Labour markets
in the green economy

Overall, people living in the EBRD regions recognise that climate change is real and has serious consequences. However, those beliefs do not necessarily translate into a willingness to pay higher taxes or forgo economic growth and job creation in order to prioritise environmental policies. Unwillingness to support green policies tends to be greater among the socio-economic groups that are most vulnerable to such changes, including those in the bottom half of the income distribution and the less educated. It is essential that the green transition – the transition to a less carbon-intensive economy – is fair and benefits most members of society, in order to maintain public support for achieving net-zero greenhouse gas emissions.

Major economic transitions of the past, such as the roll-out of digital technologies, the globalisation of trade and investment, and the phasing-out of coal, have three main implications for the green transition.1 First, such transitions entail a reallocation of employment across sectors and industries, as well as a transformation of job requirements. Second, they create substantial opportunities and benefits for workers, accompanied by new risks. And third, their impact varies across geographical areas and demographic groups, which can potentially exacerbate existing disparities.

The green transition cannot take place without an adequately skilled workforce that is able to satisfy the demands of a greener economy. New types of job will emerge as households and producers adopt greener technologies. Demand for workers with green skills is already on the rise, and this chapter documents a wage premium of 4 per cent for such workers. However, that green premium accrues disproportionately to highly skilled workers and less to individuals who are more likely to be climate sceptics.

When faced with technological change, businesses often look to replace workers with machines as they adopt new technologies, a phenomenon known as “labour substitution”. This chapter finds that green innovation generally involves less labour substitution than other types of innovation, but that this is not necessarily the case in the EBRD regions.

Despite the wage premium that green jobs command, workers’ ability to move from “brown” (in other words, polluting) industries to greener sectors remains limited, partly reflecting the inelastic supply of newly required skills in the short term. At the same time, green policies affect different areas and different labour market segments in different ways, potentially upending local labour markets. For instance, the loss of existing jobs in sectors that are associated with the highest emissions (such as the extraction of fossil fuels) is likely to be concentrated in specific regions.

With that in mind, policymakers need to combine localised short-term income support with sector specific long-term workforce development to facilitate labour market adjustments. Labour market programmes that focus on retraining and upskilling can help to ensure that the transition to a greener economy is fair and enjoys broad support, as can regional development initiatives.

In most of the economies surveyed, a large percentage of respondents either agree or strongly agree that climate change is real and are concerned about its consequences (see Chart 3.1). All in all, 63 per cent of surveyed individuals believe that “climate change is real”, with 75 per cent of them believing that “climate change will seriously affect the children of today”. Although figures vary across economies, roughly two-thirds of the people surveyed are concerned about air pollution, the loss of biodiversity, natural disasters and extreme weather, rising temperatures, the spread of disease and environmental pollution. Around half of all respondents are also concerned about a lack of action on climate change.

Although there is broad recognition that the consequences of climate change are severe, willingness to bear the economic cost of the green transition is significantly lower across much of the EBRD regions. For instance, only 43 per cent of surveyed individuals would prioritise the environment at the expense of economic growth and jobs, while between 37 and 43 per cent would pay more tax in order to fight global warming, prevent biodiversity loss or reduce/prevent pollution. In order to take a more detailed look at the tension between people’s beliefs about the environment and their willingness to pay for green policies, an unsupervised machine learning algorithm called “k-means clustering” can be used to identify groups of individuals with similar attitudes towards climate change and environmental issues. This cluster analysis reveals three distinct groups of respondents (see Chart 3.2).

The first group, “supporters” (accounting for 46 per cent of respondents), are concerned about climate change and are largely willing to pay for climate-related policies. The second group, “sceptics” (accounting for 25 per cent of respondents), are less convinced that climate change is a threat and are largely unwilling to pay climate-related taxes. The remaining 29 per cent of respondents are classified as “disengaged” (a group that can be regarded as inconsistent in the sense that 79 per cent of them agree that climate change will seriously affect them during their lifetime, but just 1 per cent are willing to pay more tax in order to fight global warming).

Numbers of disengaged individuals – those who are concerned about the environment, but unwilling to pay for climate policies – are highest in Morocco, Greece, North Macedonia and Jordan, where they account for between 38 and 45 per cent of total respondents (see Chart 3.3).

Regression analysis can then be used to estimate the likelihood of an individual belonging to one of those three groups given their sector of employment, occupation, level of education, income, country of residence and other characteristics. Chart 3.4 shows standardised estimates based on this analysis.

Environmental policies are more likely to be supported by people with a university education, those in the top quartile of the income distribution in their country, and those who prefer a democratic system and a market economy. Personal exposure to natural disasters is also associated with a greater likelihood of supporting climate policies. Climate scepticism is more likely to be found among men, individuals with no past exposure to a natural disaster, those with negative attitudes towards market economics or democracy, and those who attach more weight to present consumption than future consumption (with this “present bias” being inferred from survey questions asking whether an individual would prefer to receive (i) a guaranteed payment of a specific size today or (ii) a much larger guaranteed payment in a month’s time).

People are more likely to be disengaged if they are less educated, are in the bottom half of the income distribution, are female, have personally experienced natural disasters and prefer market economics and democracy. Lower levels of income and education tend to distinguish this group from supporters, while personal experience of natural disasters and greater support for democracy tend to distinguish this group from sceptics.

Top-down approaches classify jobs as green if they are in industries with low levels of emission intensity (in the bottom 20 per cent across all sectors, for instance). All jobs in the sector in question – renewable energy, for example – are considered green, regardless of the specific occupation or the level of skill content. Likewise, sectors can be classified as “brown” if they have high emissions per unit of value added.

In contrast, bottom-up approaches define green jobs on the basis of the skills or tasks involved, irrespective of the sectors that the jobs are in.2 Every occupation is characterised by a mixture of tasks, with those related to the reduction of carbon emissions or improvements in environmental sustainability typically being classified as green. With this kind of approach, a solar energy engineer (a high-skill occupation) and a refuse truck driver (a low-skill occupation) will both be classified as green. These two types of approach complement each other.

In order to construct measures of the greenness of occupations, this section follows in the footsteps of recent literature by taking information on the task content of occupations from the US Department of Labor’s O*NET database. First, the occupations reported by LiTS participants, which were provided as free-text answers, are matched to three-digit ISCO occupation codes. Then, each occupation is matched to the relevant eight-digit classification in O*NET, classifying it as green or non-green using an existing crosswalk.3 Green occupations include physical and earth science professionals, recycling plant technicians and installers of electrical equipment, for example.

Overall, around 27 per cent of LiTS respondents who are in work have green occupations on the basis of this classification. Similar percentages can be observed across the various economies participating in the survey.

On average, workers with green jobs earn 17 per cent more than workers in other occupations (see Chart 3.5), with this wage differential being broadly consistent across the EBRD regions. However, those raw premia do not take account of other determinants of labour income, such as a worker’s qualifications and experience. Regression analysis can be used to check whether wages differ systematically between green jobs and other jobs, while controlling for an individual’s sector of employment, country of residence and other characteristics.

When factors such as gender, age, education, skill level, years of work experience, hours worked, and sector and country of employment are considered, green jobs continue to command a wage premium, but at a much lower level of 4 per cent (see Chart 3.5). The difference between the raw premium and the adjusted premium suggests that, to a substantial extent, workers are sorted into green occupations on the basis of observable characteristics. The adjusted premium is larger for high skilled workers (at 5 per cent) than it is for low-skilled workers, while moderately skilled workers do not enjoy a green wage premium.

The green wage premium suggests that demand for green skills in the market exceeds the supply of workers with the requisite skills. The next section looks at demand for green skills across occupations, leveraging data from LinkedIn through the Development Data Partnership.

In the LinkedIn data, green jobs are defined as occupations that usually involve one or more green projects or are closely associated with green skills. In order to be defined as green, a project must include at least one green activity, such as pollution or waste prevention, energy management, generation of renewable energy, ecosystem management, sustainability education and research, environmental policy, sustainable procurement or environmental finance.5 Green skills, in turn, capture the top skills of LinkedIn members whose work predominantly involves one or more green projects (as listed in those members’ profiles).

Across the EBRD regions as a whole, green jobs accounted for 14 per cent of total paid job postings at the beginning of 2023, up from just under 9 per cent in 2018 (see Chart 3.6). In the advanced comparator economies, this ratio – a measure of demand for green skills – increased more rapidly, rising from 10 to 21 per cent on average over the same period.

The fact that the EBRD regions have lower green jobs ratios than advanced economies may reflect the slower pace of their transition to net zero, as well as an insufficient supply of qualified individuals with green skills. According to LinkedIn data, the percentage of members in the EBRD regions who had a green occupation or listed green skills in their online profile stood at around 10 per cent in April 2023, below the average for the advanced economy comparators (see Chart 3.7). Within individual countries, the accumulation of green skills has accelerated somewhat over the past year, but it does not, overall, appear to have kept pace with the rise in demand.

In every economy, women are less likely to have a green occupation or list green skills in their online profile than men (see Chart 3.8). This may reflect the fact that women are under-represented in STEM jobs, which are particularly common in sectors such as renewable energy. Employment in the renewable energy sector is already estimated to account for half of total energy-related employment worldwide, and it is expected to grow rapidly as the transition to net zero accelerates.6 Thus, the emerging green economy gender gap has the potential to further widen the gender pay gap as the green transition progresses.

These results corroborate broader evidence for the OECD economies, where women are estimated to account for just 28 per cent of all jobs with green tasks.7 The green economy gender gap is typically smaller in the EBRD regions than it is in advanced economies; however, that gap has widened since 2015.

Policy measures with the potential to arrest and reverse that trend include support for girls’ engagement with STEM subjects in schools and career guidance after university. It should also be noted that while men are more likely to have green skills, as captured by LinkedIn data, they are also more likely to be employed in sectors that are home to disappearing brown jobs (that is to say, the sectors that may well end up bearing the brunt of job losses).8

Despite a substantial green wage premium in green sectors, mobility from brown to green sectors has been limited in the EBRD regions, as the analysis in this section shows. Green industries are defined as (i) those that support – directly or indirectly – the generation of renewable energy (for instance, solar, wind and hydroelectric power, or the manufacturing of PV cells), (ii) those with no direct carbon emissions (such as nuclear power), and (iii) those that administer environmental programmes and services. Brown industries are those that produce or support fossil fuels (including coal mining, oil extraction, and the use of fossil fuels to generate electrical power). Data on job-to-job transitions for LinkedIn members are used to document the net flow of workers into green and brown activities, which are classified on the basis of a top-down approach.

LinkedIn data indicate that there was a net flow of workers into green activities between 2015 and 2023 across all economies (that is to say, more workers moved into green industries than moved out of them; see Chart 3.9). Average net rates of transition into green activities were similar across the EBRD regions and advanced comparators, with the highest net transition rate being observed in Estonia. However, a net flow into polluting industries was also observed in the EBRD regions over that period. Thus, more workers found new employment in brown industries than left those industries. In advanced comparators, by contrast, more workers moved out of brown industries than moved into them (that is to say, there was a net outflow).

On average, only around 1 in 20 workers leaving a job in a brown industry immediately find employment in a green industry (see Chart 3.10). This transition rate is almost 1 in 10 in Slovenia, but as low as 1 in 50 in Croatia.

Alternatively, industries can be classified as green or brown on the basis of their average greenhouse gas emissions per unit of value added.9 On that basis, wholesale and retail trade, public administration and various other services can be classified as green, while brown industries mainly comprise manufacturing sectors. Individual-level data from EU Labour Force Surveys can be used to track workers who move out of brown industries, looking to see whether, one year later, they are (i) employed in a green industry, (ii) employed in a non-green industry, (iii) retired, (iv) unemployed or (v) in training.

Over the period 1998-2019, an average of 16 per cent of all workers in the EBRD regions who moved out of brown sector jobs found employment in a green sector within a year (see Chart 3.11). Almost half of all workers who moved out of a brown sector were unemployed a year later, 19 per cent had retired, and only 2 per cent were undergoing retraining or upskilling.

Mobility out of brown industries and into green industries was lower in the EBRD regions than in advanced European economies. In that group of comparator economies, 26 per cent of workers leaving brown-sector jobs found employment in a green sector within a year. Around one in three workers were unemployed a year after leaving their brown-sector job, 14 per cent had retired and 5 per cent were undergoing retraining or upskilling.

Those labour force survey data are combined with information on patents filed with the US Patent and Trademark Office (USPTO) over the period 1996-2019 (which includes the title of each patent and a description of the purpose of the new technology and possible use cases). New green technologies are identified using the Y02 tagging scheme employed by the European Patent Office (EPO), which provides the most comprehensive and widely used classification of low-carbon technologies.10 That scheme covers both mitigation and adaptation across a wide range of areas.

For computational convenience, the USPTO database was queried to retrieve all patents classified as green (418,199 in total) and a random sample of 20 per cent of all non-green patents per year. The largest category of green patents (comprising nearly 140,000 patents) relates to the reduction of greenhouse gas emissions and the generation, transmission or distribution of energy, with the next largest categories relating to mitigation technologies in respect of (i) transport, (ii) the production or processing of goods, (iii) ICT aimed at reducing one’s own energy use and (iv) buildings.

The number of green patents filed per year increased from around 6,000 in 1996 to nearly 30,000 in 2017 as green innovation took off (see Chart 3.12). However, it continues to account for a relatively small share of overall innovation (just 7.7 per cent in 2019) owing to rapid increases in the numbers of patents filed in other fields, such as artificial intelligence. (The drop in patenting activity that can be seen post-2017 in Chart 3.12 is simply an artefact of the data, as patent applications are published with a significant lag.)

In terms of the geography of innovation, lead inventors located in Germany, Japan and the United States account for more than half of all green patents (whereby the first inventor listed on the patent is assumed to be the leader). In contrast, most middle-income economies continue to make a negligible contribution.11 Only 0.3 per cent of all green patents included in this analysis had a lead inventor located in the EBRD regions (with a similar share being observed for non-green patents), with the Czech Republic, Hungary and Poland being home to the most lead innovators for green patents (see Chart 3.13).

This measure shows that some occupations are more exposed to green innovation than others. For instance, installers and repairers of electrical equipment are highly exposed to green innovation, given the high levels of patenting activity in respect of the generation of renewable energy, power storage, electric vehicles and energy-efficient construction. Other occupations with large numbers of green patents as a percentage of all relevant patents include machinery mechanics and repairers, and mining and construction labourers. Workers with such occupations are more likely to be affected by the adoption of green technologies. Micro data from the EU Labour Force Survey can be used to calculate the likelihood of various groups (broken down by gender or place of residence, for instance) being employed in exposed sectors.

On average, men tend to have occupations that are more exposed to green innovation (see Chart 3.14, which shows green patents as a percentage of total innovation relevant to men and women, based on the sectors where they tend to be employed). The gender gap of 1.5 percentage points in 2019 is sizeable relative to the average exposure to green innovation across occupations in that year (8.5 per cent). On the one hand, this implies that if green innovation increases demand for labour and wages, men stand to benefit more than women, at least in the short term. On the other hand, however, it also implies that if new green technologies replace labour, men will be more likely to lose their jobs.

This analysis also reveals that workers with low levels of education are more likely to have occupations with greater exposure to green innovation (see Chart 3.15). This may entail job displacement among lower-skilled workers13 or create opportunities for them. For instance, innovation in the areas of recycling, power generation and grid maintenance is likely to increase demand for occupations that predominantly employ medium or low-skilled workers. Similarly, in construction, the need for energy-efficient buildings could, for instance, increase demand for installers of heat pump boilers, carpenters and joiners, bricklayers, and technicians.14 There are no meaningful differences between urban and rural workers or between different age groups in terms of exposure to green innovation.

In particular, the analysis links local employment growth in a given occupation to the measure of exposure to green innovation and various local labour market characteristics, as well as region-period fixed effects. Employment growth is measured as the change in the logarithm of average employment for a three-digit ISCO occupation in a NUTS-2 subnational region between the four-year periods in question (2000-03, 2004-07, 2008-11, 2012-15 and 2016-19). Occupational exposure to green and non-green innovation is lagged by one period.

The results show that an occupation’s overall exposure to technological innovation tends, on average, to be labour-substituting, in line with previous findings.16 A 1 standard deviation increase in an occupation’s exposure to non-green patents in one period is associated with an 8.9 per cent decline in the number of workers with that occupation in the region in question in the next period (see Table 3.1).

This labour-substituting effect is more muted where exposure to green innovation is greater. Conditional on an occupation’s overall exposure to innovation and controlling for local employment trends, a 1 percentage point increase in green patents’ share of an occupation’s total exposure to innovation in one period is associated with a 3.9 per cent increase in employment for that occupation in the next period. This is a large impact in economic terms, as green patents account for 6.8 per cent of all relevant patents for the average occupation. However, this effect is larger in advanced economies, which are responsible for most green patents. In the EBRD regions, the positive correlation between green patents’ share of total exposure and employment is much smaller and not estimated with statistical precision.

How quickly the demand for these new skills is met in the labour market will, to a large extent, determine the speed at which countries can roll out green technologies. Experience gained from globalisation, automation and the ICT transition suggests that labour does not chase capital (that is to say, workers are not sufficiently mobile), and neither does capital chase labour (that is to say, firms are not sufficiently attracted to areas with lower wages).17 Instead, labour shortages and skills mismatches have the potential to persist as local markets slowly adjust. For instance, in a recent survey of local municipalities across Europe, 69 per cent reported that their lack of environmental and climate assessment skills was a barrier to climate investment.18

This exercise estimates the relative elasticity of the labour supply for green occupations. Green occupations are identified using O*NET’s green job classification, which is based on the task content of each occupation. For non-green occupations, the green exposure measure is zero; for green occupations, it is based on that occupation’s share of the country’s total employment in 2000. Thus, nuclear plant operators are assumed to have greater exposure to green innovation in countries with larger nuclear power industries, and zero exposure in countries with no nuclear plants. This measure also assumes that the increase in demand for an occupation is proportionate to the size of that occupation within the economy, all other things being equal. Regression analysis relates the change in the occupation’s share of the country’s total employment since 2000 to a standardised measure of the occupation’s exposure to green innovation in that country, controlling for country fixed effects (the specificity of the local labour market) and time period fixed effects (general trends in the composition of occupations).

The results indicate that labour markets adjust fairly slowly in the face of a green transition. The employment shares of occupations that had greater exposure to green innovation in 2000 did not experience significant changes during the first decade of this century. That was true in the EBRD regions and advanced comparator economies alike (see Chart 3.16).

Stronger growth in relative employment for occupations with greater exposure to green innovation can be observed from 2015 onwards. By 2019, occupations where exposure to green innovation was 1 standard deviation greater saw extra growth in their employment shares of nearly 1 percentage point in the EBRD regions and 0.8 percentage points in advanced European economies.

There is very little difference between younger workers (age 37 and below) and older workers (age 38 and above) in terms of growth in the employment shares of occupations with greater exposure to green innovation (see Chart 3.17). This suggests that the requisite skills can be picked up through both formal education and on-the-job training.

If the task requirements of a green occupation are similar to those of a non-green occupation and the wages for that green occupation are at least as high, job reallocation should take place more smoothly relative to situations where the task requirements of green and brown jobs are far apart. Using O*NET data on the task content of every occupation, it is possible to calculate a measure of the distance between the tasks of a pair of occupations. Those distances can, in turn, be used to calculate a measure of the task distance between a given occupation and the rest of the economy as an average of all occupation task distances weighted by occupations’ employment shares in a given country.22 Occupations with a shorter distance to the rest of the economy (such as freight handlers) mostly require generic skills, rather than specialist skills. Aquatic life cultivators, in contrast, are characterised by a long distance to the rest of the economy.

In 2000, occupations classified as green had a slightly greater average distance to the rest of the economy relative to non-green occupations. The data on task distances can be incorporated into the analysis of changes in occupations’ employment shares by adding interaction terms that combine exposure to green innovation with dummy variables for each sextile of the task distance distribution.

This analysis suggests that the occupations that are most similar to the rest of the economy in terms of their task content have experienced significantly stronger growth in their employment shares relative to 2000 on account of their exposure to green innovation (see Chart 3.18). The same is also true of the occupations that are least similar to the rest of the economy task-wise, where retraining existing workers is harder but entry rates are probably high as a result of the higher wages associated with task specificity. For occupations with moderate levels of task distance, exposure to green innovation has only had a small – and often statistically insignificant – effect on employment shares.

These findings are consistent with the relative labour supply being less elastic where occupations require tasks that are sufficiently different (although not very different) from those used in the rest of the economy. As green occupations tend to require somewhat more specialist skills than the rest of the economy, this can help to explain the modest pace of labour markets’ adjustment to the roll-out of green technologies.

Where carbon-intensive activities are concentrated in specific local labour markets, communities that are heavily reliant on high-carbon industries may experience more significant job losses. This could give rise to higher local unemployment rates and a loss of social cohesion if adequate regional development plans are not put in place.

This box uses highly granular geographical data on emissions and administrative data from labour-force surveys to identify the localities with the highest concentrations of carbon-intensive manufacturing jobs in the EBRD regions. The data cover Bulgaria, Croatia, the Czech Republic, Estonia, Georgia, Greece, Hungary, Jordan, Kazakhstan, the Kyrgyz Republic, Latvia, Lithuania, Mongolia, Poland, Romania, the Slovak Republic, Türkiye and the West Bank and Gaza (with the choice of economies dictated by the availability of data).

EDGAR carbon dioxide emissions data for manufacturing facilities have been mapped to a grid with 11 km by 11 km cells. The population count for those grid cells has been calculated using the LandScan dataset. On the basis of those data, the analysis in this box focuses on the 11 km by 11 km grid cells that have manufacturing emissions in the top 40 per cent of the total distribution of such emissions.

Each of those locations is characterised by the percentage of the population that is employed in emission-intensive manufacturing sectors. Those percentage shares are approximated using data from the EU Labour Force Survey and six country-level labour-force surveys (see Chart 3.1.1). Manufacturing employment – data on which are only available at a broader, regional level – is assumed to be distributed uniformly across all high-emission locations within that region. Those estimates of manufacturing employment are, in turn, divided by population data for each 11 km by 11 km grid cell as obtained from LandScan. For instance, the city of Győr in the Nyugat-Dunántúl region of Hungary is one of the dots on the map: it is in the 10 per cent of locations with the highest levels of emissions in the EBRD regions, with an estimated 15 per cent of the population employed in emission-intensive industries.

Despite the gradual rise in green jobs’ share of total employment (as documented in this chapter), certain regions did in fact see a rise in the percentage of the population that was employed in carbon-intensive sectors between 2010 and 2018. For instance, provinces in north-western Hungary saw a 3 percentage point rise in the employment share of the manufacturing sector over that period, compared with 1 percentage point increases for construction, wholesale and retail trade, and public administration (the next biggest employers in those regions). A lack of diverse job opportunities, combined with high emission industries dominating the local labour market, can result in a large percentage of the local population continuing to work in carbon-intensive sectors.

Climate change is having a significant impact on agriculture – particularly in rural economies in sub-Saharan Africa – while credit constraints and a lack of information are impeding adaptation in the face of that impact. In such circumstances, migration out of the affected areas could serve as an important adaptation strategy. In particular, weather shocks that harm agricultural yields could drive rural populations into urban agglomerations. At the same time, weather shocks could also discourage migration if they reduce the income that is available to cover expenses associated with relocation.

This box uses rich household-level panel survey data on weather shocks, migration decisions and socio-economic characteristics covering 38,000 unique households in rural regions of Ethiopia, Nigeria, Tanzania and Uganda between 2009 and 2019.31 Those survey data contain information on whether a household has experienced a weather shock, an economic shock (a sudden drop in output prices or a sudden increase in the price of agricultural inputs, for instance) or a political shock (such as local violence) in the past year and whether a member of the household has migrated.

The frequency of weather shocks varies widely from year to year and country to country. In Tanzania, almost 50 per cent of households experienced a weather shock such as drought or flooding in 2009, compared with only 14 per cent of households in 2019. In most countries, economic shocks are more frequent than weather shocks, while political shocks are less frequent. For example, in Nigeria, 30 per cent of households were hit by an economic shock in 2013, compared with 12 per cent that suffered a weather shock and 4 per cent that experienced a political shock. On average, 1 per cent of households per year report a member of the household migrating.

Regression analysis based on data for Ethiopia and Nigeria shows that a weather shock increases the probability of a household member migrating by 0.6 percentage points. The impact of an economic shock is even greater, with the likelihood of migration rising by an estimated 1.6 percentage points. The estimated effect of a political shock is similar to that of a weather shock, but is not statistically significant.

The results also indicate that weather shocks can predict migration in households with a male head, but not in households with a female head. This may reflect the fact that female-led households lack the resources needed to migrate. Among survey respondents, households with a female head had lower average levels of education and income, as well as smaller farms, and suffered more often from food insecurity. Female-led households were also more likely to be single-parent households, making it harder for part of the household to migrate. That relative lack of ability or willingness to respond to weather shocks by migrating has the potential to further exacerbate inequality between male-led and female-led households.

Climate change may exacerbate existing gender disparities in other ways, too. For instance, in response to an adverse economic shock caused by extreme weather, a household might decide to cut back on a girl’s education rather than a boy’s.

The 2015 survey data for Nigeria also include limited information on where migrants go when they migrate. According to those data, most Nigerian migrants stay within the same local government area. Between 2 and 4 per cent go abroad – and of those, half stay on the African continent, while around a quarter move to Europe.

In conclusion, weather shocks do incentivise people to migrate, with most remaining in the same region or country; however, economic shocks are still more important predictors of migration decisions for now. Better-off households are more likely to respond to weather shocks by migrating, as they have the means to do so, while female-led households (which tend to be poorer) are less likely to migrate in response to adverse weather.

Economies in the SEMED region enjoy an advantage when it comes to renewable energy, given their abundant natural resources. In order to take advantage of low-cost green energy, governments need to ramp up their investment in the infrastructure required by renewables, including battery storage, grid upgrades and expansions, water supply, bunkering and hydrogen transport. They also need to address workforce-related challenges by identifying skill shortages and investing in green skills, while retraining, upskilling and redeploying affected workers in brown sectors.

The energy pillar of Egypt’s recently launched Nexus of Water, Food and Energy (NWFE-EP) initiative envisages decommissioning 12 fossil fuel power plants with a total capacity of 5 GW (corresponding to 9 per cent of Egypt’s total installed capacity for fossil fuels). Those plants will be replaced with 10 GW of private solar and wind energy by 2028, which will also be used to produce green hydrogen. That initiative also incorporates an investment plan aimed at scaling up the capacity of the electricity grid and a plan to support workers affected by the decommissioning of the fossil fuel plants. At the same time, the country’s largest supplier of renewable energy, Infinity Energy, is rolling out work-based learning programmes with the support of the EBRD, aiming to provide young people with skills relevant to renewable energy, as well as introducing policies that help women to play a greater role in the company’s workforce.

Meanwhile, the Société Tunisienne de l’Electricité et du Gaz (STEG), Tunisia’s national electricity and gas company, has been using three channels to promote green skills for young people in the energy sector, supported by the EBRD. First, the introduction of national occupational skills standards for key energy-sector occupations (including jobs in renewable energy, data analytics and cybersecurity) will allow companies to base their recruitment on transparent criteria and help to ensure that the skills taught by educational establishments match those required by employers. Second, in cooperation with several local technical universities, STEG has launched an innovative training programme, combining internationally accredited master’s programmes for young engineers with on-the-job learning. And third, a comprehensive gender action plan has been adopted, focusing on promoting women’s access to technical and other STEM roles.