The green potential of occupations in Switzerland

Lobsiger, Michael; Rutzer, Christian Article The green potential of occupations in Switzerland Swiss Journal of Economics and Statistics Provided in Cooperation with: Swiss Society of Economics and Statistics, Zurich Suggested Citation: Lobsiger, Michael; Rutzer, Christian (2021) : The green potential of occupations in Switzerland, Swiss Journal of Economics and Statistics, ISSN 2235-6282, Springer, Heidelberg, Vol. 157, Iss. 1, pp. 1-21, https://doi.org/10.1186/s41937-021-00076-y This Version is available at: https://hdl.handle.net/10419/259772 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 https://doi.org/10.1186/s41937-021-00076-y ORIGINAL ARTICLE The green potential ofoccupations inSwitzerland Michael Lobsiger1* and Christian Rutzer2 Abstract We use a data-driven methodology to quantify the importance of different skills in performing green tasks, aiming to estimate the green potential of occupations in Switzerland. By this we mean the potential of an occupation to be able to perform green tasks on the basis of the skills attributed to it, whereby it is irrelevant whether the occupation already bundles green tasks or not. The results show that occupations with a high green potential are predominantly those with an engineering and technical background. In order to substantiate our green potential measure, we provide evidence of a positive association between demand of employment in occupations with high green potential and an increase in the implicit tax rate on greenhouse gas emissions. The share of employment in occupations with a green potential above a reasonable threshold in the total Swiss labour force is 16.7% (number of persons employed) and 18.8% (full-time equivalents). These employed persons are, on average, younger, more often men, have a higher level of educational attainment and a higher probability of having immigrated than employed persons in occupations with low green potential. Keywords: Green potential, Green transition, Labour market JEL Classification: J23, J24, Q52 © The Author(s) 2021. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. 1 Introduction Environmental awareness has increased in the population in recent years, especially against the background of climate change. These developments have also led to regulatory activities at the political level.1 Strategies and action plans are aimed, among other things, at increasing the resource efficiency of production and achieving a more sustainable economy.2 As recognized by research and practice alike, such a transformation towards a more sustainable economy may also have strong effects on the labour market (e.g. Vona etal., 2018). Specifically, the labour market asks for employees with specific skills (knowledge, abilities, skills in the narrower sense and attitudes) that are necessary to perform green tasks— that is work activities that make a positive contribution to the green economy. These tasks will be increasingly demanded on the path to a more sustainable economy (Rutzer etal., 2020; Janser, 2018). Against this background, it is important to know which Swiss occupations have a high potential to perform green tasks. By this we mean the potential of an occupation to be able to perform green tasks on the basis of the skills attributed to it, whereby it is irrelevant whether the occupation already bundles green tasks or not (Rutzer Open Access Swiss Journal of Economics and Statistics *Correspondence: [email protected] 1 BSS Economic Consultants, Aeschengraben 9, 4051 Basel, Switzerland Full list of author information is available at the end of the article 1 Based on country case studies Strietska-Ilina et al. (2011) identify other drivers of greening economies such as changes in the physical environment, technological development, developments of markets for green products and services, and changing consumer habits. 2 In this paper, green economy and sustainable economy are used as synonyms. Footnote 6 provides a definition of the concept green economy. For Switzerland, the Federal Councils Sustainable Development Strategy 2016– 2019 (Bundesrat, 2016) can be mentioned in this context. The strategy sets out a number of objectives, one of which aims at improving resource efficiency in production. Specific measures to achieve better resource efficiency are set out in the 2013 Green Economy Action Plan and its development for the period 2016–2019 (BAFU, 2013, 2016). Footnote 2 (Continued) Page 2 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 etal., 2020). An electrical engineer, for example, can put her skills to work both overseeing a nuclear reactor at a nuclear power plant or researching renewable energy. Only in the second case does she perform a green task, but in any case she brings the potential to perform green tasks. Therefore, information about the green potential of occupations seems central to us, e.g. for education and training policy: The need to align education and training with green transformation will be less urgent if many workers are employed in occupations with high green potential. This is because one can assume that, to a certain extent, they have the necessary skills to carry out green tasks and are thus well prepared for a green transformation. In order to quantify the green potential of occupations in Switzerland, we follow a novel approach developed by Rutzer et al. (2020). The approach relies on machine-learning algorithms to quantify the importance of different skills in performing green tasks. The results show that occupations with a high green potential are predominantly those with an engineering and technical background. Once the green potential of different Swiss occupations is determined, we aim at finding empirical support for our measure of the green potential of occupations. In particular, we follow a recent strand of the literature examining heterogeneous demand responses in labour markets (measured at the occupational level) with respect to changes in environmental policy stringency (Niggli & Rutzer, 2020; Vona et al., 2018). In particular, we investigate whether occupations with high green potential are in higher demand relative to other occupations when environmental policy stringency proxied by green emission taxes increases. The results provide evidence of a positive association between demand of occupations with high green potential and an increase in the implicit emission tax. Although we do not claim a causal relationship, we interpret this result as empirical support for the estimated green potential of occupations in Switzerland. Next, we use our green potential estimates at the occupational level to determine the proportion of employees in Switzerland working in occupations considered to have a high green potential. Furthermore, we describe their socio-economic background. In particular, we characterize employment in occupations with high green potential along different dimensions such as age, sex, migration status and level of education as well as a set of labour market indicators (rate of job vacancies, rate of unemployment). For that, we divide the occupations into two groups based on a particular threshold for the green potential (which we will justify later). According to our estimates, in Switzerland in the year 2017, 739,000 persons and 670,000 full-time equivalents (FTE) have been employed in occupations with high green potential. Measured relative to the total number of persons employed and FTE, this is around 16.7% and 18.8%, respectively. Employed persons in occupations with high green potential are, on average, younger, more often men, have a higher level of educational attainment and a higher probability of having immigrated than employed persons in other occupations. Moreover, the group of occupations with high green potential has a lower unemployment rate and a higher job vacancy rate. The rest of the paper is structured as follows: Next, we review the literature (1) assessing the extent of greenness of occupations based on task level information, (2) relating the extent of greenness of occupations to skills and (3) analysing heterogeneous demand responses of the labour market due to green economy measures. Afterwards, we describe our methodology to estimate the green potential of occupations and the estimates thereof. We then outline our empirical strategy to analyse possible heterogeneous labour market effects in response to changes in environmental policy stringency and provide the respective estimation results. Consequently, we characterize employment in occupations with high green potential along different dimensions. Finally, we summarize our main findings and conclude. 2 Literature Our paper can be attributed to the growing literature analysing the impact of a green economy on the labour market by applying a task-based approach. This approach is conceptually related to a rich literature that analyses how labour market outcomes (such as employment and wages) are shaped by skills and tasks (Acemoglu & Autor, 2011).3 In particular, our approach to estimating the green potential of occupations is based on two strands of the literature: First, there is the literature that assesses the extent of greenness of occupations based on task level information. Consoli etal. (2016) use occupation and task level information provided by O*NET4 to classify occupations as green or non-green in order to elaborate on the differences between both occupational groups in terms of skill content and human capital. As Bowen etal. (2018) point out, there is considerable heterogeneity among the occupations considered as green with respect to the 3 According to Acemoglu and Autor (2011,p. 1045), “[...] a task is a unit of work activity that produces output (goods and services). In contrast, a skill is a worker’s endowment of capabilities for performing various tasks. Workers apply their skill endowments to tasks in exchange for wages, and skills applied to tasks produce output.” 4 The Occupational Information Network (O*NET) is a US American database that holds various job-related information. More information about O*NET will follow in Sect.3. Page 3 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 “greenness”—that is the use of green tasks within occupations. Again relying on O*NET, Vona etal. (2018) exploit this variation in the use of green tasks within occupations to compute a continuous greenness measure for each occupation. An alternative approach to measure the greenness of occupations based on tasks has been developed by Janser (2018). Based on job requirements stated in the German occupations database BERUFENET and applying a text mining approach, they identified green tasks in Germany. On this basis, they computed a greenness-of-jobs index per occupation by relating the identified green tasks (that is job requirements that contain key words form a pre-defined green tasks dictionary) to all job requirements.5 This approach is interesting because it shows a way to identify green tasks apart from using the O*NET database. It has the advantage that countryspecific occupational information databases can be used for task identification and do not require a diversion via O*NET. However, identifying green tasks is only one step towards estimating the green potential of occupations. In order to estimate the potential for all occupations (including those without green tasks), detailed occupation-specific information on skills is necessary. This is where the second strand of the literature on which our approach is based comes in, namely the research relating the extent of greenness of occupations to skills. Vona etal. (2018) use the continuous greenness measure (see above) to assess the importance of green skills for the exercise of green tasks. Their analysis is based on O*NET that provides, apart from information about green tasks within occupations, detailed information about skills on the level of occupations. A recent approach by Rutzer etal. (2020) starts here and offers a method to estimate the green potential of occupations based on skills in a continuous way. Rutzer etal. (2020) define the green potential of an occupation based on the skills which are required to perform green tasks. It is not important whether an occupation bundles green tasks or not, but whether the skills needed to perform those work activities would in principle allow green tasks to be performed. Essentially, the idea is that a set of skills can be applied to different (green or non-green) tasks. In this respect, there are no green skills, but skills that are better suited than others to exercise green tasks. While following Vona etal. (2018) in terms measuring the greenness of occupations and using information of skills provided by O*NET, Rutzer etal. (2020) use a different estimation approach that delivers (from a statistical point of view) more accurate predictions of the green potential of occupations. In particular, the method of Rutzer etal. (2020) relies on machine learning and not on a principal component analysis on the top of OLS estimations as Vona etal. (2018) do. The statistical superiority of the former method is probably mainly due to the fact that it also makes it possible to use information of skills that are highly needed for non-green tasks. In contrast, Vona etal. (2018) only exploits information from skills that are important for green tasks. The concept of green potential contrasts strongly with current methods of identifying so-called green jobs and determining employment in these occupations on this basis. A uniform definition of green jobs is missing (Bowen etal., 2018; Janser, 2018; Esposito etal., 2017). The literature therefore provides different definitions based on industry affiliation or the production methods used (Martinez-Fernandez etal. 2010; Consoli etal., 2016; Bowen etal., 2018). A frequently used approach to identify green jobs starts at the industry level and identifies those sectors that produce goods and services that contribute to the protection of the environment or the conservation of natural resources. In particular, the efforts of statistical offices to define the so called Environmental Goods and Services Sector (EGSS) and describe it in terms of employment and value added are to be mentioned here (Eurostat, 2016; ILO, 2018). This approach has, however, some drawbacks: As noted in ILO (2018) and Esposito etal. (2017), it neglects jobs that improve production processes with respect to their environmental impact, irrespective of the goods that are produced. Furthermore, it does not shed light on the skills necessary to carry out an activity that is expected to contribute to a sustainable economy and therefore says nothing in terms of an occupation’s potential to bundle green tasks. But precisely this information is of interest to education policy and practice in order to, for example, gear training courses to the needs of a sustainable economy. While there are estimates of employment in green industries in Switzerland (according to the definition of the EGSS, covering employment in industries that produce goods and services contributing to the protection of the environment or the conservation of natural resources), there are no estimates based on the concept of green potential. Here we make a contribution by applying the approach of Rutzer etal. (2020) to determine the green potential of Swiss occupations. To the best of our knowledge, a database that allows both the identification of green tasks and provides information on skills at the occupational level is not available in Switzerland. For this reason, we will follow Rutzer etal. (2020) and rely our analysis on O*NET data. 5 Janser (2018) differentiates between core and additional job requirements, the former being essential to the practice of the profession. In 2016, 190 job requirements (2.6% of all job requirements) were green tasks. 19.9% of all occupations (according to the Klassifikation der Berufe 2010—KldB2010) contained at least one of these 190 green tasks. Page 4 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 Once the green potential of different occupations is determined, we aim to find empirical support for our measure of the green potential of occupations. For that, we follow a recent strand of the literature examining heterogeneous demand responses in labour markets (measured at the occupational level) with respect to changes in environmental policy stringency (Niggli & Rutzer, 2020; Vona etal., 2018). The execution of occupations with high green potential requires skills that allow to perform green tasks. A central hypothesis is that these tasks and the skills associated with them will be in greater demand in case of a green shift of an economy (e.g. Vona etal., 2018). In such a case, one would accordingly expect a higher demand for occupations that require these tasks and skills compared to other occupations: On the one hand, occupations that already bundle green tasks benefit from the increased demand due to environmental policy becoming more stringent. On the other hand, occupations that do not yet bundle green tasks, but already require skills that are necessary to perform these tasks, are predestined to include green tasks in their activity bundle as a result of a development towards a green economy (Niggli & Rutzer, 2020). Niggli and Rutzer (2020) provide a recent empirical contribution with respect to heterogeneous labour market effects on the occupational level. They estimate the green potential of occupations for 19 European countries following Rutzer et al. (2020). On this basis, they then analyse the effect of increased environmental policy stringency on the occupation-level manufacturing employment for the period 1992 to 2010. In line with other literature (Vona etal., 2018; Marin & Vona, 2019), they provide evidence for heterogeneous employment changes in response to an increase in environmental policy stringency. In particular, they document a decrease in labour demand for occupations with relatively low green potential and an increase in labour demand for occupations with relatively high green potential. We add to this literature by providing further evidence of heterogeneous labour demand responses associated with increased greenhouse gas emission taxes as a proxy for environmental policy stringency, using Switzerland as an example. After this brief review of the literature, the next section deals with the methodological approach and the data we use to estimate the green potential of occupations in Switzerland. 3 Quantifying andtesting thegreen potential ofoccupations In the following, we first provide detailed information about the measurement of the green potential of occupations in Switzerland and the respective results. Afterwards, we inform about the methodological approach to analyse heterogeneous labour market effects to green economy measures and describe the data needed for this analysis. Finally, we provide the respective results. 3.1 Measurement ofthegreen potential ofoccupations The measurement of the green potential of occupations in Switzerland draws on the recent contribution of Rutzer etal. (2020). This work estimates the green potential of occupations in the US labour market, that is, for occupations classified according to the Standard Occupational Classification (SOC). This approach can be adapted so it can be used to estimate the green potential of occupations in Switzerland (and other countries) relying on a different classification system for occupations, namely the International Standard Classification of Occupations (ISCO). The following section briefly discusses this approach and highlights the modifications that need to be made to apply it to Swiss data. Here we follow the approach of Niggli and Rutzer (2020) and refer to this article for a detailed description of the procedure. The estimates of Rutzer etal. (2020) of the green potential for SOC occupations are based on data from O*NET. This database provides information about the tasks and skills contained in occupations and classifies tasks as green or non-green. The classification of tasks according to green or non-green is carried out by experts. A green task is a work activity that makes a positive contribution to the green economy6 (e.g. the development of a method for measuring water quality). Table 1 provides four examples of different O*NET occupations. For the solar photovoltaic installer, O*NET lists a total of 26 tasks, where 26 tasks are classified as “green new tasks”. The environmental engineer bundles a total of 28 tasks, three of which are green new tasks and 25 that are existing green tasks. An example of an “existing green task” is designing or supervising of the design of systems, processes, or equipment for control, management, or remediation of water, air, or soil quality. An example of a “new green task” is writing reports or articles for Web sites or newsletters related to environmental engineering issues. Architects have seven new green tasks and 18 6 According to Dierdorff etal. (2009,p. 3), “[t]he green economy encompasses the economic activity related to reducing the use of fossil fuels, decreasing pollution and greenhouse gas emissions, increasing the efficiency of energy usage, recycling materials, and developing and adopting renewable sources of energy.” Page 5 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 non-green tasks. Designing or constructing plans of green buildings projects to minimize adverse environmental impact or conserve energy is an example of a “new green task”. Tasks such as preparing scale drawings or architectural designs, using computer-aided design or other tools that has no obvious impact on the green economy are examples of “non-green tasks”. 21 tasks can be attributed to the electrician occupation, none of which are considered green. On that basis Rutzer etal. (2020) calculate the greenness of an occupation i as ηi=#green tasksi/#total tasksi . In the examples provided by Table1 (last column), two occupations have a η=1.0 (solar photovoltaic installer, environmental engineer), one a η=0.3 (architect) and one a η=0 (electrician). It is important to note that there are occupations that do not perform any green tasks, that is jobs with η=0 , but demand skills similar to occupations that perform green tasks. While these occupations do not bundle green tasks, they nevertheless have the potential to perform green tasks based on the skills required. The calculation of this potential is what is at the heart of the measurement strategy used in this analysis. For example, it is not a-priori clear that electricians that perform, according to the O*NET classification, no green tasks (cf. Table1) have a low green potential. Indeed, as the calculations later will show, electricians are assigned to an occupational group that has a relatively high green potential, because it is associated with skills that are similar to occupations that perform green tasks. In other words, employees that work as electricians are thought to be equipped with skills that are a prerequisite to perform green tasks. Besides information about tasks, O*NET provides detailed information about job-specific skills for every occupation. Following Vona et al. (2018), skills are understood as knowledge, skills (in the narrower sense) and work activities.7 O*NET provides, for each occupation, a quantitative rating for the importance ( IM ) and the level ( LV ) of a total of 114 skills. For each skill s , this information is aggregated into a single value by applying a weighting scheme of importance and level. Following Rutzer etal. (2020), the value of a skill s of occupation i is calculated as skilli,s =IMα i,s LV 1−α i,s with α=0.7 . Afterwards, the value of each skilli,s is normalized between 0 and 1 by computing (skilli,s−min(skills))/(max(skills)−min(skills)) . Using machine-learning algorithms, it is possible to predict the potential of SOC occupations to perform green tasks on a continuous scale. For that, a prediction model is trained to predict the greenness of an occupation depending on the values of different skills using data from O*NET. According to the analysis of Rutzer etal. (2020), the Ridge algorithm performs best compared to alternative estimation models (OLS, Lasso, Random forest) in terms of prediction quality (measured by the mean squared error and the multiclass receiver operating characteristic curve (Mroc) on a holdout dataset).8 For this reason, the estimation of the green potential for SOC occupations is based on the Ridge regression model. The model uses the information on skills described above as explanatory variables and the greenness ηi of an occupation i , that is the share of green tasks within the total number of tasks, as the dependent variable. The result is a coefficient ˆ βridge s for each skill s , which can be used to predict the green potential ˆηi of an occupation i : Table 1 Examples of green jobs and tasks according to O*NET Source: O*NET Title # of tasks Green Other Greenness η New Existing Solar photovoltaic installer (47-2231.00) 26 0 0 1.0 Environmental engineer (17-2081.00) 3 25 0 1.0 Architect (17-1011.00) 7 0 18 0.3 Electrician (47-2111.00) 0 0 21 0.0 7 It is important to note that there is no uniform definition of skills in the literature (OECD, 2017). We stick to the definition used by O*NET as we will base our analyses strongly on this database. For detailed information on the database, see https:// www. onetc enter. org/ datab ase. html# indiv idualfiles. The following definitions were taken from https:// www. onetc enter. org/ repor ts/ Relat ed. html (Volume I: Report): Knowledge is defined as a “[...] collection of discrete but related and original facts, information, and principles about a certain domain that is acquired through education, training, or experience” (such as “Administration and Management” or “Design”), skills (basic and cross-functional) is defined as “[...] capabilities of individuals that are acquired through experience and practice, and are used to facilitate knowledge acquisition” (such as “Mathematics” or “Writing”), work activities is defined as an “[...] aggregation of similar job activities/behaviors that underlie the accom- 8 In comparison to OLS, the Ridge regression endogenously shrinks some coefficients towards zero in order to reduce overfitting to the training data. plishment of major work functions” (such as “Analyzing Data or Information” or “Developing and Building Teams”). Footnote 7 (Continued) Page 6 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 Table 6 in the “Appendix 1” shows, for each considered skill s , the coefficient ˆ βridge s . On the one side, there are skills associated with a high predicted greenness as “Building and Construction”, “Geography” and “Physics” with coefficients of 0.15 and 0.08, respectively. On the other side are skills such as “Support”, “Fine Arts” and “Foreign Language” that are associated with a low predicted greenness carrying the coefficients of −0.13, −0.09 and −0.08, respectively. When interpreting the coefficients, however, it is important to note that they are not unbiased, since the Ridge regression shrinks coefficients of skills with low prediction power towards zero. O*NET provides information on the value of a skill s of occupation i ( skilli,s ) according to the SOC on an 8-digit level. In order to estimate the green potential for occupations according to the ISCO 08 on the 3-digit level based on Eq.(1), the values of the skills must accordingly be transferred. Here we follow Niggli and Rutzer (2020) (and the work cited therein) and carry out the following three steps: 1. The values of a particular skill s of all occupations i belonging to the same 6-digit SOC occupation are transferred to that 6-digit level SOC occupation by taking a simple (unweighted) average. 2. The skilli,s is then transferred to ISCO occupations, where a conversion table (provided by the US Bureau of Labor Statistics (BLS)) informs about the relations between SOC occupations on the 6-digit level and ISCO occupations on the 4-digit level. In the case of multiple matches, a simple (unweighted) average is used. The transfer of job information prepared for the US labour market to the European context in this second step can be seen as a limitation of this approach. However, following research of other scientists (e.g. OECD, 2017), we assume that this transfer is by and large permissible. Firstly, environmental issues arise not only in Switzerland, but internationally. Secondly, we assume that a skilled worker in both the USA and Switzerland must, with a few exceptions (e.g. in terms of nomenclature), has the same skills in order to be able to carry out the tasks assigned to her. For example, electricians must have the skills to estimate the necessary material quantities on the basis of planning documents and to install switches and sockets, no matter where they work. 3. The skilli,s for ISCO occupations on the 3-digit level are computed by taking simple (unweighted) averages of all values of a particular skill s of all 4-digit occupations that belong to a 3-digit occupation. (1) ˆ η i=ˆ β0+ p  s=1 skilli,sˆ βridge s . On that basis, it is then possible to apply the trained model (see Eq. 1) to predict the potential of 3-digit ISCO occupations to perform green tasks. Finally, we normalize the green potential of occupations on a scale between 0 and 1, whereby the value of 1 (0) is attributed to the occupation with the highest (lowest) estimated green potential. The values for the other occupations i are then calculated relative to the values of the occupations with the highest and lowest green potential: (ˆηi−min(ˆη))/(max(ˆη) −min(ˆη)) . A summary of the results is shown in Table2. It lists the ten occupations with the highest green potential estimates. As one can see, Engineering professionals have the highest green potential with a value of 1, followed by Physical and earth science professionals with a value of 0.76 and Production managers in agriculture, forestry and fisheries with a value of 0.76. The green potential of all occupations used in this study are shown in Table7 in the “Appendix”. 3.2 Emission tax, green potential andemployment In the previous section, we determined the green potential of ISCO occupations. In doing so, we argued that such occupations have skills needed to perform green tasks. Therefore, in case of a green shift of an economy one would expect a relatively high demand for such jobs compared to jobs that do not have such skills. In the following, we would like to examine whether we find signs of such a heterogeneous demand response in the Swiss labour market. Specifically, we use changes in the implicit tax rate on greenhouse gas emissions at a two-digit industry level to investigate whether this is associated with a heterogeneous change in the demand for workers with higher green potential relative to those with lower green potential. For this purpose, we estimate the following empirical model where our left-hand side variable log empi,j,t consists of the log of employment of occupation i in industry j at time t, τj,t stands for the implicit emission tax of industry j at time t. We interact the implicit emission tax with our previous determined green potential level ˆηi of an occupation i in order to analyse possible heterogeneous labour market responses. In addition, the matrix Xj,t contains control variables, such as the non-interacted implicit emission tax τj,t and the labour productivity at the industry level. In addition, we use time fixed effects ( δt ), occupational fixed effects ( ǫi ) and industry fixed effects ( vj ).9 For example, the fixed effects take into (2) log empi,j,t=β 1 ˆη i τ j,t + γ X j,t +δ t +ǫ i +v j +u i,j,t , 9 Our specification does not include a term ˆ β 3 ˆ ηi , because it is completely captured by the occupational fixed effect ǫi . Page 7 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 account unobserved heterogeneity among different ISCO occupations that may affect the level of employment and industry-specific employment differences. Finally, ui,j,t is the error term. To summarize, our estimation strategy relies on analysing variations in the number of employed persons over time within industry-specific occupational groups that differ according to their green potential. It is important to note that our specification only shows an association and not a causal relationship. Since the aim of this analysis is to empirically support our measure of the green potential of occupations, we consider establishing a controlled correlation as sufficient. 3.3 Data Next, we provide an overview of the data used to estimate equation (2). Our analysis is based on different data sources: Firstly, we use data from the O*NET database (v21.2). We have already described it in Chapter3.1 when discussing the measurement of the green potential of occupations. Secondly, for the number of employed persons we use data from the Swiss Labour Force Survey (SLFS) for the years 2008–2017. The SLFS provides information on the structure of the labour force and employment behaviour patterns. Specifically, in addition to general socio-economic information, the SLFS contains information on the job currently exercised, according to the ISCO nomenclature and the industry where the person is employed, which corresponds to the six digit NOGA-2008 nomenclature.10 Thirdly, we use data on expenditures on greenhouse gas emissions (measured in millions of Swiss Francs) and greenhouse gas emissions (measured in thousand tons of CO2 equivalent emissions) from the Federal Statistical Office (FSO). This information is available at the two digit NOGA-2008 industry level—and for some industries for a grouping of several two digit industries—from 2008 onwards and allows us, for each industry and year, to calculate an implicit tax rate by dividing the environmental expenditures through the greenhouse gas emissions. Afterwards, we multiply the result by 1000 to get an implicit emission tax measured as Swiss Franc per one ton of CO2 equivalent emission. A list of industries and their corresponding change in the implicit emission tax between our first year 2008 and last year 2017 can be found in the “Appendix 3”. In total, we have data for 37 different industries of the second (manufacturing) and third sector (services). Fourthly, we use data from the FSO to calculate the labour productivity at the industry level (value added at the industry level divided by the number of employment). Having outlined the data, we now turn to the empirical results. 3.4 Empirical results Table3 shows the main results. The first row captures whether there exists an heterogeneous association between implicit emission tax and labour demand. The variable is highly significant with a positive sign in all of our four specifications either above or almost at the 99% level. In particular, the first column shows our baseline specification. This specification controls only for general economic conditions and time-invariant occupation- and industry specific effects, such as the routine intensity of occupations. Therefore, we add yearindustry fixed effects and, additionally, year-occupation fixed effects to capture time varying effects at the most disaggregated level. As one can see in columns (2) and (3), the coefficient remains very stable. Note that, in this case, the implicit emission tax without interaction is completely absorbed by the fixed effects. In a last step, Table 2 Occupations with highest green potential Own calculations based on Rutzer etal. (2020) ISCO Occupation Green potential 214 Engineering professionals (excluding electrotechnology) 1.00 211 Physical and earth science professionals 0.76 131 Production managers in agriculture, forestry and fisheries 0.76 210 Science and engineering professionals, nos 0.75 312 Mining, manufacturing and construction supervisors 0.75 215 Electrotechnology engineers 0.73 132 Manufacturing, mining, construction, and distribution managers 0.72 216 Architects, planners, surveyors and designers 0.71 112 Managing directors and chief executives 0.68 314 Life science technicians and related associate professionals 0.68 ... ... ... 10 The NOGA nomenclature is identical to the NACE classification up to a four-digit level. Page 8 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 we use the labour productivity at the industry level and 3-digit ISCO time trends to control for major structural changes such as digitization and the increasing globalization. Again, the coefficient of our variable of interest remains almost the same (column 4). Moreover, in order to check the robustness of our results, we varied the time period, excluded the most and least polluting industries and also the most and least affected ISCO occupations. The results can be found in “Appendix 4” and generally confirm the heterogeneous association demonstrated in Table3. So far, the estimates are based on a continuum of green potential. However, for our subsequent descriptive characterization of employment in occupations with high green potential, it is necessary to divide occupations into discrete groups. For this purpose, we divide the ISCO occupations into two groups based on their green potential. In particular, we classify all ISCO occupations with a green potential larger or equal to 0.5 as occupations with high green potential and the rest as occupations with low green potential. The choice of the threshold is further explained in Sect.4.1, but can also be justified empirically. For this purpose, Table4 again shows the estimations of our previously determined baseline specification. But instead of considering a continuum of green potential, jobs are now divided into two groups with high and low green potential, respectively.11 In the first column, we interact the implicit emission tax with the binary green potential dummy. The results show a highly significant positive association between an increase in the implicit emission tax and demand for occupations of the high green potential group. Instead, for the occupation group with low green potential, the association is insignificant. Thus, our previous results based on a continuum of green potential also transfers to a binary case.12 In columns (2) and (3), we further analyse whether there exists some within-group heterogeneity. In both cases, the coefficient of the implicit emission Table 3 Association between Implicit Emission Tax and Employment The sample is an unbalanced panel covering 37 Swiss industries between 2008 and 2017. The dependent variable in all columns is the log of occupational employment. All model specifications include fixed effects for industries and years. Columns (1), (2) and (3) further include occupation fixed effects. Columns (1) and (4) additionally include the implicit emission tax. Columns (2) and (3) additionally contain yearly occupation and industry fixed effects. Furthermore, column (4) contains the labour productivity at the industry level in logarithmic form and occupation time trends. Data are from the FSO and the SLFS. Standard errors in parentheses are clustered at the industry level. Significance levels for the coefficients are indicated as: ∗p<0.1 ; ∗∗ p<0.05 ; ∗∗∗ p<0.01 Dependent variable log( employmenti,j,t ) (1) (2) (3) (4) Implicit emission taxj,t × 0.0626 ∗∗ 0.0649 ∗∗∗ 0.0655 ∗∗∗ 0.0573 ∗∗ Green potential ˆηi (0.0245) (0.0247) (0.0251) (0.0240) Implicit emission taxj,t   Occupation fixed effects    Industry fixed effects     Year dummies     Year-industry fixed effects   Year-occupation fixed effects  log(labour productivity)  3-digit ISCO time trends  Observations 22,642 22,642 22,642 22,642 Pseudo- R2 0.0691 0.0698 0.0422 0.0695 Table 4 Association between Implicit Emission Tax and Employment: Binary case The sample is an unbalanced panel covering 37 Swiss industries between 2008 and 2017. The dependent variable in all columns is the log of occupational employment. All model specifications include fixed effects for occupations, industries and years. Columns (2), (3) and (4) additionally include the implicit emission tax. It is not included in column (1) as otherwise one of the two green potential group dummies are fully absorbed. Data are from the FSO and the SLFS. Standard errors in parentheses are clustered at the industry level. Significance levels for the coefficients are indicated as: ∗p <0.1; ∗∗ p<0.05 ; ∗∗∗ p<0.01 Dependent variable: log( employmenti,j,t ) (1) (2) (3) (4) Subset of green potential: Full sample < 0.5 ≥ 0.5 Full sample Implicit emission taxj,t × η ≥ 0.5 0.0381 ∗∗∗ (0.0147) Implicit emission taxj,t × η<0.5 0.0071 (0.0106) Implicit emission taxj,t × 0.0368 0.00402 0.0626 ∗∗ Green potential ˆηi (0.0407) (0.0350) (0.0245) Implicit emission taxj,t    Occupation fixed effects     Industry fixed effects     Year dummies     Observations 22,642 17,580 5062 22,642 Pseudo- R2 0.0693 0.0696 0.1052 0.0691 11 Due to reasons of high collinearity between the binary green potential dummy and time varying occupational fixed effects, we stick to the baseline specification. 12 In Table 10 in the “Appendix”, we show additional estimations for a binary grouping based on a threshold of 0.45 and 0.55, which we also use as robustness checks in our subsequent descriptive characterization of the green potential of the Swiss labour market. The results remain stable. Page 15 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 Appendix2: Green potential ofISCO occupations See Table7 Table 6 (continued) General skill from O*NET Ridge coefficient Therapy and Counseling − 0.05 Thinking Creatively − 0.02 Time Management − 0.03 Training and Teaching Others − 0.03 Transportation 0.01 Troubleshooting 0.02 Updating and Using Relevant Knowledge − 0.01 Working Conditions 0.05 Writing 0.04 Table 7 List of ISCO occupations and their predicted green potential ISCO Job title Predicted greenness 214 Engineering professionals (excluding electrotechnology) 1.00 211 Physical and earth science professionals 0.76 131 Production managers in agriculture, forestry and fisheries 0.76 210 Science and engineering professionals, nos 0.75 312 Mining, manufacturing and construction supervisors 0.75 215 Electrotechnology engineers 0.73 132 Manufacturing, mining, construction, and distribution managers 0.72 216 Architects, planners, surveyors and designers 0.71 112 Managing directors and chief executives 0.68 314 Life science technicians and related associate professionals 0.68 110 Chief executives, senior officials and legislators, nos 0.65 111 Legislators and senior officials 0.60 213 Life science professionals 0.59 311 Physical and engineering science technicians 0.58 142 Retail and wholesale trade managers 0.57 130 Production and specialized services managers, nos 0.54 242 Administration professionals 0.53 100 Managers, nos 0.53 143 Other services managers 0.53 741 Electrical equipment installers and repairers 0.51 740 Electrical and electronic trades workers, nos 0.50 310 Science and engineering associate professionals, nos 0.50 122 Sales, marketing and development managers 0.49 212 Mathematicians, actuaries and statisticians 0.49 313 Process control technicians 0.49 120 Administrative and commercial managers, nos 0.48 742 Electronics and telecommunications installers and repairers 0.48 754 Other craft and related workers 0.48 Page 16 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 Table 7 (continued) ISCO Job title Predicted greenness 723 Machinery mechanics and repairers 0.46 140 Hospitality, retail and other services managers, nos 0.46 710 Building and related trades workers, excluding electricians, nos 0.45 711 Building frame and related trades workers 0.45 333 Business services agents 0.44 731 Handicraft workers 0.44 712 Building finishers and related trades workers 0.44 332 Sales and purchasing agents and brokers 0.43 133 Information and communications technology service managers 0.43 622 Fishery workers, hunters and trappers 0.43 240 Business and administration professionals, nos 0.43 612 Animal producers 0.43 121 Business services and administration managers 0.43 931 Mining and construction labourers 0.41 620 Market-oriented skilled forestry, fishery and hunting workers, nos 0.41 251 Software and applications developers and analysts 0.41 343 Artistic, cultural and culinary associate professionals 0.40 811 Mining and mineral processing plant operators 0.39 720 Metal, machinery and related trades workers, nos 0.39 610 Market-oriented skilled agricultural workers, nos 0.39 200 Professionals, nos 0.37 250 Information and communications technology professionals, nos 0.37 600 Skilled agricultural, forestry and fishery workers, nos 0.36 243 Sales, marketing and public relations professionals 0.36 813 Chemical and photographic products plant and machine operators 0.35 613 Mixed crop and animal producers 0.35 700 Craft and related trades workers, nos 0.35 252 Database and network professionals 0.35 621 Forestry and related workers 0.34 241 Finance professionals 0.34 721 Sheet and structural metal workers, moulders and welders, and related workers 0.34 315 Ship and aircraft controllers and technicians 0.33 611 Market gardeners and crop growers 0.33 634 Subsistence fishers, hunters, trappers and gatherers 0.33 713 Painters, building structure cleaners and related trades workers 0.32 300 Technicians and associate professionals, nos 0.32 835 Ships’ deck crews and related workers 0.31 722 Blacksmiths, toolmakers and related trades workers 0.31 752 Wood treaters, cabinet-makers and related trades workers 0.31 352 Telecommunications and broadcasting technicians 0.30 134 Professional services managers 0.30 933 Transport and storage labourers 0.29 350 Information and communications technicians, nos 0.29 232 Vocational education teachers 0.28 141 Hotel and restaurant managers 0.28 330 Business and administration associate professionals, nos 0.28 225 Veterinarians 0.27 730 Handicraft and printing workers, nos 0.27 961 Refuse workers 0.27 Page 17 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 Table 7 (continued) ISCO Job title Predicted greenness 821 Assemblers 0.27 351 Information and communications technology operations and user support technicians 0.27 834 Mobile plant operators 0.27 930 Labourers in mining, construction, manufacturing and transport, nos 0.27 261 Legal professionals 0.25 335 Regulatory government associate professionals 0.25 820 Assemblers, nos 0.24 331 Financial and mathematical associate professionals 0.24 750 Food processing, wood working, garment and other craft and related trades workers, nos 0.24 522 Shop salespersons 0.24 262 Librarians, archivists and curators 0.24 231 University and higher education teachers 0.23 632 Subsistence livestock farmers 0.23 830 Drivers and mobile plant operators, nos 0.23 411 General office clerks 0.23 630 Subsistence farmers, fishers, hunters and gatherers, nos 0.23 800 Plant and machine operators and assemblers, nos 0.23 221 Medical doctors 0.23 810 Stationary plant and machine operators, nos 0.22 921 Agricultural, forestry and fishery labourers 0.22 833 Heavy truck and bus drivers 0.22 751 Food processing and related trades workers 0.21 950 Street and related sales and service workers, nos 0.21 951 Street and related service workers 0.21 952 Street vendors (excluding food) 0.21 521 Street and market salespersons 0.21 960 Refuse workers and other elementary workers, nos 0.21 900 Elementary occupations, nos 0.21 340 Legal, social, cultural and related associate professionals, nos 0.20 520 Sales workers, nos 0.20 541 Protective services workers 0.20 814 Rubber, plastic and paper products machine operators 0.20 220 Health professionals, nos 0.20 223 Traditional and complementary medicine professionals 0.20 260 Legal, social and cultural professionals, nos 0.19 732 Printing trades workers 0.19 633 Subsistence mixed crop and livestock farmers 0.19 812 Metal processing and finishing plant operators 0.19 226 Other health professionals 0.18 816 Food and related products machine operators 0.18 832 Car, van and motorcycle drivers 0.18 524 Other sales workers 0.18 818 Other stationary plant and machine operators 0.18 513 Waiters and bartenders 0.18 430 Numerical and material recording clerks, nos 0.17 831 Locomotive engine drivers and related workers 0.17 514 Hairdressers, beauticians and related workers 0.17 512 Cooks 0.17 Page 18 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 Table 7 (continued) ISCO Job title Predicted greenness 500 Service and sales workers, nos 0.17 432 Material-recording and transport clerks 0.16 941 Food preparation assistants 0.16 815 Textile, fur and leather products machine operators 0.16 631 Subsistence crop farmers 0.15 932 Manufacturing labourers 0.15 265 Creative and performing artists 0.15 753 Garment and related trades workers 0.15 817 Wood processing and papermaking plant operators 0.15 342 Sports and fitness workers 0.15 222 Nursing and midwifery professionals 0.15 431 Numerical clerks 0.15 912 Vehicle, window, laundry and other hand cleaning workers 0.14 510 Personal service workers, nos 0.14 910 Cleaners and helpers, nos 0.14 224 Paramedical practitioners 0.14 264 Authors, journalists and linguists 0.14 320 Health associate professionals, nos 0.13 324 Veterinary technicians and assistants 0.13 230 Teaching professionals, nos 0.13 321 Medical and pharmaceutical technicians 0.13 233 Secondary education teachers 0.12 235 Other teaching professionals 0.12 516 Other personal services workers 0.12 511 Travel attendants, conductors and guides 0.11 400 Clerical support workers, nos 0.11 323 Traditional and complementary medicine associate professionals 0.11 410 General and keyboard clerks, nos 0.10 911 Domestic, hotel and office cleaners and helpers 0.10 515 Building and housekeeping supervisors 0.10 325 Other health associate professionals 0.10 523 Cashiers and ticket clerks 0.09 441 Other clerical support workers 0.09 263 Social and religious professionals 0.09 234 Primary school and early childhood teachers 0.08 420 Customer services clerks, nos 0.08 412 Secretaries (general) 0.07 422 Client information workers 0.07 413 Keyboard operators 0.06 962 Other elementary workers 0.06 531 Child care workers and teachers’ aides 0.04 421 Tellers, money collectors and related clerks 0.04 341 Legal, social and religious associate professionals 0.04 530 Personal care workers, nos 0.02 322 Nursing and midwifery associate professionals 0.02 334 Administrative and specialized secretaries 0.01 532 Personal care workers in health services 0.00 Page 19 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 Appendix3: Industries used intheempirical analysis Industry Observations τ2017/τ2008 D10–T12 1028 0.91 D13–T15 1550 1.07 D16–T18 972 0.92 D19–T20 873 0.99 D21 881 1.06 D22–T23 1140 0.77 D24–T25 1579 0.71 D26 1018 1.40 D27 749 1.30 D28 1221 1.16 D29–T30 1054 1.19 D31–T33 1548 0.96 D35 705 2.13 D36–T39 781 1.89 D41–T43 777 2.00 D45 1251 0.58 D46 985 0.77 D47 911 0.69 D49–T51 1903 0.60 D52 1721 0.60 D53 252 0.60 D55–T56 1563 0.23 D58–T60 1062 0.47 D61 390 0.22 D62–T63 1083 0.63 D64 1128 0.06 D65 1088 0.64 D68 350 1.03 D69–T70 1483 0.94 D71 738 0.94 D72 886 1.11 D73–T75 2065 0.40 D77–T82 2069 0.60 D85 2175 0.69 D87–T88 944 0.64 D90–T93 2033 2.13 D94–T96 1436 1.11 The first column shows the Noga nomenclature. The second column states the number of distinct ISCO occupations per year and industry. The third column shows the incremental change in the industry’s implicit tax rate between 2017 and 2008. A value less than one means that the implicit tax rate has decreased, a value larger than one that it has increased Appendix4: Robustness checks See Tables8, 9 and 10. Table 8 Association between Implicit Emission Tax and Employment: Varying considered occupations The sample is an unbalanced panel covering 37 Swiss industries between 2008 and 2017. The dependent variable in all columns is the log of occupational employment. All model specifications include fixed effects for occupations, industries and years. In addition, yearly industry and occupation fixed effects are considered. Column (1) excludes the 1% of observations having the largest value of the variable of interest implicit emission tax interacted with the green potential and column (2) the 1% with the lowest value. Column (3) shows the full sample for comparison. Data are from the FSO and the SLFS. Standard errors in parentheses are clustered at the industry level. Significance levels for the coefficients are indicated as: ∗p<0.1 ; ∗∗ p<0.05 ; ∗∗∗ p<0.01 Dependent variable log( employmenti,j,t ) (1) (2) (3) Excluded 1% Excluded 1% Full sample max τη min τη Implicit emission taxj,t × 0.0677 ∗∗ 0.0645 ∗∗ 0.0655 ∗∗ Green Potential ˆηi (0.0343) (0.0271) (0.0251) Occupation fixed effects    Industry fixed effects    Year dummies    Year-industry fixed effects    Year-occupation fixed effects    Observations 22,301 22,281 22,642 Pseudo- R2 0.0404 0.0411 0.0422 Page 20 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 Abbreviations SLFS: Swiss labour force survey; SOC: Standard occupational classification system; ISCO: International standard classification of occupations; FTE: Full-time equivalents; EGSS: Environmental goods and services sector. Acknowledgements We are grateful to Rolf Weder, Matthias Niggli, Wolfram Kägi and Christopher Huddleston for comments and suggestions. Comments from two anonymous reviewers and the editor contributed significantly to sharpening the content of the paper. All errors remain our own responsibility. Author contributions ML carried out the descriptive analyses to describe the green potential in Switzerland. He was also responsible for the writing of the texts. CR performed the estimation of the green potential using data from O*NET and conducted the regressions. All authors read and approved the final manuscript. Funding This work has been supported by the Swiss National Science Foundation (SNSF) within the framework of the National Research Program Sustainable Economy: resource-friendly, future-oriented, innovative (NRP 73), Grant-No 407340-172430. Availability of data and materials Data from O*NET (U.S. Department of Labor, Employment and Training Administration ) can be accessed via https:// www. onetc enter. org/ db_ relea ses. html. Data from the SLFS was made available by the OFS after signing a data usage contract. The data can be requested from the OFS by stating the purpose of use. Table 9 Association between implicit emission tax and employment: varying time span and industries The dependent variable in all columns is the log of occupational employment. All model specifications include fixed effects for occupations, industries and years. In addition, yearly industry and occupation fixed effects are considered. Column (1) shows an unbalanced panel covering 37 Swiss industries between 2008 and 2012 and column (2) between 2013 and 2017. Column (3) contains a subset of the 15 industries exhibiting the largest incremental increase in the implicit tax rate between 2008 and 2017 and column (4) a subset of the 15 industries with the lowest incremental increase between 2008 and 2017. Data are from the FSO and the SLFS. Standard errors in parentheses are clustered at the industry level. Significance levels for the coefficients are indicated as: ∗p<0.1 ; ∗∗ p<0.05 ; ∗∗∗ p<0.01 Dependent variable log( employmenti,j,t ) (1) (2) (3) (4) 2008– 2013– 15 industries 15 industries 2012 2017 Highest �τ2008−2017 Lowest �τ2008−2017 Implicit emission taxj,t × 0.0668 ∗∗∗ 0.0645 ∗∗ 0.0927 ∗∗∗ 0.0586 Green potential ˆηi (0.0254) (0.0259) (0.0360) (0.0544) Occupation fixed effects     Industry fixed effects     Year dummies     Year-industry fixed effects     Year-occupation fixed effects     Observations 11,059 11,583 7616 7926 Pseudo- R2 0.0377 0.0378 0.0196 0.0044 Table 10 Association between implicit emission tax and employment: alternative binary cases The sample is an unbalanced panel covering 37 Swiss industries between 2008 and 2017. The dependent variable in all columns is the log of occupational employment. All model specifications include fixed effects for occupations, industries and years. The implicit emission tax is not included as otherwise one of the two green potential group dummies are fully absorbed. Data are from the FSO and the SLFS. Standard errors in parentheses are clustered at the industry level. Significance levels for the coefficients are indicated as: ∗p<0.1 ; ∗∗ p<0.05 ; ∗∗∗ p<0.01 Dependent variable log( employmenti,j,t ) (1) (2) (3) x=0.5 x=0.45 x=0.55 Implicit emission taxj,t × η≥x 0.0381 ∗∗∗ 0.0343 ∗∗∗ 0.0419 ∗∗∗ (0.0147) (0.0140) (0.0105) Implicit emission taxj,t × η<x 0.0071 0.0050 0.0080 (0.0106) (0.0105) (0.0154) Occupation fixed effects    Industry fixed effects    Year dummies    Observations 22,642 22,642 22,642 Pseudo- R2 0.0693 0.0694 0.0692 Page 21 of 21 Lobsigerand Rutzer Swiss J Economics Statistics (2021) 157:8 Declarations Competing interests The authors declare that they have no competing interests. Author details 1 BSS Economic Consultants, Aeschengraben 9, 4051 Basel, Switzerland. 2 Faculty of Business and Economics, University of Basel, Peter Merian-Weg 6, 4052 Basel, Switzerland. Received: 30 November 2020 Accepted: 1 October 2021 References Acemoglu, D., & Autor, D. (2011). Skills, tasks and technologies: Implications for employment and earnings. In Handbook of labor economics (Vol. 4, pp. 1043–1171). Elsevier. BAFU. (2016). Bericht an den bundesrat, grüne wirtschaft: Massnahmen des bundes für eine ressourcenschonende, zukunfstfähige schweiz. Bern. Retrieved from https:// www. bafu. admin. ch/ bafu/ de/ home/ themen/ wirts chaftkonsum/ fachi nform ation en/ massn ahmendes- bundesfuer- resso urcen schon ung. html. BAFU. (2016). Bericht an den bundesrat, grüne wirtschaft: Massnahmen des bundes für eine ressourcenschonende, zukunfstfähige schweiz. Bern. Retrieved from https:// www. bafu. admin. ch/ bafu/ de/ home/ themen/ wirts chaftkonsum/ fachi nform ation en/ massn ahmendes bundesfuer- resso urcen schon ung. html. Bowen, A., Kuralbayeva, K., & Tipoe, E. L. (2018). Characterising green employment: The impacts of ‘greening’ on workforce composition. Energy Economics, 72, 263–275. Bundesrat. (2016). Strategie nachhaltige entwicklung 2016–2019. Retrieved from https:// www. are. admin. ch/ are/ de/ home/ nachh altigeentwi cklung. html. Bundesrat. (2018). Schlussbericht zur fachkräfteinitative. Retrieved from https:// www. seco. admin. ch/ seco/ de/ home/ Arbeit/ Fachk raeft einit iative. html. Consoli, D., Marin, G., Marzucchi, A., & Vona, F. (2016). Do green jobs differ from non-green jobs in terms of skills and human capital? Research Policy, 45(5), 1046–1060. Degen, K., Ragni, T., Bieri, D., & Marti, S. (2016). Fachkräftemangel in der schweiz: Indikatorensystem zur beurteilung der fachkräftenachfragel. Bern: Staatssekretariat für Wirtschaft SECO. Retrieved from https:// www. seco. admin. ch/ seco/ de/ home/ wirts chaft slage--- wirts chaft spoli tik/ wirts chaft spoli tik arbei tsmar kt/ fachk raeft einit iative. html Dierdorff, E. C., Norton, J. J., Drewes, D. W., Kroustalis, C. M., Rivkin, D., & Lewis, P. (2009). Greening of the world of work: Implications for o*netsoc and new and emerging occupations. The National Center for O*NET Development. Esposito, M., Haider, A., Samaan, D., & Semmler, W. (2017). Enhancing job creation through green transformation. In T. Altenburg & C. Assmann (Eds.), Green industrial policy: Concept, policies, country experiences (pp. 51–67). Geneva and Bonn: UN Environment and German Development Institute/ Deutsches Institut für Entwicklungspolitik). Eurostat. (2016). Environmental goods and services sector accounts—Practical guide—2016 edition [Computer software manual]. Retrieved from https:// ec. europa. eu/ euros tat/ web/ produ ctsmanua ls- and- guide lines/-/ KSGQ- 16- 011. https:// doi. org/ 10. 2785/ 688181. ILO. (2018). Greening with jobs—World employment and social outlook 2018. International Labour Office. Retrieved from https:// www. ilo. org/ global/ resea rch/ globalrepor ts/ weso/ green ingwith- jobs/ lang-- en/ index. htm. Janser, M. (2018). The greening of jobs in germany: First evidence from a text mining based index and employment register data (Tech. Rep.). IAB- Discussion Paper. Marin, G., & Vona, F. (2019). Climate policies and skill-biased employment dynamics: Evidence from eu countries. Journal of Environmental Economics and Management, 98, 102253. Martinez-Fernandez, C., Hinojosa, C., & Miranda, G. (2010). Green jobs and skills: The local labour market implications of addressing climate change. Working document, CFE/LEED, OECD. Retrieved from www. oecd. org/ datao ecd/ 54/ 43/ 44683 169. pdf? conte ntId= 44683 170. Niggli, M., & Rutzer, C. (2020). Environmental Policy and Heterogeneous Labor Market Effects: Evidence from Europe (Working paper No. 2020/09). Faculty of Business and Economics—University of Basel. Retrieved from https:// ideas. repec. org/p/ bsl/ wpaper/ 2020- 09. html. OECD. (2017). Getting skills right: Skills for jobs indicators. OECD Publishing, Paris. Retrieved from https:// doi. org/ 10. 1787/ 97892 64277 878- en. Rutzer, C., Niggli, M., & Weder, R. (2020). Estimating the Green Potential of Occupations: A New Approach Applied to the U.S. Labor Market (Working paper No. 2020/03). Faculty of Business and Economics—University of Basel. https:// ideas. repec. org/p/ bsl/ wpaper/ 2020- 03. html. Strietska-Ilina, O., Hofmann, C., Haro, M. D., & Jeon, S. (2011). Skills for green jobs: A global view. synthesis report based on 21 country studies. International Labour Office, Skills and Employability Department, Job Creation and Enterprise Development Department - Geneva: ILO. Vona, F., Marin, G., Consoli, D., & Popp, D. (2018). Environmental regulation and green skills: An empirical exploration. Journal of the Association of Environmental and Resource Economists, 5(4), 713–753.