Promoting
structural change

Recently, however, several countries (such as Ghana, India and Zambia) have more or less moved straight from agriculture to services.1 This “premature deindustrialisation” is sometimes viewed as troubling owing to the unique role that manufacturing plays in aiding economic growth and development. Unlike services, manufacturing exhibits unconditional convergence – that is to say, convergence of manufacturing output per worker is not, in principle, dependent on the quality of economic institutions, governance and education.1

Indeed, before 1990 the growth models of many developing economies prioritised industrialisation, supported by investment in capital equipment, technology, education and infrastructure. Post-communist economies in the EBRD regions began the 1990s with larger manufacturing sectors than other countries at a similar level of development. Their proximity to the European Union also meant that they became integrated into European value chains relatively quickly and were able to pursue manufacturing export-led growth. As a result, their manufacturing sectors remained larger than those of their peers.

Over time, the pursuit of manufacturing export-led growth has become increasingly challenging for many countries, largely as a result of competition from China and other developing economies. Moreover, in most economies in the EBRD regions, as well as China and India, the service sector’s contribution to economy-wide labour productivity growth has far exceeded that of the manufacturing sector in the period since the 1990s.

At the same time, the advent of digital technologies has transformed the service sector, making services easier to trade across borders. Manufacturing has also become increasingly reliant on service inputs. Within services, digitally enabled, tradeable services – especially global innovator services such as information and communication technology (ICT)-related services (which are defined, for the purposes of this chapter, as telecommunications, computer and information services), financial services, insurance services, professional services, and scientific and technical services – exhibit particular growth potential. These have increasingly driven improvements in the labour productivity of the service sector. These services rely on high levels of skill, can be traded across borders and have strong linkages to the rest of the economy.

The legacy of central planning meant that post-communist economies in the EBRD regions initially had underdeveloped service sectors and experienced a slower shift from manufacturing to services. That shift has indeed taken place, however, and those economies’ historical legacy of strong human capital focused specifically on engineering and technical skills has allowed them to develop ICT services, professional services and other services with high levels of output per worker (that is to say, high levels of labour productivity). While many post-communist economies in the EBRD regions are top exporters of computer and information services, others need to upgrade their infrastructure, skills and institutional capabilities if they are to excel in a service-based world.

Service trade liberalisation and targeted industrial policies can help to support the shift towards high value-added services, provided that the necessary fundamentals are in place. For instance, economies with stronger state capacity see a marked increase in service-related FDI projects after investment promotion agencies (IPAs) start to target foreign investment in specific service sectors. At the same time, no such effects are observed in economies with weaker state capacity, and such targeting of service sectors has no impact on manufacturing-related FDI projects. Similarly, tax incentives granted to computer and information service firms in Romania have been effective in supporting employment growth in the computer and information service sector, but primarily in regions with strong historical endowments of specialist human capital.

This chapter starts with an overview of structural change in the EBRD regions since the early 1990s. It then discusses the rising importance of services, including as an input for manufacturing, and analyses the conditions and policies that are necessary for a structural shift to more productive service sectors, such as service trade liberalisation and FDI promotion. The chapter ends with several policy recommendations.

Since 2008, however, manufacturing exports and FDI have stagnated as a share of total output, with newcomers facing far stiffer competition. The world’s top 10 countries in terms of the production of manufactured goods have accounted for around 71 per cent of gross global production since 1995, but production has become more geographically concentrated. In 1995, the world’s top manufacturing producer was the United States of America, accounting for 21 per cent of gross global production. By 2020, however, China was at the top of the list with 35 per cent (up from just 5 per cent in 1995).5 Against this backdrop, economies are increasingly looking for an alternative growth model based on a shift from manufacturing to highly productive services.

Large shifts from manufacturing to services are a conventional post-industrialisation pattern in advanced economies, with employment typically shifting to services once manufacturing has achieved a certain level of productivity. In the post-communist economies of the EBRD regions, however, this shift started when manufacturing productivity was still relatively low. Nevertheless, in 2022 manufacturing’s average share of total value added in those post-communist economies was still around 5 percentage points higher than in comparator economies with equivalent levels of income per capita and similar characteristics.

This premature deindustrialisation is a more general trend, rather than a phenomenon specific to the EBRD regions. The most plausible explanation for this trend is globalisation. When these developing economies were first exposed to global markets, those without a comparative advantage in manufacturing became net importers of manufactured goods. Moreover, in advanced economies, the relative price of manufactured products had been declining owing to productivity improvements and the ability to import cheaply. When they were exposed to these price declines, the developing economies effectively “imported” deindustrialisation.7

However, the experiences of the EBRD regions have not been uniform. The post-communist EBRD economies in the EU8 have benefited from (i) better initial conditions, (ii) reforms that were largely driven by the EU accession process, and (iii) their membership of the EU’s single market, which has allowed them to maintain a stronger manufacturing core than other post-communist economies (notably those in the EEC region and Central Asia).9 In 1997, the two groups of economies were roughly similar in terms of exports’ share of GDP: 39.8 per cent in the EEC region and Central Asia, and 43.3 per cent in the post-communist EBRD economies in the EU. By 2021, that figure had risen to 69.3 per cent in the second group, while it had dropped to 35.7 per cent in the first. In that same year, trade-weighted import tariffs averaged 4.6 per cent in the EEC region and Central Asia, compared with 1.4 per cent in the post-communist EBRD economies in the EU.

The Western Balkans and SEMED regions experienced similar shifts, but from a much smaller industrial base, exemplifying the premature deindustrialisation phenomenon. This is concerning, since manufacturing has historically played an important role in terms of driving unconditional convergence in labour productivity,10 absorbing unskilled labour and providing opportunities for export-led growth, as it is tradeable and not constrained by the size of the domestic market.

Türkiye stands out as an economy in the EBRD regions that has managed to buck the downward trend in the importance of manufacturing. Since 2008, manufacturing’s share of total value added in Türkiye has increased by almost 7 percentage points. After the “lost decade” of the 1990s, which was marked by three major economic crises, a number of reforms have been implemented in Türkiye since 2001. The establishment of closer links with the EU through Türkiye’s membership of the customs union for manufactured goods as of 1995 and the start of accession negotiations in 2005 have resulted in increased trade and investment opportunities for Turkish companies and triggered significant improvements in the sophistication and quality of export products through the adoption of EU standards and the transfer of knowledge.11

A common way of quantifying the share of aggregate labour productivity growth that is due to structural change involves using a shift-share decomposition.12 This separates growth in aggregate labour productivity into two components: fundamentals and structural change. The fundamentals component captures intra-sector contributions to growth through innovation and the upgrading of capital stock (which result in improvements in the labour productivity of firms within a given sector), whereas the structural change component captures the productivity dividend that is derived from workers shifting into sectors with relatively higher levels of productivity (see Chart 2.2).13

Average labour productivity growth in the EBRD regions between 1991 and 2018 was weaker than in China and India, reflecting the EBRD economies’ different starting points and thus the smaller remaining gaps in labour productivity. Most of the overall growth in labour productivity was driven by intra-sector growth. However, growth-enhancing structural change was also observed in advanced European economies, China and India, as well as all EBRD regions except the EEC region and Central Asia.

In the EEC region and Central Asia, structural change was growth-reducing between 1991 and 2018 – a development that was driven primarily by manufacturing. This reflected a shift from industry to low-productivity services and informality.14 Intra-sector productivity growth was significantly stronger in those economies, primarily owing to the wider margin for improvement as a result of their lower initial productivity levels.

In contrast, EBRD economies in the EU, the economies of the SEMED region and Türkiye all experienced a small growth-enhancing structural change over the same period (with manufacturing making a negative contribution in both EBRD economies in the EU and the SEMED region). In EBRD economies in the EU, structural shifts accounted, on average, for around 20 per cent of total labour productivity growth, compared with 28.8 per cent in advanced Europe and 41.4 per cent in Türkiye. Increasingly, the remaining labour productivity gaps in the EBRD regions reflect differences between the productivity levels of manufacturing and other sectors that cannot absorb unskilled labour to the same extent, such as business services.

In China, an average of 17.5 per cent of the country’s labour productivity growth over that period was attributable to structural shifts (primarily shifts from agriculture to manufacturing and services). In India, meanwhile, structural change consisted mostly of shifts straight from agriculture to the service sector, possibly owing to numerous restrictions on the manufacturing sector (including industrial licensing, tariff and non-tariff barriers to imports and restrictions on FDI).15 In 1995, agriculture accounted for 27 per cent of India’s total value added and manufacturing accounted for a further 20 per cent. By 2018, the shares of manufacturing and agriculture had fallen to 16 per cent each.

At the same time, services cover many different economic activities, ranging from retail shops, restaurants, hairdressers, hotels and transport on the one hand to education, health, R&D, and information and computer activities on the other. These activities vary in terms of the extent to which they can be traded internationally, as well as in their scalability, the extent to which they can benefit from innovation and digitalisation, their linkages to other sectors in the economy and their capacity to absorb low-skilled workers. On the basis of these characteristics, services can be grouped together in four broad categories: global innovator services, low-skill tradeable services, skill-intensive social services and low-skill domestic services.16

Global innovator services consist of ICT services, financial services, insurance services, professional services, and scientific and technical services.17These services can be traded internationally through remote cross-border delivery, they mostly employ skilled workers, and they have strong links to other domestic sectors. ICT services and financial services are relatively capital-intensive, while ICT services, professional services, and scientific and technical services are highly R&D-intensive.

Low-skill tradeable services such as wholesale trade, transport and logistics services, and accommodation and food services are also traded internationally, but they mostly employ low-skilled workers. Transport and logistics services and wholesale trade have greater linkages to other sectors, making them amenable to offshoring. Accommodation and food services, meanwhile, are exported by being consumed by tourists abroad – they cannot be provided remotely.

Skill-intensive social services encompass education and healthcare services. These also have a comparatively high percentage of skilled workers and are tradeable (albeit to a more limited extent): they can be exported through FDI, enrolment of foreign students or “medical tourism”. In the EBRD regions, the Kyrgyz Republic, Serbia, North Macedonia and Slovenia were all among the world’s top 10 exporters of personal healthcare services in 2022 as a percentage of GDP.

In contrast, low-skill domestic services are not typically traded internationally, they employ a comparatively high percentage of low-skilled workers and they tend to have fewer linkages to other sectors. Such services include retail trade, administrative and support services, arts, entertainment and recreation services, and other community and personal services.

In most EBRD regions, skill-intensive social services and low-skill tradeable services account for the majority of value added in the service sector (see Chart 2.3). In advanced Europe, global innovator services account for 37 per cent of all value added in services, compared with about 30 per cent in EBRD economies in the EU. In other EBRD regions, global innovator services account for smaller percentages of total value added in services, which is limiting the potential for service-led productivity growth in the short term.

The service sector’s contribution to economy-wide labour productivity growth has exceeded that of the manufacturing sector in all EBRD regions, as well as India and China (see Charts 2.2 and 2.4). Advanced European economies have less scope for improvements in service-sector labour productivity than the EBRD regions, China and India, since services have historically accounted for a much larger share of total value added in those advanced economies. Most of the improvements there have been driven by global innovator services, reflecting both shifts towards the global innovator service sector and improvements in productivity within that sector. In EBRD economies in the EU, labour productivity growth in the service sector has slowed since 2009, reflecting the fact that the service sector’s share of total value added has increased substantially since 1990 and services have already reached a high level of sophistication. Elsewhere, global innovator services’ contribution to total productivity growth in the service sector has been significant, but relatively modest, partly reflecting those services’ smaller share of total value added as shown in Chart 2.3.

Within exports of other commercial services, the average share of digitally enabled global innovator services is, if anything, higher in the EBRD regions than in advanced Europe (with the exception of the SEMED region). Since 2020, exports of these services have also been growing faster than goods exports in all EBRD regions apart from the SEMED region, the EEC region and Central Asia. In EBRD economies in the EU, the Western Balkans and Türkiye, the average annual compound growth rate for exports of digitally enabled services exceeded the equivalent rate for their non-digital counterparts by a factor of 1.8 between 2005 and 2022.

While economies in the SEMED region have liberalised trade in goods, that region is one of the most restrictive when it comes to trade in services, with an estimated service trade restrictiveness score that is twice that of Europe and Central Asia.18 In those economies, trade in services and the competitiveness of services have been held back by restrictive policies that (i) limit the entry of competitors seeking to take on incumbent state-owned enterprises (in the telecommunications sector, for instance) or (ii) impose licensing requirements and charge high operating fees (especially for professional services).19 Despite improvements since 2016, trade in services remains highly restricted in all SEMED economies: a global dataset spanning 134 countries considers that Tunisia and Egypt have the 5th and 10th most restrictive practices of all the countries covered, with Jordan in 17th place and Morocco in 21st.20

Estonia, Ukraine, Serbia, Armenia, North Macedonia and Moldova were all among the top 10 exporters of computer services worldwide in 2022, measured as a share of GDP (see Chart 2.6). In the same year, Bulgaria, Estonia, Romania, Georgia, Slovenia and Czechia were among the top 10 exporters of information services as a share of GDP.21 Most of these economies have seen marked increases in the ratio of computer service exports to GDP since 2005, catching up with the early movers in that industry (which include countries such as Israel, India and the Netherlands; see Chart 2.6).

Moreover, there is a strong positive correlation between (i) medium- and high-technology-intensive sectors’ share of manufacturing value added in 1990 and (ii) global innovator services’ share of total employment in 2018 (see Chart 2.8). This correlation reflects the importance of human capital for both sets of industries – which, in turn, facilitates the transition from technology-intensive manufacturing to highly productive services.

Educational upgrading and structural change also have implications for the types of task for which there is demand in the local labour market. Analysis drawing on individual- level data from the EU Labour Force Survey (LFS) and the O*NET-SOC occupational taxonomy shows that since 1998, the importance of non-routine cognitive tasks – that is to say, tasks that require creativity, problem solving and complex communication skills – has increased across EBRD economies in the EU, almost as much as in advanced European economies (see Chart 2.9).24

This trend reflects both the evolving demands of a service-oriented economy and the ways in which technology responds to available sets of skills.25 On the supply side, educational advancements have helped to increase the supply of skilled workers who are capable of performing complex tasks. On the demand side, technological change has simultaneously reduced demand for routine tasks that are susceptible to computerisation and increased demand for non-routine cognitive skills that are complementary to computer technology.26 This shift has supported the growth of occupations in high-productivity service sectors, such as ICT services, which rely heavily on abstract tasks that cannot easily be automated.27

Using a shift-share decomposition, the observed increase in the intensity of non-routine cognitive tasks can be broken down into (i) a change in the composition of tasks within occupations (for example, the fact that a secretary’s job now involves complex tasks using computers and software, which are more complicated than the simple clerical work carried out in the past) and (ii) a structural change component reflecting changes in the occupational structure of employment. The latter can be broken down further into changes attributable to (i) shifts in the occupational structure of individual sectors and (ii) the movement of workers across sectors (see Chart 2.10).

The results reveal that while the abstract task content of specific occupations has remained relatively stable, structural change has been the primary driver of the increased intensity of non-routine cognitive tasks in European labour markets. This structural change is playing two roles. First, existing industries are transforming to incorporate more occupations with greater intensity of non-routine cognitive tasks. For example, professional occupations’ share of total occupations within sectors increased by an average of 57 per cent in EBRD economies in the EU between 1998 and 2019 and by an average of 86 per cent in advanced European economies over the same period (with the ISCO definition of “professional occupations” including professions such as lawyers and IT professionals). Second, there has been a broader shift towards sectors that require more abstract skills. In particular, global innovator services’ share of total employment in EBRD economies in the EU increased by 7 percentage points between 1998 and 2019, reaching 20 per cent.28

Data suggest that growth is often still export-led, but it is now more likely to be led by exports of services. Before 2008, growth was led by manufacturing exports in the majority of the EBRD economies outside the EU, and it was led solely by service exports in most advanced European economies. It was led by both manufacturing exports and service exports in other emerging market economies (see Chart 2.11). Since 2008, by contrast, growth in EBRD economies in the EU has tended to be led by both manufacturing exports and service exports, and in a significant percentage of other EBRD economies growth has been non-export-led. In other emerging market economies, meanwhile, growth is now just as likely to be led by service exports as non-export-led. In some economies, growth has gone from being led solely by service exports before 2008 to being led by both manufacturing exports and service exports since 2008 (see Box 2.3, which describes the example of Morocco).

Countries with highly skilled workforces and other strong fundamentals (such as robust governance and liberalised trade in services) have the best potential to achieve service export-led growth (see Chart 2.12). In the EBRD regions, the economies that fall into this category are the EU member states and Albania. Jordan, Kazakhstan, Moldova, Serbia and Ukraine already have relatively highly skilled workforces, but need to improve their fundamentals in order to realise their potential. Egypt, Morocco, Tunisia and Türkiye, on the other hand, need to improve both the skills of their workforces and their fundamentals.

In emerging market economies outside the EBRD regions, value added originating in the service sector accounted for almost a third of the total value added in manufacturing exports in 2020. In advanced Europe, it accounted for a third (see Chart 2.13). In those advanced economies, global innovator services such as ICT services and financial services typically accounted for between 8 and 23 per cent of the total value added in manufacturing exports. However, in all regions shown in Chart 2.13, the service sector’s largest contribution to total value added in manufacturing exports came from low-skill tradeable services such as transport, followed by global innovator services.

In addition, service-related functions such as R&D, logistics, marketing and ICT services now account for a larger share of total employment in manufacturing firms. EU LFS data show that service-related occupations in the manufacturing sector (referred to as “embodied services”) accounted for an average of 55 per cent of all manufacturing-sector occupations in advanced European economies in 2019, up from about 45 per cent in 2000 (see Chart 2.14). In EBRD economies in the EU, that share increased by an average of 5 percentage points over the same period, standing at 40 per cent in 2019.

Manufacturing value added per worker is often higher in countries where services play a larger role in the manufacturing process. This is probably because advanced European economies have focused on retaining intangible high-skill production activities within their global value chains. These activities include pre-production tasks such as R&D and product design, as well as post-production tasks such as after-sales services and marketing.

At the same time, those economies have outsourced or automated labour-intensive low-skill production activities such as assembly. As a result, more value tends to be added to manufactured products in the pre- and post-production stages than in the intermediate production stage. This creates a “smile curve” pattern in the distribution of value added along global supply chains, with higher values at the beginning and the end of the process and lower values in the middle.34

That kind of policy-light approach would not work as well now. Innovation in manufacturing is increasing demand for specific skills, while the use of robots, 3D printing and other forms of automation have reduced the benefits of having plenty of cheap unskilled labour.42

Services, meanwhile, have different requirements. Global innovator services such as ICT services and business process outsourcing require (i) skilled labour, (ii) investment in physical capital, technology and innovation (private fundamentals), and (iii) strong infrastructure, robust economic institutions and a conducive business environment (public fundamentals).43 The liberalisation of trade in services may allow economies to target some low-hanging fruit in terms of facilitating a structural shift towards services with higher value added. However, most other enabling factors cannot be changed overnight and will require a sustained policy effort over the medium term.

The EBRD economies in the EU became part of the European single market – the European Economic Area (EEA) – when they joined the European Union. In addition to the free movement of goods, capital and people, the single market also includes the free movement of services following the adoption of the EU’s Services Directive in 2006. As a result, restrictions on trade in services are much lower inside the EEA than outside it, as illustrated by Chart 2.17 on computer services – a sector that is playing an increasingly important role. As the chart shows, EEA countries such as Iceland, Belgium and Greece have STRI scores in the top quartile of the global distribution for trade with non-EEA countries, with much lower restrictions for intra-EEA trade. In the EBRD regions, trade in computer services is less restricted than in middle-income comparator economies, but more restricted than in advanced economies.

The cost of trade in services is almost double that of trade in goods, with differences in the quality of governance, trade policy and regulations accounting for more than a quarter of total variation in the cost of bilateral trade in services.45 In addition, the extent of ICT adoption is more important for services than goods in terms of facilitating trade. The WTO estimates that the cost of trade in services dropped by 9 per cent between 2000 and 2017 thanks to digital technologies, investment in infrastructure and the lowering of policy barriers to trade.

Analysis suggests that market access matters for service exports and that liberalising your own service market does more to boost service exports than exporting to a liberalised service market (see Chart 2.18). For example, if all countries of origin with higher STRI scores reduced their restrictions to match the 25th percentile of the STRI distribution in the relevant sector, their service exports would grow by 9.1 per cent and their service imports would increase by 5.5 per cent. A similar reduction in the STRI score of a destination country is associated with smaller increases in exports and imports (increases of 2.1 and 3.4 per cent, respectively).46 The reason for this is twofold: (i) a country’s own services are more competitive if its service market is liberalised, and (ii) imported services are an input for service exports.47

A common spoken language and a common legal system are more important for trade in services than trade in goods. At the same time, having a shared border does not appear to be a significant determinant of bilateral trade in services, unlike trade in goods. As with goods, bilateral trade in services tends to be stronger when the two countries are larger, and it tends to be weaker when the two countries are further apart geographically.

Further analysis suggests that relaxing restrictions on digital trade in services in the country of origin or destination is also associated with increases in exports and imports of services. This analysis is conducted by incorporating OECD Digital Services Trade Restrictiveness Index (DSTRI) scores (and excluding transport services, digital delivery of which is uncommon), as well as adding an indicator for the application of the EU’s General Data Protection Regulation (GDPR) or GDPR-equivalent legislation. Relaxing restrictions on digital trade in services in the country of origin to match the 25th percentile of the DSTRI distribution is associated with a 20.4 per cent increase in service exports and a 25.5 per cent increase in service imports. A similar reduction in the DSTRI score of a destination country, on the other hand, is associated with a 25.4 per cent increase in service exports and a 20.4 per cent increase in service imports. These estimates are statistically significant at the 1 per cent level (while including DSTRI scores in the model results in the estimated coefficients for the two variables involving STRI scores becoming smaller and losing their statistical significance). In addition, less restrictive regimes for digital trade in services have been shown to be associated with increases in the productivity of manufacturing firms.48

Having the GDPR (or equivalent legislation) in force in the country of origin is associated with an increase in service exports, while having such legislation in force in the destination country is associated with an increase in service imports, probably because having clear privacy and security regulations – even if the rules are strict – reduces ambiguity around data protection and supports trade in services.

In 2023, the EBRD conducted an online survey of the national IPAs in its shareholder economies, gathering data on the sectors targeted, the strategies employed and the timing of the relevant initiatives.49 The information collected was combined with data in the FT fDi Markets database – a project-level dataset on FDI projects – to assess the effectiveness of sector-targeting policies.

The effects of sector targeting are evaluated here using a difference-in-differences approach. The outcome of interest is the number of investment projects in a given country, sector and year. The analysis compares the actual outcomes for targeted sectors with counterfactual outcomes in the absence of policy intervention. The counterfactual outcomes are estimated by using sectors that are never targeted or have not yet been targeted as a control group.50

The results suggest that sector-targeting policies tend to have a positive effect on FDI.51 Ten years after a policy has been rolled out, targeted sectors see, on average, 2.8 times as many FDI projects as non-targeted sectors. This increase is driven predominantly by investment in service-related projects, with no significant effect being observed for manufacturing projects. This mirrors global trends in FDI, which is dominated by services: service-related projects’ share of total cross-border greenfield projects increased from 66 per cent in 2004 to 81 per cent in 2023. Moreover, service-related projects’ share of total cross- border greenfield projects within manufacturing industries (involving activities such as R&D, for instance) has nearly doubled to about 70 per cent, propelled by rapid technological advances. In contrast, manufacturing FDI has recently experienced a significant downturn (and was already stagnating before that).52

Furthermore, the effectiveness of sector targeting in terms of service-related FDI is also contingent on state capacity.53 Indeed, countries with stronger state capacity see a marked increase in service-related FDI projects following the targeting of sectors, whereas countries with weaker state capacity see no differences between targeted and non-targeted sectors in terms of the number of projects (see Chart 2.19).

Investment promotion can be used to foster structural change and a shift towards services, and policymakers should prioritise services in such strategies. At the same time, they should also implement reforms aimed at enhancing governance, improving the efficiency of public administration and strengthening the rule of law, which can amplify the impact of those investment promotion efforts.

Balance of payment statistics are the main source of data on trade in services. However, there are differences across countries in terms of both the availability of certain data and the methodologies applied. This often leads to asymmetries between reported exports and imports of services. This chapter relies mainly on the WTO-OECD Balanced Trade in Services dataset,57 which contains bilateral data on trade in services for the period 2005-21 and aims to reconcile these asymmetries.

The BaTIS dataset uses broad sectors; for example, telecommunications, computer and information services are grouped together. The WTO’s experimental Trade in Services data by Mode of Supply dataset contains information on each of these three subsectors, including details of how the service is supplied (cross-border, consumption abroad, commercial presence or movement of people). Thus, TiSMoS allows trade in services to be broken down into non-digital and digitally enabled components. It does not, however, contain bilateral data, and it relies on a set of assumptions to allocate trade to different modes of supply, as most countries do not distinguish between different modes in their official statistics.58

Each of the composite indices shown in Chart 2.9 is constructed as the sum of constituent task items.59 First, the individual O*NET task item scores are standardised within each country so that they have a mean of 0 and a standard deviation of 1. These standardised scores are summed to obtain five different composite task intensity indices (“non-routine cognitive analytical tasks”, “non-routine cognitive interpersonal tasks”, “routine cognitive tasks”, “routine manual tasks” and “non-routine manual physical tasks”), which are then standardised within each country. Next, the average of these occupation-level composite measures is computed for each country-year cell, and a two-year moving average is applied to the resulting country-level indices to smooth year-to-year volatility. Lastly, the country-level task intensity measures are aggregated into broader country groupings using unweighted cross-country averages.

The structural break in task intensities stemming from the switch from ISCO-88 to ISCO-08 in 2011 was corrected by equalising the means of the task importance measures for the two years immediately before and after the classification change.60

The analysis covers the period between 2008 and 2019 and focuses on firms which had at least five employees in at least one year between 2000 and 2021. The manufacturing sector is defined as NACE Rev. 2 codes 10-33, while the tradeable service sector is defined as NACE Rev. 2 codes 58-63, 66 and 69-82. Data are not available for the financial and insurance industries (NACE Rev. 2 codes 64 and 65).

Trade in digital services – the fastest-growing segment – is less affected by conventional restrictions such as barriers to foreign entry and the movement of people. However, all services that are traded digitally can be constrained by the quality of digital infrastructure and connectivity, cross-border payment systems, intellectual property rights and other barriers, regardless of the sector. In order to take account of these issues, the OECD compiles the Digital Services Trade Restrictiveness Index, which covers 90 countries, including 17 economies in the EBRD regions.66

The sectors used in the OECD’s STRI database tend to be more detailed than those used in other databases. Table 2.1.1 shows the mapping used between the BaTIS, TiSMoS and STRI databases. Where more detail was available in TiSMoS, the average STRI score for a BaTIS sector was calculated using weights based on the value of exports in the relevant TiSMoS subsectors.

To distinguish between different types of FDI project within sectors, projects were categorised as either manufacturing-oriented or service-oriented investment. This categorisation was based on the specific function or purpose of each project as recorded in the FT fDi Markets database. Manufacturing-oriented projects were those explicitly listed as engaging in manufacturing activities, while service-oriented projects encompassed activities such as business services, customer contact centres, ICT infrastructure, logistics, R&D, and sales and marketing support.

Since 2010, Romania has had the highest labour productivity in computer and information services and the greatest revealed comparative advantage in that sector of any EBRD economy in the EU (see Chart 2.2.1). This advantage stems from two key factors: targeted policy interventions and strong human capital. Since the early 2000s, Romania’s computer and information service sector has undergone significant liberalisation. In 2001, a personal income tax break effectively reduced the tax rate for programmers to zero, down from 40 per cent. This policy was then broadened in 2013 to cover a larger portion of the computer and information service sector. Romania also liberalised its telecommunications sector in 2002.

A recent study found that the 2001 tax cut and its expansion in 2013 had significantly boosted growth in the computer and information service sector relative to other EBRD economies in the EU. Eligible firms had experienced substantial, long-lasting growth in employment and revenue. Moreover, downstream sectors dependent on computer and information services had also seen stronger growth, indicating that the policy had been effective in helping Romania to transition to a knowledge economy.68

The success of these industrial policy interventions was dependent on Romania having a well-educated population.69 Romania’s educational reforms in the 1970s and 1980s had strongly prioritised science and technology and laid the foundations for an educational system that channelled high-achieving students into specialist secondary schools at a young age.70

This box builds on existing studies and explores the importance of human capital for the success of industrial policies’ success. Computer and information service firms that were eligible to benefit from the 2013 tax reform grew faster than ineligible firms in the control group, while eligible firms located in NUTS-3 regions with above-median STEM-focused human capital endowments outperformed their counterparts in below-median regions (see Chart 2.2.2).71 Moreover, the income tax reform had a positive effect on FDI inflows in Romania’s computer and information service sector, with FDI projects increasing by an estimated 20 per cent relative to the computer and information service sectors of other EBRD economies in the EU, holding other factors constant.

As development policy shifts its focus from the promises of industrialisation to service-based growth, education is becoming increasingly crucial as an enabling factor for successful industrial policies. If economies are to accelerate structural transformation and lay the foundations for sustained economic growth in the 21st century, they need to expand access to education and build a skilled workforce. It remains to be seen whether Romania can continue this impressive growth, with the government revoking part of that income tax exemption in 2024 and a full rollback expected by 2028.

Over the past 15 years, Morocco has implemented a series of industrial strategies aimed at developing globally competitive manufacturing sectors, with a focus on the automotive and aerospace sectors. One of those initiatives was the 2009-15 National Industrial Emergence Plan (PNEI), which sought to create 220,000 jobs and increase exports by US$ 11 billion, primarily by attracting FDI, training 94,000 skilled workers and increasing cooperation between the public and private sectors in the target areas.73 This was followed by the even more ambitious 2014-20 Industrial Acceleration Plan (PAI), which more than doubled the employment target.74 In addition, the Moroccan government invested US$ 15 billion in infrastructure between 2010 and 2015, and established special economic zones in key locations such as Casablanca, Kenitra and Tangier.75

As a result, medium- and high-technology-intensive exports’ share of total goods exports rose from 33 per cent in 2009 to 65 per cent in 2022. In two of the target areas, the automotive and aerospace sectors, exports per capita rose, in real terms, by 2,390 per cent and 550 per cent, respectively, over that period (see Chart 2.3.1).76 Moreover, between 2008 and 2020 Morocco managed to achieve manufacturing export-led growth – something it had been unable to do between 1995 and 2008.

In order to separate the effect of Morocco’s policy interventions from the impact of concurrent global trends, Morocco’s export performance in the automotive and aerospace sectors is compared with synthetic controls – weighted averages for a group of economies (Algeria, Croatia, Egypt, Greece and Tunisia) that were similar to Morocco prior to the adoption of policies promoting those specific sectors. The synthetic control matches Morocco’s performance in terms of the average value of exports per capita (in constant US dollars) between 2002 and 2009, average GDP, average GDP per capita, trade in goods as a share of GDP, the manufacturing sector’s share of total value added, an indicator for EU membership and average years of schooling in 2005. This analysis confirms the exceptional nature of Morocco’s export performance in the target sectors. Similar results can be observed if the analysis is repeated at the HS4 level (separating automobiles from automotive components, for example).

While Morocco’s industrial policy has significantly boosted exports and employment in the target sectors, spillovers to the rest of the local economy have been limited. Few Moroccan firms have joined the country’s automotive clusters, with foreign-owned firms accounting for most of the automotive sector’s production and employment. Attempts to foster the integration of local small and medium-sized enterprises (SMEs) have been hindered by obstacles to investment and scalability, as well as the prevalence of informality. The Moroccan government is aware of these issues and is attempting to improve local sourcing in industrial ecosystems with its new 2020-25 Industrial Acceleration Plan (PAI2).77