.001) (0.001) (0.003) (0.003) Degree 0.034*** 0.042*** -0.233*** -0.198*** (0.001) (0.001) (0.002) (0.002) base=black/African Coloured 0.022*** 0.018*** 0.012*** -0.019*** (0.001) (0.001) (0.002) (0.002) Indian 0.014*** 0.007*** -0.097*** -0.079*** (0.002) (0.002) (0.004) (0.004) White 0.033*** 0.028*** -0.007*** -0.040*** (0.001) (0.001) (0.002) (0.002) Urban 0.011*** 0.011*** -0.086*** -0.021*** (0.001) (0.001) (0.002) (0.002) Industry dummies No Yes No Yes Observations 609,628 609,472 609,628 609,472 Standard errors in parentheses *** p<0.01, ** p<0.05, *p<0.1 Logit estimate (equation 2), reporting marginal effects. The dependent variable takes a value of 1 if the observation reports a strictly (columns 1 and 2) or broadly (columns 3 and 4) green job (GTI-strict > 0). Survey weights applied in all regressions. Sample includes employed individuals aged 15-65. All columns include survey quarter dummies, province dummies. Source PALMS V3.3. Black Africans are least likely to be in strictly green jobs, even when controlling for education level, but most likely to hold broadly green jobs. This may be expected due to the disproportionately high share of high-level jobs in strictly green occupations, where Black Africans are under-represented due to the persistent occupational segregation by race in the 20 South African labor market (Gradín 2019). 33 In contrast, Black male workers are overrepresented in manual occupations such as machine plant operators and assemblers, which are classified as broadly green jobs and are numerous. Relative to rural workers, urban workers are more likely to be in strictly defined green jobs while rural workers are more likely to be in broadly green jobs. The results for strictly green jobs do not change when industry dummies are added to the regression. This may reflect the more urban nature of strictly green jobs, in professional fields as well as public services. However, the point estimate for broadly green jobs decreases, while remaining negative, suggesting that manufacturing industries, which are broadly green jobs, may be partly driving the results. 4.2 Which Green Workers Are in the Greenest Jobs? When restricting the sample to only those working in strictly green occupations, 34 results show that not only are women less likely to be in strictly green jobs (Table 4), but they are also less likely to be in the greenest jobs (Table 5). Women are 8.2 percent less likely to be in the greenest jobs. The coefficient estimate remains negative when controlling for industry. Relative to the youngest age group, older individuals are more likely to be in the greenest jobs. The coefficients are positive and statistically significant for the oldest age groups (age 45-54 and 55-65) with the oldest age group being 8.5 percent more likely to be in the greenest jobs as compared to the youngest age group (15-24). Thus, while the oldest age groups were the least likely to be in green jobs, those who do end up in green jobs are in the greenest. Individuals with less than complete high school education are more likely to be in the greenest jobs (Table 5), even though they were less likely to be in green jobs as compared to other age groups (Table 4). Every education group is 30 percent less likely than workers without completed high school to be in the greenest jobs. This result stems from the fact that while most green jobs are high skilled, some low-skill, high employment occupations, such as garbage collectors and motor vehicle mechanics and fitters have high GTI scores, resulting in the greenest jobs belonging to those with lower than grade 12 level of education (see Figure 6). Individuals from the African subpopulation are more likely to be in the greenest of the green jobs. All other population groups are on average less likely to be in the greenest jobs, even when controlling for education level, industry, and urban residence. Once controlling for industry, the gap reduces, but remains negative. So, while Black Africans are less likely to work in a strictly green job, when they do have that job, it is greener. Urban workers are more likely to work in the greenest of the strictly green jobs. Individuals working in the urban areas are 15 percent more likely to be in the greenest jobs. This seems to reflect the industrial distribution across locations, though. When controlling for industry, the 33 The legacy of apartheid labor market policies such as the job reservation act and the Bantu education act is still reflected in the present labor market, with Black South Africans being underrepresented in professional occupations and in institutions of higher learning and overrepresented in elementary occupations. 34 We do not do this exercise for potentially green jobs since the broad green definition includes tasks that could become green (hence the green task intensity (GTI_broad)), but this transition is far from certain. 21 correlation disappears. Table 5: Characteristics of those in a green job Model OLS, y=ln(GTI) Female -0.082*** -0.060*** (0.014) (0.014) base=age 15-24 25-34 0.002 0.018 (0.020) (0.019) 35-44 0.020 0.020 (0.020) (0.019) 45-54 0.046** 0.046** (0.022) (0.020) 55-65 0.084*** 0.084*** (0.024) (0.023) base=less than matric Matric -0.344*** -0.307*** (0.013) (0.012) Other tertiary -0.449*** -0.418*** (0.019) (0.018) Degree -0.475*** -0.399*** (0.015) (0.015) base=black/African Coloured -0.146*** -0.067*** (0.019) (0.017) Indian -0.289*** -0.178*** (0.026) (0.025) White -0.369*** -0.188*** (0.015) (0.014) Urban 0.155*** 0.008 (0.016) (0.014) Constant 2.944*** 3.390*** (0.040) (0.041) Industry dummies No Yes Observations 31,638 31,637 R-squared 0.131 0.273 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 OLS estimates, as per equation (3). The dependent variable is ln(GTI-strict). Sample includes employed individuals aged 15-65 who work in strictly green occupations. All models include survey quarter dummies and province dummies. Model 1 excludes industry dummies while model 2 is the full sample of those in strictly defined green jobs. Survey weights applied in all regressions. Source PALMS V3.3. 4.3 Is There a Wage Premium for Being in a Green Job? There is a wage premium to green jobs. Table 6 presents regression results as per equation 4. Model 1, where we only include a strictly green job dummy, dummy variables for location 22 and province, and time fixed effects, shows that those in a strictly green job earn nearly 43 percent higher wages than those who are in jobs without any green tasks. When adding the full set of demographic, industry, and occupation controls, which are often correlated with wages, the green premium is still a positive and statistically significant 2.7 percent. 35 If we include the broad definition of a green job, the wage premium disappears. The coefficient on green_broad is negative. Workers in broadly green jobs earn 14.9 percent less than those in non-green jobs. The gap narrows to 9.9 percent when including the full set of demographic, occupation, and industry controls. This follows from the descriptive section where we showed that once we expand the definition of a green job to include broadly green occupations, mid-level and elementary occupations (1-digit ISCO codes 7, 8, and 9) are more heavily weighted in the broadly green sample relative to the non-green sample. Table 6: Wage regression, using the strict and broad definition of green jobs Models Strictly Green Jobs Broadly Green Jobs Variables Baseline Full Baseline Full green_strict (green_broad) 0.432*** 0.027*** -0.149*** -0.099*** (0.011) (0.010) (0.005) (0.005) Female -0.263*** -0.281*** (0.004) (0.004) Age 0.032*** 0.032*** (0.001) (0.001) Age2 -0.000*** -0.000*** (0.000) (0.000) base=less than matric Matric 0.287*** 0.283*** (0.005) (0.005) Other tertiary 0.480*** 0.479*** (0.009) (0.009) Degree 0.821*** 0.811*** (0.009) (0.009) base=black/African Coloured 0.054*** 0.054*** (0.007) (0.007) Indian 0.110*** 0.108*** (0.015) (0.015) White 0.270*** 0.271*** (0.008) (0.008) Urban 0.478*** 0.200*** 0.466*** 0.198*** (0.005) (0.005) (0.005) (0.005) Constant 8.117*** 7.234*** 8.201*** 7.287*** (0.012) (0.038) (0.012) (0.038) Observations 419,148 415,553 419,148 415,553 R-squared 0.070 0.345 0.067 0.346 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 35 Doan et al. (2022) carry out a similar conditional estimate and do not find a wage premium for green jobs relative to non-green jobs. 23 OLS estimates, as per equation (4). Dependent variable is ln(earnings). Sample includes employed individuals aged 15-65. Income bracket weights applied. The ‘baseline’ columns control for survey quarter dummies and 9 province dummies. The ‘Full’ columns add demographic characteristics (gender, age, education, and race), 9 main industry dummies, a location dummy and 9 main occupation dummies to the variables in the baseline model. Source: PALMS v3.3 data. Our variable of interest is the variable green_strict which is a binary variable equal to one if an individual is in a green job. 5 Conclusion and Policy Directions This paper aims to profile green jobs in the South African labor market. We utilize the PALMS labor force survey data 2012-2019 and apply an occupational task-based approach to identify green jobs, count them, and profile the workers and the wages for green jobs. Through the process, we adapt a green terms dictionary to be relevant for the South African labor market. We find that 5.5 percent of South Africa’s jobs can be classified as strictly green, while 32 percent can be classified as broadly green. The share of strictly green jobs is in the range of estimates from other countries, spanning 2.3 percent in Indonesia to 39 percent in the UK. South Africa’s share of strictly green jobs has barely changed over the last 10 years, while broadly green jobs are slowly increasing. This may be a statistical anomaly driven by the limitations in our data that do not allow us to update the task content of jobs as they evolve. Alternatively, it may be a signal that strictly green job growth is stagnant in the South African economy, even as broadly defined green sectors are growing. This brings into question if today’s green policies are sufficiently transformational to shift the South African economy, and its labor force, to greener jobs. While 65 percent of South Africa’s strictly green occupations can be classified as high-level, only 55 percent of green workers are in those jobs. South Africa has a higher number of strictly green occupations among high-level occupations (e.g., science and engineering professionals and technicians) than among mid-level and elementary occupations. Those working in these jobs enjoy a significant wage premium. However, elementary and mid-level strictly green occupations employ a disproportionately larger share of people in South Africa than high- level strictly green jobs do. In short, a narrow set of occupations have the potential to sweep up large numbers of typically excluded workers as the economy greens. The green economy offers jobs across the socio-economic spectrum in South Africa. While studies from the global north find that green jobs are concentrated in professional and managerial jobs (Vona 2021; Consoli et al. 2016; Bowen et al. 2018; Valero et al. 2021), our results, especially under the broad definition of green jobs, show that there is an opportunity for green jobs in mid-level and elementary occupations. Similar results emerge in the middle- income country of Viet Nam (Doan et al 2022). This is a significant finding for an economy such as South Africa that has a large low-skilled workforce and high rates of unemployment among more vulnerable groups. The profile of green job workers illustrates exclusion of some groups, but inclusion of others. Most strictly green occupations are dominated by older males, a fact that reflects persistent gendered occupational segregation and high levels of youth unemployment. Green occupations are concentrated in engineering and mathematical professional occupations or mid-level and elementary skill blue collar jobs where women are underrepresented. Women in professional occupations are concentrated in teaching, health professions and office clerks, all of which have a very low (or zero) GTI. Younger individuals (15-24) are also less likely to be in green jobs. However, workers in the greenest of the green occupations are those who tend 24 to be most excluded from the general labor market: Black African workers with lower than matric qualifications. We attribute this result to the fact that the few mid-level and elementary occupations that can be classified as strictly green (e.g., refuse and garbage collectors), exhibit some of the highest GTIs and are responsible for a large share of employment. Five policy areas emerge from the analysis. First, as the green economy grows, there will be a need to upskill the current work force and prepare the incoming work force with skills used in strictly and broadly green occupations. The 2022 critical skills list identifies 101 occupations that cannot be filled by South Africans; 40 are on our strictly green list, pointing to the current scarcity of local skills to fill (mostly) high-level green jobs. 36 Even more worrisome is that while most of the skills required for a green economy could be delivered by some slight adjustments to the education system, it is not happening on the required scale (Duncan 2023). A skills development strategy to carry out the tasks defined in the green jobs identified in this paper can create some order and efficiency in the scattered and ad hoc current offering of green skills programs (ILO 2019). It can be used to incentivize universities to expand curriculum that aligns with the green occupations identified in this paper to build the missing cohort of scientists and managers needed to spur the green revolution. It can also be a means for expanding and organizing the TVET system to prepare workers in mid-level occupations, particularly for broadly green jobs. Partnerships with the private training sector, which is already active in providing short courses in a range of green-related technologies, will allow for a rapid expansion of training opportunities to engage less-skilled youth in green occupations (Duncan 2023). Even workers in elementary occupations will need to upskill and reskill as new technologies and industries come on the market. This group is best served through practical, short courses provided through municipalities or other local private sector players. Second, the low prevalence of youth in green jobs suggests a need to proactively engage youth to view green jobs as a viable job opportunity. Three types of interventions can bring more youth to green jobs. First, career counseling programs, such as the SkillCraft career guidance portal on SAYouth.mobi, can present green occupations as career options and guide youth toward training programs to prepare to enter these fields. Second, scholarships for TVET study can be earmarked for green occupations. Third, the Presidential Youth Employment Initiative, Extended Public Works Program, and other programs that provide youth a first job experience can expand their offering of strictly or broadly green jobs. Third, women’s under-representation in green jobs needs to be addressed before they are left out of the green transition. Girls can be encouraged to engage in STEM fields, as is shown through computer programming clubs in high schools (Yabas et al 2022), scholarships for girls entering STEM fields in TVET or university, and career guidance that provides earnings information (among other) for STEM careers. Publicly recognizing and celebrating successful women in STEM or mentorship programs between female STEM students or young women in the STEM workforce with established women in STEM fields can help newcomers navigate the male world of green jobs. A societal shift in mindset – that STEM is for girls and women – through public campaigns has been successful in other countries (Hill et al 2010). 36 Department of Home Affairs. 2022. Government Gazette (Vol. 680) #45860 of February 2022. Republic of South Africa. 25 Fourth, elementary strictly green jobs are home to many excluded workers that play a vital role in the green economy value chain but may be even more productive through organization. For example, waste pickers in South Africa are beginning to organize by registering with municipal authorities, organizing their work streams to increase the value added, and gaining some worker rights. As the green economy expands, such organized groups of workers may play a stronger role (Medina 2008). Finally, as South Africa moves toward a green economy, it will be even more important to understand and track the impact on jobs. This paper used proxies to estimate the number and profile of green jobs in South Africa. But better measures will be needed. This highlights the need to develop detailed and regularly updated occupation-task databases, such as through web scrapes or other big data sources. It will also require academia-industry collaboration to more succinctly define the South African skill profile needed in emerging green jobs, as well as profiling those occupations that will disappear through the transition and related necessary reskilling. 26 References Acemoglu, D., and Autor, D. (2011). ‘Skills, tasks and technologies: Implications for employment and earnings’. In Handbook of labor economics (Vol. 4, pp. 1043–1171). Elsevier. Arndt, C., Gabriel, S., Hartley, F., Strzepek, K. M., and Thomas, T. S. (2021). ‘Climate change and the green transition in South Africa’. In A. Oqubay, F. Tregenna, and I. Valodia (eds), The Oxford Handbook of the South African Economy. Oxford: Oxford University Press. Autor, D. H., Levy, F., and Murnane, R. J. (2003). ‘The skill content of recent technological change: An empirical exploration’. The Quarterly journal of economics, 118(4): 1279– 1333. Bowen, A., Kuralbayeva, K., and Tipoe, E. L. (2018). ‘Characterising green employment: The impacts of ‘greening’ on workforce composition’. Energy Economics, 72(): 263–275. Branson, N., and M. Leibbrandt (2013). ‘Educational Attainment and Labour Market Outcomes in South Africa, 1994–2010’. Economics Department Working Paper 1022. Paris: OECD. Branson, N., and Wittenberg, M. (2014). ‘Reweighting South African National Household Survey Data to Create a Consistent Series Over Time: A Cross‐Entropy Estimation Approach’. South African Journal of Economics, 82(1), 19-38. Consoli, D., Marin, G., Marzucchi, A., and Vona, F. (2016). ‘Do green jobs differ from non- green jobs in terms of skills and human capital?’ Research Policy, 45(5): 1046–1060. Cunningham, W., Moroz, H., Muller, N, and Solatorio, A. (2022). "The Demand for Digital and Complementary Skills in Southeast Asia." Policy Research Working Paper #10070. World Bank: Washington, DC. DEA. (2011). National climate change response white paper. Department of Environmental Affairs, Republic of South Africa Pretoria. Department of Higher Education and Training (DHET) (2013). Organising Framework for Occupations (OFO). Pretoria: DHET. Dierdorff, E. C., Norton, J. J., Drewes, D. W., Kroustalis, C. M., Rivkin, D., and Lewis, P. (2009). Greening of the world of work: Implications for o*net®-soc and new and emerging occupations. Raleigh, NC: The National Center for O* NET Development. Doan, D., Luu, T., Nguyen, N.T., and Safir, A. (2022). Green Jobs: Upskilling and Reskilling Vietnam’s Workforce for a Greener Economy. World Bank: Washington, DC. https://documents1.worldbank.org/curated/en/099091823063517461/pdf/P1781121 e19c8506719ff81042e6be4e00b.pdf Duncan, K. (2023). A Review of Green Skills Training Programmes in South Africa. World Bank/AFRE2, mimeo. Goos, M., Manning, A., and Salomons, A. (2014). ‘Explaining job polarization: Routine-biased technological change and offshoring’. American economic review, 104(8): 2509–2526. Engelbrecht F, Adegoke J, Bopape MJ, Naidoo M, Garland R, Thatcher M, McGregor J, Katzfey J, Werner M, Ichoku C, Gatebe C. (2015). ‘Projections of rapidly rising surface temperatures over Africa under low mitigation’. Environmental Research Letters, 10(8), 085004. Granata, J. and Posadas, J. (2024). ‘Why Look at Tasks When Designing Skills Policy for the Green Transition ? A Methodological Note on How to Identify Green Occupations and the Skills They Require.’ Policy Research Working Paper #10753. Washington, DC: World Bank. 27 Hill C., Corbett C., St. Rose A. (2010). Why So Few? Women in Science, Technology, Engineering, and Mathematics. AAUW (American Association of University Women): Washington, DC. Inglesi-Lotz, R. (2021). ‘Socio-economic Aspects of Energy and Climate Change in South Africa’. In A. Oqubay, F. Tregenna, and I. Valodia (eds), The Oxford Handbook of the South African Economy (p. 305). Oxford: Oxford University Press. International Labour Organisation (ILO). (2018). Skills for green jobs in South Africa. International Labour Office: Geneva, Switzerland. ILO (International Labour Organisation) and AFD (Agence Française de Développement). 2019. Skills for Green Jobs in South Africa. Geneva: ILO. Janser, M. (2018). ‘The greening of jobs in Germany: First evidence from a text mining based index and employment register data’, IAB-Discussion Paper, No.14/2018, Institut für Arbeitsmarkt- und Berufsforschung (IAB), Nürnberg. Kerr, A., Lam, D., and Wittenberg, M. (2019). Post-Apartheid Labour Market Series: 1993– 2019 [Dataset]. University of Cape Town: DataFirst [producer and distributor]. (Version 3.3) McLean, D. (2018). ‘Unlocking green jobs in South Africa: A catalytic intervention synthesis report’. Trade and Industrial Policy Strategies. Pretoria Maia, J., Giordano, T., Kelder, N., Bardien, G., Bodibe, M., Du Plooy, P., Jafta, X., Jarvis, D., Kruger-Cloete, E., Kuhn, G., et al. (2011). ‘Green jobs: An estimate of the direct employment potential of a greening South African economy’. Makgetla, N. (2021). ‘The Just Transition in Coal’. Working paper for the Presidential Climate Commission. Pretoria: Trade and Industrial Policy Strategies (TIPS) Makgetla, N., and Patel, M. (2021). ‘The coal value chain in South Africa’. Pretoria: Trade and Industrial Policy Strategies (TIPS). Medina, M. (2008). “The informal recycling sector in developing countries : organizing waste pickers to enhance their impact.” Gridlines, #44. Washington, D.C.: World Bank Group. Maseko, N. (2021). ‘Unemployment and Sustainable Livelihoods: Just Transition Interventions in the Face of Inequality’. Pretoria: Trade and Industrial Policy Strategies (TIPS) Merven, B., Hartley, F., McCall, B., Burton, J., & Schers, J. (2019). Improved representation of coal supply for the power sector for South Africa. SA-TIED working paper, 84. Mincer, J. (1974). ‘Schooling, experience, and earnings. Human behaviour and social institutions no. 2.’ Mosomi, J., Thornton, A., and Branson, N. (2020). ‘Unpacking the Potential Implications of COVID-19 for Gender Inequality in the SA Labour Market’. Working Paper No. 269. University of Cape Town: Southern African Labour and Development Research Unit. Montmasson-Clair, G. (2012). Green Economy Policy Framework and Employment Opportunity: A South African Case Study. Pretoria: Trade and Industrial Policy Strategies (TIPS), Working Paper Series 2012-02. Nhamo, G. (2010). Green economies and green jobs: Implications for South Africa. WIT Transactions on Ecology and the Environment, 131, 257-268. NPC. (2011). National Development Plan: Vision 2030. Tech. Rep. Pretoria: National Planning Commission. Rutovitz, J. (2010). ‘South African energy sector jobs to 2030: How the Energy [R] evolution will create sustainable green job’. Gradín, C. (2019). ‘Occupational segregation by race in South Africa after apartheid’. Review of Development Economics, 23(2), 553-576. 28 Presidential Climate Commission (PCC) (2021). ‘Laying the foundation for a Just Transition for South Africa.’ A presidential climate commission report. Stanef-Puică, M. R., Badea, L., Șerban-Oprescu, G. L., Șerban-Oprescu, A. T., Frâncu, L. G., and Crețu, A. (2022). ‘Green Jobs—A Literature Review’. International Journal of Environmental Research and Public Health, 19(13), 7998. Statistics South Africa (2022). Quarterly Labour Force Surveys quarter 1 2022. Statistical release P0211. Pretoria: Statistics South Africa UNEP/ILO/IOE/ITUC. (2008). Green Jobs: Towards Decent Work in a Sustainable, Low-Carbon World. United Nations Office at Nairobi (UNON). Valero, A., Li, J., Muller, S., Riom, C., and Nguyen-Tien, V. (2021). ‘Are ‘green’ jobs good jobs? How lessons from the experience to-date can inform labour market transitions of the future’. Grantham Research Institute on Climate Change and the Environment and Centre for Economic Performance, London School of Economics and Political Science: London. Vona, F. (2021). ‘Labour markets and the green transition: a practitioner’s guide to the task- based approach (Vol. 126681)’. Publications Office of the European Union. Vona, F., Marin, G., and Consoli, D. (2019). ‘Measures, drivers and effects of green employment: evidence from us local labor markets, 2006–2014’. Journal of Economic Geography, 19(5): 1021-1048. Vona, F., Marin, G., Consoli, D., and Popp, D. (2018). ‘Environmental regulation and green skills: an empirical exploration’. Journal of the Association of Environmental and Resource Economists, 5(4): 713–753. World Bank Group. (2018). ‘Overcoming poverty and inequality in South Africa: An assessment of drivers, constraints and opportunities’. Book. Washington DC: World Bank. World Bank Group. (2022). ‘South Africa Country Climate and Development Report. CCDR Series’. Tech. Rep. Washington, DC: World Bank. Yabas, D., Kurutas, B.S., Corlu, M.S. (2022). “Empowering Girls in STEM: Impact of the Girls Meet Science Project.” School Science and Mathematics, 122(5): 247-258. 29 Table A1: Strictly Green Occupations ISCO code Occupational title GTI strict Total employ 1221 Production and operations departmen managers in agriculture, hunting, forestry and fishing 4,17 9 464 1222 Production and operations department managers in manufacturing 9,17 47 087 1311 General managers in agriculture, hunting, forestry and fishing 2,83 21 310 1312 General managers in manufacturing 9,17 11 751 2112 Meteorologists 77,78 463 2113 Chemists 25,00 1 109 2114 Geologists and geophysicists 58,33 3 048 2141 Architects, town and traffic planners 16,67 12 629 2142 Civil engineers 28,57 15 455 2147 Mining engineers, metallurgists and related professionals 11,11 4 388 2149 Architects, engineers and related professionals not elsewhere classified 36,67 6 971 2159 Physical sciences technologists 11,11 342 2190 Physical, mathematical and engineering science professionals not elsewhere classified 11,11 76 2210 Scientist 50,00 1 581 2211 Biologists, botanists, zoologists and related professionals 75,00 3 603 2212 Pharmacologists, pathologists and related professionals 16,67 2 097 2213 Agronomists and related professionals 33,33 6 101 2412 Personnel and careers professionals 16,67 54 620 3111 Chemical and physical science technicians 16,67 21 817 3112 Civil engineering technicians 22,22 10 641 3114 Electronics and telecommunications engineering technicians 7,14 69 608 3117 Mining and metallurgical technicians 12,50 3 383 3119 Physical and engineering science technicians not elsewhere classified 40,00 3 112 3151 Building and fire inspectors 11,11 4 398 3152 Safety, health and quality inspectors 16,68 101 189 3211 Life science technicians 5,00 8 159 3212 Agronomy and forestry technicians 16,25 1 617 3213 Farming and forestry advisers 33,33 3 831 3222 Sanitarians 45,00 5 058 3471 Decorators and commercial designers 2,22 46 105 5161 Fire-fighters 33,33 19 234 6141 Forestry workers and loggers 20,00 12 121 6142 Charcoal burners and related workers 20,00 36 6151 Aquatic-life cultivation workers 11,11 317 6152 Inland and coastal waters fishery workers 4,17 3 322 7132 Floor layers and tile setters 25,00 34 631 7134 Insulation workers 33,33 275 7216 Underwater workers 8,33 278 7231 Motor vehicle mechanics and fitters 22,92 201 292 8161 Power-production plant operators 16,67 6 335 8162 Steam-engine and boiler operators 16,67 2 024 8163 Incinerator, water-treatment and related plant operators 50,00 22 455 8232 Plastic-products machine operators 7,14 19 367 9153 Vending-machine money collectors, meter readers and related workers 14,29 13 288 9161 Garbage collectors 79,17 61 023 9321 Assembling laborers 41,67 2 803 Total 879 818 Source: authors’ calculations using PALMS v3.3 data. Employment totals of less than 3000 individuals are estimated from small cells (few observations of less than 100 in an occupation) therefore these totals should be interpreted with caution. 30 Table A2: Broadly green jobs complete list GTI- Total No ISCO-88 code Occupational Title Potential employed 1 1221 Production and operations department managers in agriculture, hunting, forestry and fishing 69.23 9464 2 1222 Production and operations department managers in manufacturing 9.17 47087 3 1311 General managers in agriculture, hunting, forestry and fishing 47.35 21310 4 1312 General managers in manufacturing 9.17 11751 5 2112 Meteorologists 88.89 463 6 2113 Chemists 25.00 1109 7 2114 Geologists and geophysicists 58.33 3048 8 2141 Architects, town and traffic planners 20.83 12629 9 2142 Civil engineers 57.14 15455 10 2144 Electronics and telecommunications engineers 13.39 3882 11 2145 Mechanical engineers 14.29 16116 12 2147 Mining engineers, metallurgists and related professionals 11.11 4388 13 2149 Architects, engineers and related professionals not elsewhere classified 36.67 6971 14 2159 Physical sciences technologists 11.11 342 15 2190 Physical, mathematical and engineering science professionals not elsewhere classified 11.11 76 16 2210 Scientist 62.50 1581 17 2211 Biologists, botanists, zoologists and related professionals 81.25 3603 18 2212 Pharmacologists, pathologists and related professionals 20.83 2097 19 2213 Agronomists and related professionals 100.00 6101 20 2222 Dentists 8.33 6237 21 2412 Personnel and careers professionals 20.00 54620 22 3111 Chemical and physical science technicians 16.67 21817 23 3112 Civil engineering technicians 33.33 10641 24 3113 Electrical engineering technicians 33.33 31598 25 3114 Electronics and telecommunications engineering technicians 24.29 69608 26 3115 Mechanical engineering technicians 12.50 25245 27 3116 Chemical engineering technicians 20.00 1434 28 3117 Mining and metallurgical technicians 25.00 3383 29 3118 Draughtspersons 37.50 13400 30 3119 Physical and engineering science technicians not elsewhere classified 40.00 3112 31 3132 Broadcasting and telecommunications equipment operators 10.00 4759 32 3141 Ships' engineers 40.00 135 33 3142 Ships' deck officers and pilots 12.50 692 34 3145 Air traffic safety technicians 25.00 192 35 3151 Building and fire inspectors 16.67 4398 36 3152 Safety, health and quality inspectors 29.91 101189 37 3211 Life science technicians 5.00 8159 38 3212 Agronomy and forestry technicians 75.00 1617 39 3213 Farming and forestry advisers 100.00 3831 40 3222 Sanitarians 55.00 5058 41 3241 Traditional medicine practitioners 8.33 42474 42 3471 Decorators and commercial designers 2.22 46105 43 4223 Telephone switchboard operators 10.00 70541 44 5112 Transport conductors 10.00 1800 31 45 5161 Fire-fighters 33.33 19234 46 5169 Protective services workers not elsewhere classified 10.00 573030 47 6111 Field crop and vegetable growers 81.82 12789 48 6112 Tree and shrub crop growers 81.82 4138 49 6113 Gardeners, horticultural and nursery growers 38.38 5994 50 6114 Mixed-crop growers 63.64 763 51 6121 Dairy and livestock producers 15.38 10186 52 6122 Poultry producers 16.67 1805 53 6123 Apiarists and sericulturists 66.67 171 54 6130 Market-oriented crop and animal producers 50.00 3678 55 6141 Forestry workers and loggers 80.00 12121 56 6142 Charcoal burners and related workers 80.00 36 57 6151 Aquatic-life cultivation workers 44.44 317 58 6152 Inland and coastal waters fishery workers 22.50 3322 59 6153 Deep-sea fishery workers 12.50 1086 60 6154 Hunters and trappers 20.00 423 61 6210 Subsistence agricultural and fishery workers 15.82 61 62 6211 Subsistence farmers 28.57 10009 63 7111 Miners and quarry workers 5.56 63534 64 7112 Shotfirers and blasters 9.09 5355 65 7113 Stone splitters, cutters and carvers 14.29 6045 66 7114 Diamond drivers 14.29 147 67 7121 Builders, traditional materials 14.29 779 68 7122 Bricklayers and stonemasons 23.81 243612 69 7123 Concrete placers, concrete finishers and related workers 20.00 31728 70 7124 Carpenters and joiners 60.00 98070 71 7129 Building frame and related trades workers not elsewhere classified 4.76 83473 72 7132 Floor layers and tile setters 25.00 34631 73 7134 Insulation workers 33.33 275 74 7136 Plumbers and pipe fitters 20.00 122045 75 7137 Building and related electricians 25.00 143051 76 7211 Metal moulders and coremakers 14.29 1598 77 7212 Welders and flamecutters 12.50 117626 78 7213 Sheet-metal workers 42.86 87238 79 7214 Structural-metal preparers and erectors 16.67 18192 80 7215 Riggers and cable splicers 50.00 1808 81 7216 Underwater workers 25.00 278 82 7221 Blacksmiths, hammer-smiths and forging-press workers 14.29 519 83 7222 Tool-makers and related workers 36.36 8080 84 7224 Metal wheel-grinders, polishers and tool sharpeners 14.29 13649 85 7231 Motor vehicle mechanics and fitters 37.50 201292 86 7232 Aircraft engine mechanics and fitters 40.00 2176 87 7233 Agricultural- or industrial-machinery mechanics and fitters 41.07 90254 88 7241 Electrical mechanics and fitters 57.14 41065 89 7242 Electronics fitters 45.24 16256 90 7243 Electronics mechanics and servicers 45.24 26238 91 7244 Telegraph and telephone installers and servicers 57.14 4349 92 7245 Electrical line installers, repairers and cable jointers 45.24 10459 32 93 7311 Precision-instrument makers and repairers 17.50 2495 94 7312 Musical instrument makers and tuners 9.09 445 95 7313 Jewellery and precious-metal workers 9.09 4074 96 7341 Compositors, typesetters and related workers 5.56 14108 97 7345 Bookbinders and related workers 14.29 1540 98 7346 Silk-screen, block and textile printers 11.11 1181 99 7414 Fruit, vegetable and related preservers 20.00 1357 100 7415 Food and beverage tasters and graders 20.00 2461 101 7422 Cabinet-makers and related workers 16.67 9409 102 7423 Woodworking-machine setters and setter-operators 16.67 748 103 7432 Weavers, knitters and related workers 3.85 3363 104 7433 Tailors, dressmakers and hatters 18.18 30299 105 7434 Furriers and related workers 18.18 104 106 7436 Sewers, embroiderers and related workers 8.33 24965 107 7437 Upholsterers and related workers 16.67 12899 108 7442 Shoe-makers and related workers 38.46 16347 109 8111 Mining-plant operators 5.56 31888 110 8112 Mineral-ore- and stone-processing-plant operators 10.00 19001 111 8113 Well drillers and borers and related workers 10.00 312 112 8161 Power-production plant operators 16.67 6335 113 8162 Steam-engine and boiler operators 16.67 2024 114 8163 Incinerator, water-treatment and related plant operators 50.00 22455 115 8212 Cement and other mineral products machine operators 10.00 7565 116 8231 Rubber-products machine operators 8.33 9003 117 8232 Plastic-products machine operators 14.29 19367 118 8240 Wood-products machine operators 8.33 3087 119 8251 Printing-machine operators 5.56 15614 120 8252 Bookbinding-machine operators 7.14 713 121 8262 Weaving- and knitting-machine operators 3.85 4268 122 8264 Bleaching-, dyeing- and cleaning-machine operators 3.70 4182 123 8265 Fur- and leather-preparing-machine operators 4.55 699 124 8311 Locomotive-engine drivers 20.00 24423 125 8320 Driver, taxi 28.57 188508 126 8322 Car, taxi and van drivers 12.50 147914 127 8323 Bus and tram drivers 28.57 34886 128 8324 Heavy truck and lorry drivers 33.33 250167 129 8331 Motorised farm and forestry plant operators 62.50 51772 130 9141 Building caretakers 25.00 34029 131 9153 Vending-machine money collectors, meter readers and related workers 14.29 13288 132 9161 Garbage collectors 87.50 61023 133 9162 Sweepers and related laborers 26.19 193167 134 9211 Farm-hands and laborers 21.75 957792 135 9212 Forestry laborers 87.50 31848 136 9321 Assembling laborers 50.00 2803 137 9332 Drivers of animal-drawn vehicles and machinery 11.11 37 Total 5096474 Source: authors’ calculations using PALMS v3.3 data. Employment totals of less than 3000 individuals are estimated from small cells (few observations of less than 100 in an occupation) therefore these totals should be interpreted with caution. 33