The SkiLMeeT conference brought together more than 70 researchers and policymakers to examine how rapid digitalisation and the green transition are reshaping Europe’s labour markets: what workers do, how firms are organised, and which skills are rising in value. Held at the Bulgarian Academy of Sciences on 2 – 3 October 2025 in Sofia, the two-day event featured 28 papers across 10 sessions, with a headline keynote by Matias Cortes (York University, Toronto) and policy keynotes by Simone Rosini (DG EMP) and Mariya Mincheva (Bulgarian Industrial Association). Across two days, participants engaged in thematic sessions including skills in the green transition, education and occupational choice, worker mobility, green jobs, firms and inequality, job quality, artificial intelligence, workers coping with technology, and skills & competences.

The conference opened with a keynote by Matias Cortes (York University), who presented “Firm Organization and Worker Outcomes: The Role of Occupational Specialization”. Drawing on Portugal’s matched employer–employee data (2010–2019), he showed that workers in more specialised (“fissured”) firms earn less than the same workers at more occupationally diverse firms – a penalty that persists over individual careers and across occupations. Cortes traced the gap mainly to lower productivity (≈60–70%) and, second, weaker rent-sharing (≈20–30%) in specialised firms, and documented worse five-year career outcomes: slower earnings growth, more firm switching, and a higher risk of non-employment. The keynote can be watched here: xxx

Federico Fabio Frattini (FEEM) opened Session 1 with a paper, “Exploring skills in the green transition: new insights from Italian data world,” co-authored with Irene Brunetti, Martino Kuntze, Andrea Ricci and Francesco Vona. The paper develops a methodological framework to study how skills shape occupational mobility during Italy’s green transition. Using multiple datasets on tasks, skills, and employment transitions, the authors identify 28 General Green Skills (mainly analytical, technical, and monitoring competencies) and 18 General Brown Skills, with substantial overlap between the two groups. They also assess the role of soft skills, finding no significant impact in this context. The study introduces and compares different ways to measure “skill distance” between occupations, showing that greater distance reduces mobility, though transitions toward green jobs mitigate this effect, suggesting that workers tend to move into green occupations closer to their existing skill sets.

Giulia Santangelo (Cedefop) discussed her paper “The Impact of European Climate Actions on Labour-Market Skills,” using Europe-wide online job ads (2018–2023) to track demand for green skills. She introduced two indicators, green pervasiveness (share of ads with at least one green skill) and greenness (green skills as a share of all skills in an ad), to compare regions and sectors. Linking these to ESIF climate-related funds (2014–2020) and using a shift–share IV, she finds that the effect of EU funding on green-skill demand is highly heterogeneous across member states, helping pinpoint leaders and laggards in the transition. The paper was co-authored by Alessia De Santo and Giovanni Mellace.

Daniele Gasparini (Maastricht University) presented “How green are jobs for the energy transition?” He showed that only a minority of the jobs created by planned energy-transition investments qualify as “green occupations,” and that most roles rely more on non-green and digital skills than on narrowly green ones. Using Dutch forecasts of labour demand (MRIO model) linked to O*NET and ESCO skill taxonomies, he compared the “green content” of these occupations and found it small relative to that of classic green jobs, suggesting policy should look beyond creating green jobs to prioritising digital skills and broad reskilling. The paper was co-authored by Jessie Bakens, Peter Mulder, and Nico Pestel.

David Marguerit (LISER) opened Session 2 with “AI and Human Capital Formation: How occupational AI exposure shapes college major selection?” He tracked over 1,100 U.S. fields of study (2010–2022), distinguishing automation-AI from augmentation-AI, and used a shift–share design to isolate causal effects. The results show that in fields linked to occupations exposed to automation-AI, bachelor’s graduations fall, while exposure to augmentation-AI boosts graduations – patterns that are weaker at the master’s level but stronger at the doctorate. The paper was co-authored by Christina Gathmann.

Andreas Fridolin Bühler (University of Zurich) presented “Natural Disasters and the Green Occupational Aspirations of Adolescents.” Using Australian data that link teens’ stated career goals to a task-based “greenness” measure and local disaster records, he showed that disaster exposure is associated with greener occupational aspirations. The effect is stronger for youths who are more environmentally aware or optimistic (measured via PISA), pointing to the importance of awareness-building alongside climate shocks. The paper was co-authored by Patrick Lehnert and Harald Pfeifer.

Karol Madoń (IBS) discussed his paper “Automation and Educational Choice: Evidence from Norway,” co-authored with Piotr Lewandowski. Using student-level data from Norway (2006–2018) and regional data on robot adoption, the study examines whether industrial automation discourages young people from pursuing vocational tracks that lead to acquiring skills increasingly substituted by robots.The preliminary results show that in regions more exposed to robots, students are less likely to choose automation-prone vocational tracks. The effect is strongest among boys and among students whose fathers work in robot-exposed sectors. Lower-achieving students are more likely to select these tracks, revealing negative selection that may widen future inequalities.

In the first policy keynote Simone Rosini (DG EMP) presented “European Skills and Labour Shortages. An Update on the Evidence Available and Policies put in Place at EU Level”. Rosini showed what the latest data says about skills and labour demand in Europe and how EU policy is responding. He argued that today’s broad labour shortages are relatively new, caused by demographics, the speed of green and digital change, and training systems that don’t adapt quickly enough. He walked through the EU response: improve skills intelligence and forecasting so training matches real vacancies; make re- and upskilling shorter and stackable; recognise skills more easily across countries; and attract talent where gaps are structural. He emphasised job quality – pay, conditions, and career paths – as essential to participation and retention. He also highlighted ways to bring in under-used talent (women returning to work, older workers, migrants) through validation of prior learning, apprenticeships that fit SMEs, and better guidance from public employment services. The main message: link evidence to practical actions so people can move into good jobs quickly, with credentials that count across regions. The keynote can be watched here.

Kicking off Session 3, Roy Peijen (TNO) presented “Skill Mismatches and Worker Reallocation During the Green Energy Transition: Scenarios from Fossil Fuel-Intensive Regions in the Netherlands.” Using company workforce data, regional vacancies and ESCO-based skill-similarity, the TNO team simulated closures at Tata Steel IJmuiden and Yara Sluiskil, showing that technicians and electricians could switch jobs with limited training, while metal-rolling/furnace operators, metallurgical engineers and especially chemists face large skill gaps and scarce local demand. The takeaway is that successful reallocation hinges on region-specific demand and targeted re-/upskilling, or else highly specialised workers risk long-term unemployment. The paper was co-authored by Diane Confurius, Joost van Genabeek, and Pascal Kampert.

Lodewijk Visschers (Universidad Carlos III de Madrid) spoke about the paper “Occupation Growth and Worker Flows in Dynamic Labour Market Equilibrium.” He showed new evidence on how workers move into and out of occupations that grow at different speeds, using community detection to flag moves that are easy versus those that face bigger frictions. He also sketched a dynamic general-equilibrium model to quantify how workers’ life-cycles interact with uneven occupation growth, allowing for uncertain growth prospects and possible deterministic occupation life-cycle patterns. The paper was co-authored by Carlos Carrillo-Tudela, Tommaso Santini, and Frank Leenders.

Rumiana Stoilova and Petya Ilieva-Trichkova (IPS-BAS) presented “Digitalisation and Inequality: The Impact on Adult Education Participation across Social Classes and Genders.” Using ESS 2021/22 data for 16 countries merged with the EU’s DESI index, they showed that stronger national digital performance is linked to higher participation in non-formal, job-related adult learning, with frequent internet users especially likely to train. DESI also narrows gaps: women participate more as DESI rises, and differences by class shrink – particularly for lower-grade service, skilled and unskilled workers. They argue that boosting digital infrastructure and skills can expand adult-learning uptake while reducing class and gender inequalities.

Ronald Bachmann (RWI) opened the session with a presentation of a paper, “Wages in the Green Transition,” co-authored by Jessica Wiest, Markus Janser, Christina Vonnahme. The study explores whether environmentally friendly (“green”) jobs offer higher wages, what drives wage differences between green and non-green occupations, and how worker mobility shapes wage trajectories. The findings reveal no clear green wage premium; jobs requiring green skills do not offer significantly higher wages than others. While wages differ across occupations based on their environmental profile (green, brown, or neutral), there is also considerable variation within each group. This could be partly explained by individual characteristics such as gender, education, and the technical nature of tasks, including how routine they are. The lack of a wage incentive may make green jobs less attractive, potentially slowing the green transition. The paper.

Cesar Barreto (OECD) discussed the paper “The ‘Clean Energy Transition’ and the Cost of Job Displacement in Energy-Intensive Industries”. Barreto analysed the costs of job displacement in energy-intensive industries across 14 OECD countries, focusing on sectors such as energy supply, heavy manufacturing, and transport. His findings show that workers displaced from the energy supply and heavy manufacturing sectors face greater earnings losses than those from other sectors, mainly due to lower re-employment wages, job instability, and difficulty in finding new jobs. These outcomes are linked to the characteristics of displaced workers, who are often older, less skilled, and previously employed in high-wage firms, highlighting the challenges of transitioning labour in carbon-intensive industries. The paper was co-authored by Jonas Fluchtmann, Alexander Hijzen, Stefano Lombardi, Patrick Bennett, Antoine Bertheau, Winnie Chan, Andrei Gorshkov, Jonathan Hambur, Nick Johnstone, Benjamin Lochner, Jordy Meekes, Tahsin Mehdi, Balázs Muraközy, Gulnara Nolan, Kjell Salvanes, Oskar Nordström Skans, Rune Vejlin.

Elena Magrini (Lightcast) presented “The Green Transition in Labour Markets Across Europe: An Analysis of Job Postings Data,” showing that demand for green skills is rising across France, Germany, Italy, Spain and the UK, with the highest shares clustered in utilities and construction and strong links to engineering roles. Using Lightcast’s skills taxonomy and 1 billion+ online job ads, the team classifies “green” postings by the presence of environmental skill terms, and finds that climate/environment, energy management and waste management drive most green demand – often alongside digital skills like AutoCAD and automation. The paper was co-authored by Julia Nania, Mauro Pelucchi, Rudra Sett, Will Hanvey, Layla O’Kane, Duncan Brown and Bledi Taska.

Hannah Liepmann (ILO Research Department) discussed the paper “Green Jobs and Green Skills in Emerging Economies: Evidence from Online Vacancy Data.” Using large-scale job ads from Brazil, Uruguay, South Africa, and the Russian Federation, and an ILO-augmented green-skills dictionary, she classified postings by their degree of “greenness” and compared skill profiles and job amenities. The results show that the green transition is reshaping skills demands and job characteristics also in middle-income countries. Yet, the demand for green tasks remains modest. Green vacancies typically require higher skill requirements. Often, green jobs offer higher wages and desirable amenities; however, this is not uniform across all countries or occupational groups. Skills development policies will be crucial for accelerating the green transition and equipping workers with the necessary skills. Policies need to go beyond promoting green skills alone and also foster job quality. The paper was co-authored by Isaure Delaporte and Verónica Escudero.

Mirjam Fitzthum (RWI) presented a paper, “The Regional Dimension of the Green Transition,” co-authored by Markus Janser and Christina Vonnahme. Using Germany-wide administrative worker data linked to a task-based Greenness-of-Jobs Index (GOJI), the study maps county-level greenness from 2012–2022 and finds an overall greening of employment with strong regional variation – especially gains in Mecklenburg-Western Pomerania, Brandenburg, Lower Saxony, and the Ruhr. Early results show greenness correlates with population density and unemployment, and the next step is to quantify how far workers must re/skill, switch occupations, or move regions to avoid wage or job losses. The takeaway: just-transition challenges are region-specific, so support must match local industry structures and worker profiles.

Dimitar Nikoloski (St. Kliment Ohridski University) closed Session 5 with the presentation “Green skills mismatch in North Macedonia: Assessing the difference among sector-specific, cross-sectoral and soft skills.” The study uses a 2025 survey of 530 firms, to measures gaps in green skills across regions, industries and firm sizes. The study finds that older firms tend to experience fewer sector-specific but more cross-sectoral green skills mismatches. Public companies face greater mismatches in both technical and soft skills, while SMEs show lower mismatches across all areas. Integration into global value chains produces mixed effects: foreign investment and exports increase certain mismatches, while imports reduce sector-specific ones but raise cross-sectoral and soft skill gaps. Membership in chambers of commerce appears to widen cross-sectoral mismatches yet ease soft skill ones. In contrast, firms with environmental certifications, corporate social responsibility initiatives, or in-house training capacity display lower green skills mismatches overall. The paper was co-authored by Branimir Jovanovic, Marija Midovska Petkoska, and Dijana Jovanoska.

Opening Session 6, Luisa Braunschweig (IAB) presented “The Role of Industries, Occupations and Firms in the Evolution of Wage Inequality.” To capture the heterogeneity in wage dynamics, the researcher distinguishes worker wages across 272 industrial sectors and four task-based occupational classes, yielding 1020 industry-by-occupation cells that can be consistently observed over the 35-year period. The finding shows that wage inequality increased by 58% between 1985 and 2020. Two-thirds of this increase is between industries and occupations. The paper was co-authored by Johannes Seebauer and Matteo Targa.

Elena Fumagalli (Utrecht University) presented “Willful Ignorance in Hiring,” a study exploring how humans interact with algorithmic decision-making systems and whether they deliberately overlook potential bias when relying on algorithmic recommendations. Together with her colleagues Cielo Cai and Sarah Rezaei, Fumagalli conducted a randomised controlled trial to examine whether people prioritise convenience and financial incentives over verifying fairness and accuracy. In the experiment, participants acted as recruiters, choosing the better candidate from pairs of workers. Their decisions were informed by algorithmic suggestions and additional information about each worker, including demographic characteristics and performance on cognitive tasks. Participants received a higher financial reward for following the algorithm’s recommendation than for overruling it. The study, motivated by EU human-in-the-loop rules, shows how incentive design and information frictions can foster willful ignorance, implying that human review alone may not prevent biased AI-assisted hiring.

Elena Grinza (University of Turin) presented the study “The Productivity Impact of Global Warming: Firm-Level Evidence for Europe” co-authored with Nicola Gagliardi and François Rycx. The paper examines how rising temperatures affect firm productivity across Europe, using longitudinal balance sheet data from private firms in 14 countries, combined with detailed local weather records. The authors find that higher temperatures significantly reduce total factor productivity (TFP), with the strongest effects observed at extreme heat levels. The decline in productivity primarily operates through the labour channel, as labour productivity falls with rising temperatures, while capital productivity remains largely unaffected. The negative impact is most pronounced for firms engaged in outdoor activities such as agriculture and construction, as well as for manufacturing and blue-collar-intensive firms. Geographically, productivity losses are greatest in temperate and Mediterranean regions, showin the need for widespread adaptation measures to mitigate the economic consequences of climate change in Europe.

In our second policy keynote, Mariya Mincheva of the Bulgarian Industrial Association laid out a roadmap for addressing the skills mismatch crisis, which she placed in the context of challenges including demographic shifts, digital disruption and continued perceptions of vocational education as a secondary option. The Bulgarian Industrial Association is calling for a transformation of vocational education to offer flexible learning pathways, enhancing agility through bite-sized programmes, micro-qualifications, and stronger apprenticeships to connect classroom learning to the workplace – all of it backed by a strategic communications campaign to improve the image of vocational education and training. Within this vision, universities also have a role, as skills laboratories, learning hubs and sources of labour market intelligence. The bottom line, Mincheva said, is that there is no competitiveness without skills; skills are the key currency of Europe’s future.

Francesco Berlingieri (European Commission, JRC) presented “Working from home after the pandemic: impact on job quality and well-being.” Using the German Socio-Economic Panel and a difference-in-differences design, he showed that pandemic-induced WFH raised men’s contracted hours but reduced women’s contracted hours while increasing their overtime in 2020–21. By 2022, both men and women working from home report higher job satisfaction, with no clear effects on hourly wages, total hours, or life satisfaction. The paper was co-authored by Sarra Ben Yahmed.

Patrice Jalette (Université de Montréal/CRIMT) presented “How can employers cope in the short term with labour and skills shortages? The case of Québec’s unionized and non-unionized workplaces.” Using interviews, focus groups and a survey of HR/IR managers, the study looks at what firms actually do when vacancies are hard to fill. Early results show employers in both unionized and non-unionized settings adopt similar mixes of measures: boosting attraction and retention, tapping new labour pools, and reorganising work to make better use of current staff. Ongoing analysis will map distinct “bundles” of HR/IR practices by workplace context. The paper was co-authored by Amadou Thierno Diallo.

Mina Kostova (Bulgarian Academy of Sciences) presented “Quality of work and professional development of nurses in Bulgaria”. The study tackles Bulgaria’s acute nursing shortage and asks whether digitalisation can improve job quality and attract new entrants. Drawing on 2025 in-depth interviews within the SkiLMeeT project, the authors frame “quality of work” broadly – pay and conditions, but also skill development, voice, careers, and work–life balance – and examine how technology is reshaping tasks, skills, and work organisation in nursing. The research seeks to translate these insights into practical recommendations for education, employers, and policymakers to support skill upgrading and better retention. The paper was co-authored by Vassil Kirov.

Starting Session 8, Najada Feimi (LISER) presented a study, “Skill Demand in the Age of AI: Evidence from Europe”, co-authored with Christina Gathmann, Terry Gregory, and David Marguerit. The study examines how artificial intelligence (AI) is reshaping employers’ demand for skills across Europe. Using a dataset of 31 million online job advertisements from Belgium, France, Germany, and Luxembourg (2018–2022), the authors apply advanced machine learning models to extract and classify skills in multiple languages and link them to a new measure of occupational AI exposure. The findings show that AI adoption has driven a strong rise in demand for AI and data-related skills, a decline in prediction skills, and stable demand for judgment skills. Moreover, AI exposure increases the need for decision-making and social skills, highlighting the growing importance of human–machine collaboration. Overall, the study shows that AI broadens skill diversity across occupations while deepening specialisation within them, with the strongest effects seen in high-skill and emerging jobs.

Piotr Lewandowski (IBS) discussed the paper “Workers’ Exposure to AI Across Development,” co-authored with Karol Madoń and Albert Park. The paper develops a new, task-adjusted, country-specific measure of workers’ exposure to artificial intelligence (AI) across 103 countries, representing about 86% of global employment. Building on the AI Occupational Exposure Index by Felten et al. (2021), the authors map AI-related abilities to worker-level tasks using international survey data and predict exposure for countries without such data. The study shows that differences in what people actually do at work explain nearly half of the global variation in AI exposure. Workers in richer countries are much more exposed to AI, particularly those in high-skilled, non-routine jobs, while exposure levels are far lower in poorer economies. These differences largely stem from varying levels of technology use, digital infrastructure, and education. The study also finds that AI exposure has risen over the past decade as the nature of work tasks has evolved,

Ulrich Zierahn-Weilage (Utrecht University) ended Session 8 with the presentation of the paper “Expertise at Work: New Technologies, New Skills, and Worker Impacts,” co-authored with Cäcilia vom Baur and Anna Salomons. The paper examines how educational content adapts to advancing technology, focusing on whether training curricula in occupations exposed to digital innovation are updated more frequently and how the skill content of those updates evolves. Using evidence from Germany, the study finds that technology-exposed vocational training programs underwent more frequent revisions and at a faster pace between 1971 and 2021. These updates increasingly emphasise skills that cannot be automated, such as non-routine tasks, digital literacy, and teamwork. The research also shows that curriculum updates have tangible labour market effects: workers who acquire new, technology-relevant skills earn higher wages and remain in their occupations longer, while those whose skills become outdated face a greater risk of obsolescence. In response, firms in technology-intensive sectors increase capital investments to align with evolving skill demands.

Elena Stoykova (Sofia University) presented “Skill Trajectories and Labour Market Integration of Sofia University Sociology Graduates: A 20-Year Longitudinal LinkedIn Analysis.” Using a novel dataset of 400+ alumni profiles, she traced how sociology graduates adapt to the digital – and to a lesser extent green – transition, documenting shifts toward digital skills, new industry paths, and increased international mobility. By comparing LinkedIn-listed skills with the undergraduate curriculum, the study flags skill gaps for non-STEM graduates and calls for curriculum updates and lifelong-learning links to better align social-science training with market demand.

Verónica Escudero (ILO) presented a study, “The Role of Skills in Mediating the Effects of Emerging Digital Technologies on Employment,” co-authored by Isaure Delaporte and Fabien Petit. The study examines how emerging digital technologies impact employment worldwide and identifies the skills that enable workers to benefit from digital transformation. Using data on exposure to 40 digital technologies across 70 countries and detailed skill information derived from online job vacancies, the study finds that digital exposure generally boosts employment, but the effects are modest and uneven across countries, regions, and population groups. The strongest gains are observed in low-income countries, particularly among women, young people, and less-educated workers, while high-income countries experience mixed outcomes, including signs of job polarisation and routine job losses. Skills play a critical mediating role: cognitive and organisational skills amplify employment gains, especially in advanced economies, whereas advanced digital skills, such as machine learning and software development, benefit highly educated workers but can reduce employment prospects for those with lower education. People management skills also enhance employment outcomes, while manual skills are increasingly vulnerable to automation in middle-income countries. The findings highlight the need for targeted, context-specific skill development strategies to ensure a more inclusive digital transition.

Starting Session 10, Vassil Kirov (IBS) presented “Understanding Construction Firms’ Digital Skill Needs: Survey-Based Insights and Predictive Modelling.” Using a novel survey of 264 Bulgarian construction firms, he showed that companies with lower current digitalisation, in urban locations, and with better-educated workforces anticipate broader digital-skill needs, with strong subsector differences (e.g. demolition, environmental/hydro, equipment rental). A firm’s digitalisation level is the strongest predictor of demand for VR, drones and AI; larger firms prioritise database management, smaller firms lean toward 3D printing, and many rural firms foresee rising needs in mobile apps, AI and BIM – implying targeted, sector-aware upskilling policies. The paper was co-authored by Rumiana Jeleva and Svetlomir Zdravkov.

Magdalena Parcheva (University of Economics, Varna) presented “Enhancing Entrepreneurial Competences in the Digital Age – Changing Skills Requirements and Educational Programme Challenges.” She argued that AI and digitalisation are reshaping the entrepreneurial skill set – beyond mindset – toward digital creativity, opportunity recognition from IT, integrating digital tech into business models, and using AI for socially innovative solutions, with platform work expanding self-employment pathways. Drawing on EU skills agendas (2025) and a Varna student case study, she highlighted sizeable skills gaps and called for stronger entrepreneurship education in universities and lifelong learning – built with local innovation ecosystems and updated curricula and training approaches.