OIT - Skills dynamics in the Arab region - New evidence from online vacancy data

 ILO Brief 1 Skills dynamics in the Arab region: New evidence from online vacancy data  Regional Research Brief 08/12/2025 Skills dynamics in the Arab region: New evidence from online vacancy data Sama El Hage Sleiman, Hannah Liepmann, and Tia Mokdad1

 Combining big vacancy data from UNESCWA’s Skills Monitor and the ILO’s skills taxonomy developed for such data, this brief analyses skills dynamics in Egypt, Jordan, and the United Arab Emirates (UAE).  In the online markets analysed, enterprises demand a range of socio-emotional and cognitive skills and fewer manual skills. Skills are demanded in bundles, combining core and sophisticated cognitive skills, for example, with socio-emotional skills.  Wage analysis for the UAE suggests a positive association of wages with cognitive skills but not with manual and socio-emotional skills (except for people management skills). Socio-emotional skills are nevertheless important foundational skills.  In the three countries, there have been shifts towards employment in information and communication technology and service-oriented roles, suggesting opportunities for private-sector growth.  Such growth opportunities need to be paired with adequate supply-side measures to train workers and reduce skills mismatch.

 1. Introduction Structural transformations, including those related to technological change, demographic shifts and the green transition, make it important to have a good understanding of skills dynamics in the labour market. Such an analysis informs how job demands are c hanging as well as which

1 This brief was written as part of a collaboration between the International Labour Organization (ILO) and the United Nations Economic and Social Commission for Western Asia (UNESCWA), as an input to the 2026 ILO Report on Lifelong Learning and Skills for the Future. It was made possible thanks to substantial contributions of colleagues from both institutions. Salim Araji (UNESCWA) and Verónica Escudero (ILO) initiated a nd coordinated the

Commission for Western Asia (UNESCWA), as an input to the 2026 ILO Report on Lifelong Learning and Skills for the Future. It was made possible thanks to substantial contributions of colleagues from both institutions. Salim Araji (UNESCWA) and Verónica Escudero (ILO) initiated a nd coordinated the collaboration, in addition to providing inputs to the brief. Willian Boschetti A damczyk, Maksym Hermez, and Le Wu were responsible for data science. Ludovica Gerbino, Elvire Jegu and María Moreno Barrera provided expert research assis tance. Simon Böhmer, Isaure Delaporte, Roy Doumit, Caroline Fredrickson, Aya Jafaar, Amal Mowafy, and Jad Yassin contributed excellent feedback. type of skills development should be fostered to increase the resilience of workers and enterprises facing the structural transformations. In this context, a rich literature has examined skills dynamics, especially in the United States. This literature has documented marked changes in skills demand even within granular occupations (Atalay et al. 2020). Moreover, this literature shifted the emphasis to complementary skills that are understood as skills bundles. Socio-emotional skills combined with technical skills were Key points ILO Brief 2 Skills dynamics in the Arab region: New evidence from online vacancy data in this context found to improve labour market outcomes (Borghans et al. 2014; Deming 2017; Deming and Kahn 2018; Weinberger 2014). However, skills dynamics may not be the same in other regions with fundamentally different labour market features. For example, several countries in the Arab region face challenges of labour underutilization, growing working-age populations, and political instabilities. Other countries, especially the Gulf Cooperation Countries, experience labour market segmentations, with nation als predominantly working in the public sector and nonnationals in the private sector. Discrepancies in labour market outcomes between women and men, younger and older workers, and workers in urban and rural areas are

countries, especially the Gulf Cooperation Countries, experience labour market segmentations, with nation als predominantly working in the public sector and nonnationals in the private sector. Discrepancies in labour market outcomes between women and men, younger and older workers, and workers in urban and rural areas are moreover pronounced in the region. Finally, there are supply-side concerns regarding the adequacy of education and skills development initiatives (ESCWA and ILO 2021; ILO 2024; 2025b). All of these developments highlight the importance of a good understanding of skills demand and the availability of suitable data that allows monitoring skills dynamics to inform policy decisions. To respond to this need, as part of its Future of Work Initiative, the United Nations Economic and Social Commission for Western Asia (ESCWA) created the Skills Monitor, which collects real -time vacancy data from online platforms (ESCWA 2025) . This initiative is motivated by the understanding that the needed shift towards greater economic diversification of the private sector will only be possible in the region if the demand and supply of skills evolve accordingly (ESCWA 2024). Against this background, the present brief analyses Skills Monitor data from three countries with distinct features: Egypt, Jordan, and the United Arab Emirates (UAE). To distil skills variables from these data that are suitable for research purposes, the brief employs the skills classification and natural language processing (NLP) techniques developed by the ILO (Adamczyk et al. 2025; Escudero, Liepmann, and Podjanin 2024) for the 2026 ILO Report on Lifelong Learning and Skills for the Future (ILO Forthcoming). Key findings from this brief will feature in the report, where they will be discussed in comparison to findings from countries of other regions. On this basis, the brief addresses the following questions:

2 This question will be answered for the UAE alone, which was the only country with readily available wage data of good quality.

report, where they will be discussed in comparison to findings from countries of other regions. On this basis, the brief addresses the following questions:

2 This question will be answered for the UAE alone, which was the only country with readily available wage data of good quality. ● What are the observed patterns in skills demanded by firms in Egypt, Jordan, and the UAE? ● How are skills interconnected? ● Which skills are associated with higher wages2? The brief proceeds as follows. Section 2 describes the methodology. Section 3 analyses skills demand, followed by an analysis of skills interconnectedness (Section 4) and of the relationship between skills and wages in the UAE (Section 5). Section 6 discus ses the results, while Section 7 presents recommendations.

 2. Methodology This brief focuses on Egypt, Jordan, and the UAE. As illustrated by selected economic, demographic, and digital indicators in Table 1, the three countries have distinct features, which makes it interesting to study these countries’ skills dynamics in comparison: The GDP per capita figures reflect that t he UAE is a high -income economy belonging to the Gulf Cooperation Council, whereas Egypt and Jordan are lower -middle income economies. While Jordan and the UAE share similar population sizes, Egypt has by far the largest population of the three countries. Finally, the employment-to-population ratio and the share of the population using the internet are highest in the UAE. The data source of this brief are online vacancy data, which provide granular information on the tasks sought in vacancies and enable the identification of country -specific trends in skills demand . In many countries, including the three countries studied, such analyses are particularly informative given that it would otherwise not be possible to study skills dynamics due to the absence of longitudinal survey or expert -derived data on skills. While online vacancy data are not fully representative of national labour markets and tend to overrepresent jobs requiring higher

three countries studied, such analyses are particularly informative given that it would otherwise not be possible to study skills dynamics due to the absence of longitudinal survey or expert -derived data on skills. While online vacancy data are not fully representative of national labour markets and tend to overrepresent jobs requiring higher levels of qualification, Escudero, Liepmann, and Podjanin (2024) show for Uruguay that jobs requiring medium and even low levels of formal qualifications are reasonably captured by vacancy data. Nevertheless, vacancies that are posted offline or advertised through personal contacts only are not captured in online vacancy data and thus not acc- ILO Brief 3 Skills dynamics in the Arab region: New evidence from online vacancy data  Table 1: Key economic, demographic, and digital indicators for Egypt, Jordan, and the United Arab Emirates Indicator Egypt Jordan United Arab Emirates GDP per capita (current US$) 3,338.5 (2024) 4,618.1 (2024) 49,377.6 (2024) Population (in thousands) 116,538 (2024) 11,553 (2024) 11,027 (2024) Employment-to-population ratio (%) 45.1 (2024) 33.5 (2023) 79.8 (2024) Share of population using the Internet (%) 72.7 (2023) 92.5 (2023) 100.0 (2024) Note: The table reports the latest available data for each indicator and country.

Sources: GDP per capita: World Bank (2025), total population: UN DESA (2024), employment-to-population ratio: ILO (2025a), share of population using the internet: International Telecommunication Union (2025). The data were accessed in October 2025. counted for in this brief (see Fabo and Kureková 2022 for a detailed discussion of the representativeness of online vacancy data). As such, the brief provides evidence for the online labour markets of the three countries, yielding

counted for in this brief (see Fabo and Kureková 2022 for a detailed discussion of the representativeness of online vacancy data). As such, the brief provides evidence for the online labour markets of the three countries, yielding insights especially into growing and future -oriented occupations and sectors. To systematically identify and classify skills within the online datasets, this brief employs a new skills taxonomy that the ILO developed for the forthcoming 2026 Report on Lifelong Learning and Skills for the Future. The taxonomy categorizes skills into three overarching groups: cognitive, socio-emotional, and manual. It is further divided into 15 subcategories, with keywords derived from academic literature from economics (e.g., Deming and Kahn 2018; Deming and Noray 2020; Hershbein and Kahn 2018) and psychology (e.g., Almlund et al. 2011) , along with complementary analyses. The skills subcategories are defined through unique keywords that can be mapped into online vacancy data. The taxonomy’s theoretical foundation makes it more comprehensive than existing classifications and applicable to different countries’ realities. To ensure comprehensiveness and measurability, the taxonomy largely focuses on technical and non -technical skills that are transferable and thus applicable across occupations, rather than on skills that are specific to the tasks of an occupation. Transferable skills are crucial because they tend to be more enduring and adaptable in the face of a fast-changing w orld of work. For more details on the taxonomy, see Escudero, Liepmann, and Podjanin (2024) and Adamczyk et al. (2025). The ILO’s taxonomy can be applied to online vacancy data using natural language processing (NLP) techniques, enabling the systematic extraction and categorization of unstructured information. Open -text descriptions of vacancies are thus pre -processed to align with the structured format of the skills taxonomy and its associated keywords. A skill is identified if at least one relevant

using natural language processing (NLP) techniques, enabling the systematic extraction and categorization of unstructured information. Open -text descriptions of vacancies are thus pre -processed to align with the structured format of the skills taxonomy and its associated keywords. A skill is identified if at least one relevant keyword from the taxonomy appears in a vacancy. Adamczyk et al. (2025) provide a detailed explanation of the implementation of this methodology across countries , which interested practitioners and researchers are welcome to use in their own work . The present brief employs the same NLP methodology to code skills variables. The data analysed stem from the ESCWA Skills Monitor. ESCWA’s database is a n AI -driven platform that collects online job postings and uses machine learning to track the skills in demand across the Arab region. Since June 2020, it has collected and analysed over 4 million job postings from around 90 online job portals. The platform uses advanced deep learning and NLP algorithms to classify job advertisements into standardized job titles and extract skills based on a predefined taxonomy. Each job title is standardized according to the International Standard Classification of Occupations (ISCO -08) and linked to a corresponding title from the European Skills, Competences, Qualifications, and Occupations (ESCO) framework , extending the classification beyond the 4-digit ISCO level to ESCO’s highly granular 6-digit level through a custom-built text classification algorithm. Additionally, the monitor employs a named -entity-recognition (NER) -based algorithm to classify skills . In the context of the questions addressed in this brief, however, the ILO skills taxonomy was employed to ensure comparability with the analysis conducted for the 2026 Report on Lifelong Learning and Skills for the Future . The online vacancy data differ in size, by country: the UAE’s sample includes 824,723 online job advertisements (OJAs), while the data for Egypt and Jordan

conducted for the 2026 Report on Lifelong Learning and Skills for the Future . The online vacancy data differ in size, by country: the UAE’s sample includes 824,723 online job advertisements (OJAs), while the data for Egypt and Jordan include 222,587 and 44,271 OJAs, respectively. ILO Brief 4 Skills dynamics in the Arab region: New evidence from online vacancy data The ESCWA Skills Forest provides a visual representation of the skills 3 demanded in the labo ur market . The forest reflects key skills diversification that provide s strategic insights for economic planning: ● Skill Variety : Reflecting the breadth of distinct skills demanded. ● Skill Mobility : Showing how transferable or versatile specific skills are between occupations. Each node represents a job title. The size of the nodes reflects the number of OJAs in each role, with larger nodes signifying higher demand. The links between these nodes represent the degree of similarity between the skill requirements of different jobs. The proximity and clustering of nodes reveal relationships between occupations based on shared skills sets. The heart of the forest lies a densely interconnected "core," representing the most central and interrelated jobs in the virtual labo ur market. Thi s core reflects occupations that share a high degree of skill overlap, making transitions between these roles through reskilling or upskilling comparatively easier. Conversely, jobs on the periphery are less connected, requiring greater effort for workers to adapt to the skills needed for these roles.

 3. Exploring the specific skills in demand The overall distribution of skills during the four-year period (2020 to 2024 ) for Egypt, Jordan, and the UAE reveals a n emphasis on cognitive skills (shown in green in Figure 1) and socio-emotional skills (shown in orange), while manual skills (shown in blue) are less frequently demanded in vacancies. It is noticeable that the UAE market demands more socio -emotional skills than cognitive skills, while

emphasis on cognitive skills (shown in green in Figure 1) and socio-emotional skills (shown in orange), while manual skills (shown in blue) are less frequently demanded in vacancies. It is noticeable that the UAE market demands more socio -emotional skills than cognitive skills, while

3 In this brief, the Skills Forest is based on the ILO Skills taxonomy, classified as in Adamczyk et al. (2025). Egypt and Jordan show almost equal shares of demand between the two broad skill types. Turning to more detailed skill demand, s ophisticated cognitive skills and core cognitive skills together are the most demanded skills across the three countries, at 24 per cent for Egypt, 22 for Jordan and 21 for UAE . This reflects the sustained importance of skills needed to solve problems and think creatively and critically. People management and social skills likewise represent significant shares of demand , particularly in the UAE, where social skills represent 20 per cent of the distribution, indicatin g a strong focus on leadership and service -oriented roles that require personal interactions . Meanwhile, physical and finger-dexterity-related skills and hand -foot-eye coordination, although slightly more demanded in the UAE, are the least demanded skills in the three countries, at least according to the online labour market s analysed. This suggests that the three countries’ online labour markets have features of a knowledge -based economy, with a stronger focus on interpersonal, managerial, and cognitive expertise tailored to the evolving demands o n the workforce. Figure 1 also highlights a small shift in the skills distribution of Egypt, Jordan, and the UAE from 2020 to 2024, emphasizing the different growths patterns of the demand for cognitive, socio-emotional and manual skills. There is a general increase in the demand for socio-emotional skills in Egypt and Jordan, whereas in the UAE, demand growth concerns manual skills, marginally shrinking the demand for socio-emotional skills. More specifically, in Egypt, demand for general computer

for cognitive, socio-emotional and manual skills. There is a general increase in the demand for socio-emotional skills in Egypt and Jordan, whereas in the UAE, demand growth concerns manual skills, marginally shrinking the demand for socio-emotional skills. More specifically, in Egypt, demand for general computer skills saw a decline of 2 percentage points, while demand for social skills increased by 4 percentage points. In Jordan, there has been an increase in the demand for people management skills from 6 t o 8 per cent, and a decline in sophisticated cognitive s kills from 10 to 8 per cent. In the UAE, there was a decline in the demand for customer service skills, by 2 percentage points. ILO Brief 5 Skills dynamics in the Arab region: New evidence from online vacancy data  Figure 1: Skills compositions of vacancies and their evolution over time, Egypt, Jordan, and the United Arab Emirates

Note: The figure displays the percentage of occurrences of a specific skill relative to the total occurrences of all skills across time (top panel) and in given years (bottom panel). The years were selected based on data availability and to maximize the timespan.

Source: The vacancy data for three countries stem from ESCWA’s Skills Monitor Database. Skills variables are defined and coded according to the ILO’s skills taxonomy. ILO Brief 6

Skills dynamics in the Arab region: New evidence from online vacancy data  4. Uncovering interconnections between skills Figure 2 highlights the interconnections between skills as reflected in vacancies, showing that skills do not operate in isolation. Each node represents one of the 15 subcategories of skills from the ILO’s taxonomy, with the density of the connecting lines indicating the strength of the relationship between them: The thicker the line between two skills, the more often both skills are simultaneously required by employers. Strong connections are particularly evident among the four types of socio -emotional skills, as well as between these skills and cognitive skills. For instance,

connecting lines indicating the strength of the relationship between them: The thicker the line between two skills, the more often both skills are simultaneously required by employers. Strong connections are particularly evident among the four types of socio -emotional skills, as well as between these skills and cognitive skills. For instance, across all three countries, core and sophisticated cognitive skills consistently appear in conjunction with character and social skills. These skills also serve as the foundation for various technical and managerial skills across all three countries. Yet, there are also country -specific patterns. Connections among socio -emotional skills are especially pronounced in Egypt and the UAE, whereas in Jordan, the relevance of skill bundles is more evenly distributed across the various skill subcategories. Both Egypt and the UAE show a prominent role of customer service skills, which are closely linked with core cognitive skills and other socioemotional skills. In Egypt, there tend to be relatively many links between financial skills and other domains, reflecting the relevance of finance in its labour market. Even skills that are less frequently mentioned in vacancies (recall Figure 1 above) are typically required in combination with core and sophisticated cognitive and social skills. For instance, in Egypt, 8 7 per cent of vacancies requiring machine learning and AI skills also mention core cognitive skills, 66 per cent sophisticated cognitive skills, and 93 per cent social skills. Similar patterns are observed for machine learning and AI skills in Jordan and the United Arab

4 On the ESCWA Skills Monitor portal, under “Country Profiling”, users can investigate the skills forests of different countries. The optimal choice of nodes that show the diversity difference is 150 nodes per forest. Emirates. Within the broad manual skills category, fingerdexterity skills are more strongly linked with core cognitive and social skills than other manual skills. Across all three countries, at least 60 per cent of vacancies requiring fingerdexterity skills also specify core cognitive skills, and at least 75 per cent require social skills. Next, the skills forests below offer a structural overview of

dexterity skills are more strongly linked with core cognitive and social skills than other manual skills. Across all three countries, at least 60 per cent of vacancies requiring fingerdexterity skills also specify core cognitive skills, and at least 75 per cent require social skills. Next, the skills forests below offer a structural overview of the online labour markets in the UAE, Egypt, and Jordan, highlighting how occupations cluster based on shared skill requirements. Jobs are grouped according to the nine major ISCO categories, including Managers, Professionals, Technicians and Associate Professionals, Cler ical Support Workers, Service and Sales Workers, Craft Workers, Plant and Machine Operators, and Elementary Occupations, and skills are classified using the ILO Skills Taxonomy. In the skills forests, each node represents a distinct occupation . Occupations are captured according to the ESCO Framework, at its highly granular 6 -digit level. T he spatial proximity between nodes reflects the degree of shared skills between them. The closer two occupations appear, the greater the overlap in their underlying skill requirements, indicating higher potential for mobility between them. Although derived from the demand side, this approach captures the latent transferability of individuals' skills across roles, thereby offering an inclusive and practical perspective on labo ur mobility across the region. Importantly, this form of skills-based mobility does not consider formal entry barriers such as academic credentials, licensing requirements, or occupation -specific certifications. In most Arab countries, the skills forest indicates a single, centralized core, which represents the most central and interrelated occupations. In contrast, jobs linked to the Fourth Industrial Revolution (4IR), such as those requiring IT, digital, and technological skills, are either positi oned at the edges of the core or relegated to the periphery. This positioning reflects their still limited integration within the broader labour market. 4 ILO Brief 7 Skills dynamics in the Arab region: New evidence from online vacancy data  Figure 2: Skills networks among vacancies in Egypt, Jordan, and the United Arab Emirates

positioning reflects their still limited integration within the broader labour market. 4 ILO Brief 7 Skills dynamics in the Arab region: New evidence from online vacancy data  Figure 2: Skills networks among vacancies in Egypt, Jordan, and the United Arab Emirates

Note: Each skills network shows the relationships between skills in each country. The thicker the line between two skills, the more frequently both skills are jointly required in job adverts. The time span covered is 2020-2024. ILO Brief 8

Skills dynamics in the Arab region: New evidence from online vacancy data Egypt: A skills forest shaped by economic diversification challenges Figure 3 provides a detailed visualization of Egypt's labour market distribution by skill type and occupation, offering insights into the diversity of roles and the balance between traditional and emerging sectors. Egypt’s digital skills variety could be considered an opportunity to overcome its premature deindustrialisation 5 by creating new types of jobs that connect well with the Fourth Industrial Revolution (see Ayed Mouelhi and Ghazali 2021; Fedi et al. 2019). As explained before, the size of each node corresponds to the number of vacancies, with larger nodes indicating higher frequency. A central cluster pertains to administrative roles, often related to the broad occupational groups of managers and professionals, while there are an additional technical support cluster (top left quadrant) and a customer service cluster (lower part of the forest). Regarding the technical support cluster , positions like ICT help desk manager and ICT system administrator indicate demand for cognitive and other skills used in information and communication technology and show an interconnectedness of skills that allow for comparatively straight-forward mobility for workers amongst related occupations6. The concentration of these roles suggests that Egypt’s online labour market features jobs requiring medium to high levels of education and professional training. Moreover, this concentration suggests the potential for the creation of a new forest core emphasising ICT and related fields and the labour market's adaptation to digital and technological requirements.

that Egypt’s online labour market features jobs requiring medium to high levels of education and professional training. Moreover, this concentration suggests the potential for the creation of a new forest core emphasising ICT and related fields and the labour market's adaptation to digital and technological requirements. Even though Egypt appears to have fewer high-technology roles such as AI specialists and cloud engineers, compared to advanced economies, Egypt’s forest is more diversified

5 Egypt has experienced prolonged stagnation and a relative decline in its manufacturing sector. Since the 1990s, the share of manufacturing in both GDP and employment has remained largely stagnant or declined, reflecting a trajectory consistent with premature deindustrialization. 6 The skills forest reveals some noteworthy and perhaps unexpected patters. For example, the occupation of “ICT system integration consultant“ , in the lower part of the forest, exhibits skills that, while they are tightly related to the ICT occupations in the upper left cluster, are more closely related to enterprise development workers than to any other occupation in terms of a measured distance of 2.37. This short distance reflects the overlapping competencies in systems implementation and organizational development. This is followed by very similar distances between this occupation ( i.e., ICT system integration consultant s) and than the ones of other countries in the region, according to analysis by ESCWA7. Another interesting aspect is the moderate representation of clerical support workers and service and sales workers in the online vacancy data (nodes in purple and orange), with roles like customer service representatives and ICT system integration consultants (at the lower part of the forest). This reflects an online labour market that serves both traditional sectors which pre -date more recent technological developments and less traditional industries. The graph also captures the diversity within managerial and office roles, with positions such as contract manager , executive assistant, and administrative assistan t ( in the mid-upper part) indicating a structured demand for office work in various industries and relatively straightforward labour mobility – in terms of skill requirements - between these occupations. Moreover, healthcare -related positions like

roles, with positions such as contract manager , executive assistant, and administrative assistan t ( in the mid-upper part) indicating a structured demand for office work in various industries and relatively straightforward labour mobility – in terms of skill requirements - between these occupations. Moreover, healthcare -related positions like healthcare institution manager and medical practice manager (lowest part of the forest) underline the importance of health services in the employment landscape. Although the connection s between the nodes do not show at the current threshold applied to the forest, the location of these occupation s in the forest and their proximity indicates straightforward movements between occupations based on the sha