This chapter examines the relationship between inward foreign direct investment (FDI) and local firms’ productivity and employment in Germany, Italy, and Romania. It focuses on how this relationship differs between small and large firms across different sectors and regions. The analysis suggests that FDI spillovers are strongest for small and medium-sized enterprises (SMEs) and when foreign firms connect with local industrial value chains and land in less developed regions. Differences in SMEs’ benefits from FDI across regions and sectors highlight the importance of coherent and integrated investment promotion, SME, and regional policies.
Connecting FDI and SMEs for Productivity and Innovation in Europe
Abstract
Countries and regions in the European Union – and the rest of the world – attract FDI with the expectation that it creates jobs and supports productivity growth through knowledge transfer and technology diffusion, benefiting local firms and fostering regional development. Productivity growth has slowed down in advanced economies in recent decades, with limited technology diffusion from the most productive and innovative firms to the rest of the economy. Less is known about the extent to which the activities of foreign firms, including in the European Union, generate knowledge spillovers – and through which transmission channels – and whether SMEs benefit from these spillovers and under what conditions.
This chapter examines the relationship between FDI and local firms’ productivity and employment in Germany, Italy, and Romania. FDI plays an important but uneven role in the three countries, which are characterised by high shares of SMEs and strong disparities in their geography and performance (see also Chapters 1 and 2). Germany combines the largest FDI inflows with a diversified industrial base and strong regional conditions, such as high-skilled labour, offering broad opportunities for spillovers into local firms. Italy’s linkage potential is concentrated in industrial hubs such as Lombardia and Emilia-Romagna. The prevalence of micro-enterprises, however, potentially constrains the capacity to scale these opportunities across the wider economy. Romania’s profile is different: investment is concentrated in services and in Bucharest-Ilfov, with weaker industrial anchors and more limited absorptive capacity in other regions.
In the three EU countries, the analysis shows that local firms in regions that attract more FDI record significantly higher total factor productivity levels, with effects stronger for SMEs than for large firms. Foreign investment is not associated with higher employment growth of local firms, however, indicating that foreign firms contribute more to efficiency gains than to additional employment generation. Schumpeterian selection mechanisms can be at play, where competitive pressures from foreign firms’ entry leads SMEs either to innovate – and possibly connect to global value chains (GVCs) – or exit.
The benefits of FDI on German, Italian, and Romanian SMEs’ performance depend on sectoral and regional conditions. Overall, they are strongest when foreign firms connect with local industries and in less developed regions that manage to attract investment. SMEs in industrial sectors show stronger productivity gains than in services, particularly when they are in sectors supplying inputs to foreign firms. FDI helps smaller firms scale up production, meet higher quality standards and access new markets. Large industrial firms also benefit from such linkages, but to a lesser extent, reflecting their stronger pre-existing role in GVCs. Importantly, less developed regions with high unemployment and low skills basis show stronger productivity spillovers when they attract FDI, underlining the potential for place-based policies to amplify benefits, including through well-targeted regional investment promotion policies.
Differences in the relationship between FDI and SME’s performance are marked across the three countries, although further investigation relying on higher quality data is needed. Italian SMEs show the most consistent productivity benefits from FDI. They record higher performance whether they are in the same sector as foreign firms or are suppliers (forward linkages) or customers (backward linkages) to these foreign firms. In Germany, SMEs benefit from own-sector and customer-sector FDI but face negative effects when foreign projects enter supplier industries. Romanian SMEs and large firms record fewer productivity gains, and in some cases negative effects, highlighting structural barriers to connecting with foreign investors. However, their employment levels are positively associated with foreign activity.
The High-Level Group on EU’s Future of Cohesion Policy calls for placing GVC participation and FDI attraction more centrally in investment priorities. It urges the creation of regional ecosystems that can absorb and leverage global knowledge flows by coupling local firms with international networks. Policy tools to operationalise this agenda could include dedicated regional linkage programmes and investment-promotion measures that act as matchmakers between foreign investors and local suppliers, alongside supplier-development support to upgrade SME capabilities. This approach aligns with the EU’s evolving focus on competitiveness and open strategic autonomy, recognising that resilient, diversified supply chains and innovation diffusion depend on engaging all regions – and their SMEs – in global networks.
Well-designed investment promotion strategies can help align FDI with regional specialisations and value chains, reinforcing productivity gains for SMEs. Evidence shows that SMEs benefit more when FDI is aligned with sectoral specialisations and value chains.
Policies that support knowledge transfer and value-chain integration can strengthen linkages between foreign investors and local SMEs. This includes supplier-development and matchmaking programmes, inter-firm collaboration, and innovation ecosystems.
Place-based strategies remain essential. Investments in skills, infrastructure and institutional capacity are particularly important in less developed regions to maximise the knowledge spillovers from FDI.
Monitoring and evaluation are key to assess policy impact. Systematic evidence on linkage effectiveness is limited; Regular monitoring and rigorous evaluation – underpinned by linked firm-level microdata on SMEs’ buyer–supplier networks, trade intensity, and GVC participation – are needed to pinpoint approaches that deliver the greatest productivity and employment gains.
Strengthening FDI spillovers is critical for supporting technology diffusion, productivity growth and regional convergence in the European Union. FDI serves as a relevant channel for productive capital inflow, skills and knowledge transfer as well as technology diffusion for host national and regional economies, benefiting domestic firms and fostering regional development when well-embedded in the local economy and coupled by the necessary local absorbative capacity. Thus, beyond its direct contributions to investment and job creation, FDI has the potential to support productivity growth. Productivity growth has slowed down significantly in advanced economies in recent decades (Goldin et al., 2024[1]; Rodríguez-Pose and Ganau, 2021[2]) and research highlights the limited technology diffusion from frontier firms to the rest of the economy (Andrews, Criscuolo and Gal, 2015[3]; Crescenzi, Dyèvre and Neffke, 2022[4]).
Adoption of new technologies is particularly challenging for SME as shown by significant technology gaps relative to larger firms (OECD, 2017[5]; OECD, 2023[6]). Yet, SMEs account for around a third of total employment and around 3% of business sector value added in the European Union. Therefore, bolstering FDI and SME linkages can potentially contribute to building stronger SME ecosystems that, in turn, could facilitate knowledge and technology diffusion and productivity gains for the wider economy and especially in less advanced regions where smaller firms are often over-represented (OECD, 2024[7]).
Knowledge spillovers from FDI can occur through multiple channels. Direct linkages with multinational enterprises (MNEs) enable domestic firms to access advanced technologies, management practices, and operational know-how. These differences can take different forms: backward linkages, when local firms act as suppliers to MNEs; forward linkages, when MNEs act as suppliers to domestic firms; and horizontal linkages, when domestic firms operate in the same sector as MNEs and may learn through demonstration or collaboration (Blomström and Kokko, 1998[8]; Javorcik, 2004[9]; Newman et al., 2015[10]) Indirect mechanisms, such as competition, demonstration effects, and labour mobility, may encourage domestic firms to optimise resource use and adopt new technologies (Blomström and Kokko, 1998[8]). However, FDI exposure may also lead to negative crowding-out effects that occur when foreign firms compete with incumbents for market shares, skilled labour, finance, and other resources (Barry, Gorg and Strob, 2005[11]; Jude, 2015[12]; Nguyen, Sun and Welters, 2024[13]). The mixed results on the impacts of inward FDI on their host economies are generally attributed to competitive pressure that may crowd out domestic firms. often less able to benefit from supplier relationships, innovative opportunities or competitive pressures. The crowding out effects of FDI entry can be particularly acute for smaller and less competitive firms.
Evidence indicates that FDI spillovers are not automatic. They depend on firm-to-firm interactions and firms’ absorptive capacity (Girma, Görg and Pisu, 2008[14]; Alfaro-Urena, Manelici and Vasquez, 2019[15]; Amiti et al., 2024[16]). For SMEs, the realisation of spillover benefits more likely depends on enabling conditions and targeted policy interventions, as they often face greater barriers to forming linkages with foreign subsidiaries, and have limited capacity to adopt new technologies or compete through investments in research and innovation – this is especially true compared to their larger domestic counterparts (Crescenzi, Dyèvre and Neffke, 2022[4]; Crescenzi and Harman, 2023[17]). Therefore, FDI spillover mechanisms are likely to be different for SMEs as opposed to large firms. Small firms may derive limited gains from FDI spillovers, as they are less likely to benefit from supplier relationships, less prepared to engage in innovation, and more vulnerable to heightened competitive pressures than larger firms. By contrast, medium-sized firms are better positioned to capture positive spillover effects from FDI (Lembcke and Wildnerova, 2020[18]).
This chapter examines FDI-SME linkages and their association with productivity and employment in selected EU countries and regions. Despite extensive research on FDI spillovers, evidence focusing on SMEs remain limited, leaving a significant gap in understanding the effect of FDI on SME ecosystems. This chapter provides new evidence, focussing on Germany, Italy and Romania - three economies with strong SME presence and pronounced regional disparities. The analysis compares SMEs with large firms, differentiates between industry and services, and explores the role of regional characteristics in shaping outcomes. It looks beyond direct job effects to assess how exposure to foreign investment in the same sector, in supplier industries or in customer industries is associated with firm performance.
The analysis draws on a representative sample of more than 24 000 firms. The sample links firm-level information on productivity and employment with data on regional greenfield FDI and other subnational indicators (see Box 3.1). Further methodological detail and descriptive statistics available the Annex 3.A. The approach is descriptive rather than causal, but it identifies consistent patterns that shed light on how and when SMEs benefit most from inward FDI.
This chapter uses firm-level evidence to explore how inward FDI relates to productivity and employment outcomes in Germany, Italy, and Romania. The analysis relies on a tailored dataset combining information from three sources:
Firm characteristics (Orbis, Bureau van Dijk). A representative sample of 24 220 firms in industry and services was drawn for the period 2015–2023. To ensure comparability, the sample covers around 5% of firms, stratified by sector, size, and region, using benchmarks from Eurostat’s Structural Business Statistics (SBS). Available variables include employment, total factor productivity (TFP) - estimated using robust methods (Levinsohn and Petrin, 2003[19]) - firm age, sector of activity (two-digit), and location (NUTS-1 for Germany; NUTS-2 for Italy and Romania).
Regional characteristics (Eurostat Regio). Firm-level observations were enriched with subnational indicators such as GDP per capita, population density, tertiary education share, unemployment rate, and the sectoral composition of employment.
Regional greenfield foreign investment (fDi Markets, Financial Times). Projects recorded between 2011 and 2023 were used to construct three measures of inward foreign investment:
Own-sector FDI: projects in the same region and sector as the firm.
Backward linkages: projects in the same region as the firm and in its supplier sectors, weighted using national input–output tables.
Forward linkages: projects in the same region as the firm and in its customer sectors, also weighted with input–output tables.
This combined dataset offers a unique perspective on how foreign investment interacts with different types of firms across diverse regional contexts. While the evidence provides rich descriptive insights, it should be interpreted with caution: the associations identified are not causal. Further methodological detail and descriptive statistics are presented in the Annex 3.A.
The three EU countries are attractive destinations for foreign investors, though Germany dominates, and regional variation is substantial. Between 2011 and 2023, more than 5 600 greenfield projects were recorded across Germany, Italy, and Romania. Germany accounted for 3 680 projects (65%), confirming its role as the main entry point for foreign investors in the sample. Italy and Romania received 1 003 (18%) and 954 (17%) projects, respectively.1 Annual trends were relatively stable, with all three countries experiencing a sharp dip in 2020 during the COVID-19 pandemic followed by recovery in 2021 (Figure 3.1, Panel A). Within each country, projects are highly concentrated in a few leading regions (see also Chapter 1). Lombardia alone hosted one-third of Italy’s projects, while Bucharest-Ilfov captured nearly 30% of Romania’s total. By contrast, foreign investment in Germany was more evenly distributed, though major regions such as Nordrhein-Westfalen, Bayern and Baden-Württemberg stand out as leading destinations (Figure 3.1, Panel B).
Source: Authors’ calculations based on fDi Markets
FDI is concentrated in services and manufacturing, with primary activities playing only a marginal role. Between 2011 and 2023, most projects in Germany, Italy and Romania were recorded in services and manufacturing sectors, reflecting broader European specialisation patterns. Agriculture and mining attracted only a handful of investments, almost all in a few German regions, with Italy and Romania registering only isolated cases, making clear the diminished role of primary sectors as destinations for foreign projects (Figure 3.2, Panel A). By contrast, manufacturing investment was substantial and geographically anchored in major industrial hubs: Nordrhein-Westfalen, Bayern and Baden-Württemberg in Germany, Lombardia and Emilia-Romagna in Italy, and several regions in Romania at lower intensity (Figure 3.2, Panel B). Services accounted for the largest overall share of projects, especially in wholesale and retail trade, transport and ICT (Figure 3.2, Panel C). These were more widely distributed than manufacturing but strongly concentrated in capital-city regions such as Lombardia, Lazio and Bucharest-Ilfov, which act as gateways for multinational service activities. These sectoral trends are highly relevant for SMEs, as opportunities to link with foreign investors are typically greater in manufacturing value chains and high-productivity services than in primary or low-productivity activities.
Source: Authors’ calculations based on fDi Markets.
SMEs form the backbone of the business population, though structures differ markedly across countries. In Italy, the business landscape is dominated by micro firms, which account for two-thirds of all enterprises, with small firms representing a further 29% (Figure 3.3, Panel A). Germany presents a more balanced profile: only 4% of firms are micro, while over half are medium-sized and more than a quarter are large. This reflects Germany’s tradition of Mittelstand enterprises with strong industrial roots. Romania lies between these two extremes: the majority of firms are micro (78%), with 17% small and just 4% medium-sized, leaving large enterprises at less than 1%. As these figures come from an estimation sample rather than the full firm population, they should be read as stylised patterns.
Sectoral composition varies across the three countries, with notable differences between industry and services. In Germany, industry accounts for just over half of all firms, split between manufacturing (35%) and utilities and construction (16%) (Figure 3.3, Panel B). Services are also significant, with wholesale and retail trade alone representing 27% of firms. Italy shows a more even balance: industry makes up 41% of firms, while services represent 59%, shaped by large shares in retail (27%) and other market services (24%). Romania, by contrast, is overwhelmingly service-oriented: more than 74% of firms are in services, led by retail (34%) and transport (11%), while industry accounts for only one-quarter. Primary activities such as agriculture and mining are excluded from the firm-level dataset given limited coverage in Orbis, and in any case represent only a marginal share of inward FDI projects in the three countries.
Note: Figures refer to the estimation sample of 24 220 firms used in the analysis. See Annex 3.A for details on sampling and representativeness.
Source: Authors’ calculations based on Orbis data.
Smaller firms face persistent performance gaps compared to larger enterprises. Micro firms employ fewer than four workers on average and record the lowest productivity levels across all three countries (Table 3.1). By contrast, large firms employ more than 500 workers on average and display productivity levels more than twice as high as those of micro firms. These gaps are mirrored in firm age: micro enterprises average around 28 years, while large firms average more than 60 years. The patterns are consistent across Germany, Italy, and Romania and underline the challenges faced by SMEs in upgrading, adopting new technologies, and linking into international value chains. Structural differences in productivity, scale and maturity mean that smaller firms are less well placed to benefit from foreign investment without targeted support.
Note: Approximate values are derived by transforming logged means back to levels. They are rounded to the nearest whole number for employment and age, and to two decimals for TFP. Figures should be interpreted as indicative averages across the sample rather than exact measures.
Source: Authors calculations based on Orbis.
Large SME presence in Germany, Italy, and Romania, combined with stark contrasts in their industrial structures and regional conditions, provides fertile grounds to examine how FDI influences local firms’ productivity and employment. Germany combines strong inflows of foreign projects with a diversified industrial base and highly skilled workforce, offering broad scope for spillovers. Italy, by contrast, shows more concentrated opportunities in hubs such as Lombardia and Emilia-Romagna, where the dominance of micro-enterprises constrains the capacity to scale linkages across the wider economy. Romania presents yet another profile, with investment concentrated in services and Bucharest-Ilfov, and weaker industrial anchors in other regions limiting absorptive capacity.
This section provides evidence on how inward FDI relates to productivity and employment among SMEs and large firms in the three EU countries. The analysis examines regional-level patterns in firm demography, employment, and productivity, and, using firm-level data, explores three channels through which FDI can influence local firms’ performance: (1) own-sector effects (foreign investors entering the same sector as domestic firms), (2) backward linkages (foreign investors entering sectors that supply inputs to domestic firms), and (3) forward linkages (foreign investors entering sectors that purchase outputs from domestic firms). Sectoral and regional features are considered. Results are correlational rather than causal, but they highlight patterns that can guide the design of more targeted policy interventions and future evaluations. The consolidated database underlying this work is described in Box 3.1.
3.4.1. SMEs capture the largest productivity spillovers, while employment effects are limited
Regional-level results show a positive link between inward FDI and productivity, but no systematic relationship with employment. Regions with higher volumes of foreign projects tend to host local firms with higher average total factor productivity (Figure 3.4). Only SMEs appear to benefit, however, potentially due to smaller productivity gaps between foreign firms and larger companies. By contrast, FDI do not translate into faster growth in the number of firms, shifts in their size distribution, or stronger employment growth. This holds across all firm size classes, suggesting that FDI is not correlated with the demography of firms in terms of employment or scale.2 The results are consistent with competitive pressures that raise the bar for firms: weaker firms adjust or exit, while more productive SMEs improve their efficiency.
Note: Bars show the estimated association between inward FDI and regional outcomes (productivity and employment), disaggregated by firm size (SMEs vs. large firms). Estimates are based on log–log fixed-effects panel regressions with regional controls, region and year fixed effects. Only full bars are statistically significant at the 10% level or below. As the analysis is correlational, magnitudes should be interpreted with caution. See Annex 3.A for the empirical methodology and the main empirical results.
Source: Authors’ calculations based on Orbis and fDI Markets.
3.4.2. Productivity spillovers are strongest when foreign buyers’ source from local SMEs
SMEs record the clearest productivity improvements when foreign investors purchase their outputs, while employment effects remain weak. SMEs benefit most from demand-side channels: productivity gains are strongest when foreign investors buy their outputs, with smaller but still positive effects when investors enter their own sector (Figure 3.5). These linkages help smaller firms expand production, meet higher quality standards, and access new markets. Large firms also benefit from foreign customers, though to a lesser extent given their stronger position in supply chains. By contrast, employment effects are modest. Only micro and large firms record marginal gains when foreign investment enters their input-supplying industries, while most SMEs show no systematic job creation.3 This suggests that FDI is more effective in supporting efficiency upgrades than in driving employment expansion.
Note: Bars show the estimated association between inward FDI, and firm total factor productivity and employment, disaggregated by firm size (SMEs vs. large firms) and linkage channel. Estimates are based on log–log fixed-effects panel regressions with regional controls, region and year fixed effects. Only full bars are statistically significant at the 10% level or below. As the analysis is correlational, magnitudes should be interpreted with caution. See Annex 3.A for the empirical methodology and the main empirical results.
Source: Authors’ calculations based on Orbis, fDI Markets, and EU Regio data.
3.4.3. Spillovers are concentrated in industrial SMEs, especially through foreign buyers
FDI spillovers to productivity are stronger in industrial sectors than in services, concentrated among SMEs, and mainly transmitted through forward linkages. Industrial sectors shows the broadest and most consistent associations, reflecting the role of industrial value chains as conduits for knowledge and technology transfer (Alfaro-Urena, Manelici and Vasquez, 2019[15]; Javorcik, 2004[9]) (Figure 3.6). SMEs benefit most, as they typically have greater scope to upgrade practices and absorb new technologies compared to large firms already closer to the productivity frontier (Gorg, 2004[20]; Crespo and Fontoura, 2007[21]). The strongest effects appear through forward linkages, where foreign buyers demand higher quality and reliability from local suppliers, prompting SMEs to raise standards and improve efficiency (Javorcik, 2004[9]; Blalock and Gertler, 2008[22]). By contrast, own-sector FDI is least associated with productivity gains: while it heightens competition, this often pressures weaker firms to exit rather than directly boosting the productivity of survivors (Aitken and Harrison, 1999[23]; Smeets, 2008[24]). In services, spillovers are more limited, in part because many activities are less tradable and more fragmented, offering fewer opportunities for sustained linkages with foreign investors (Alfaro, 2003[25]). Even within industrial sectors, spillovers appear strongest in capital-intensive industries, where integration into global value chains provides local SMEs with greater opportunities to absorb technology and process upgrades. Employment effects remain negligible overall, with only isolated positive associations - for instance, among SMEs in services through backward linkages.4
Note: Bars show the estimated association between inward FDI and firm total factor productivity, disaggregated by sector (manufacturing vs services), firm size (SMEs vs. large firms), and linkage channel. Estimates are based on log–log fixed-effects panel regressions with regional controls, region and year fixed effects. Only full bars are statistically significant at the 10% level or below. As the analysis is correlational, magnitudes should be interpreted with caution. See Annex 3.A for the empirical methodology and the main empirical results.
Source: Authors’ calculations based on Orbis and fDI Markets.
3.4.4. Italy shows clear productivity gains, Germany has mixed results with some negative effects, and Romania shows weak outcomes overall
The country comparison shows that linkages between inward FDI and firm productivity vary markedly across contexts. Italy exhibits the most systematic spillovers: SMEs benefit across own sector, backward, and forward linkages, and large firms show positive associations when foreign investors act as customers (Figure 3.7). This breadth of channels suggests that Italian firms - particularly SMEs - are well positioned to connect to global value chains and upgrade through demand from foreign buyers. Germany, by contrast, records fewer and more selective spillovers. SMEs show some positive effects in their own sector but also negative associations when foreign investment enters their input-supplying industries, pointing to competitive pressures in already highly efficient markets. Romania presents the weakest pattern: associations are generally small and not statistically significant, reflecting more limited integration of domestic firms into foreign investment networks. Overall, the results suggest that national conditions -including industrial structure, competitive dynamics, and SME absorptive capacity - shape the extent and direction of productivity spillovers from FDI. Employment effects remain weak: Italy shows some positive associations for large firms in their own sector, Romania records negative effects for SMEs in forward-linked industries and large firms in backward-linked industries, while Germany shows little evidence of systematic employment impacts.5
Note: Bars show the estimated association between inward FDI and firm total factor productivity, disaggregated by country, firm size (SMEs vs. large firms), and linkage channel. Estimates are based on log–log fixed-effects panel regressions with regional controls, region and year fixed effects. Only full bars are statistically significant at the 10% level or below. As the analysis is correlational, magnitudes should be interpreted with caution. See Annex 3.A for the empirical methodology and the main empirical results.
Source: Authors’ calculations based on Orbis and fDI Markets.
Finally, regional conditions shape the extent of spillovers. SMEs and large firms in less developed regions - those with higher unemployment, weaker skills, and less diversified economies - show relatively stronger productivity benefits when they succeed in attracting inward FDI. For firms in these regions, foreign investors provide an important channel for accessing knowledge, markets and international networks that would otherwise remain out of reach. This highlights the role of place-based strategies: the same investment can have very different effects depending on whether it lands in a strong industrial hub with abundant skills, or in a weaker region struggling with structural barriers.
Overall, FDI contributes more to efficiency than to scale, with spillovers shaped by firm size, sector, and place. Productivity gains are concentrated among SMEs, especially in industrial sectors, where forward linkages with foreign buyers create the strongest incentives for local firms to upgrade. Country differences underline the role of national conditions: Italy shows the broadest and most consistent spillovers, Germany more selective and sometimes negative effects, while Romania records weaker and less systematic outcomes. Regional context is also important: less developed regions with higher unemployment, weaker skills, and less diversified economies tend to benefit more when they succeed in attracting FDI, as foreign investors provide critical access to knowledge, markets, and international networks. Taken together, the results suggest that spillovers are not automatic but conditional - depending on sectoral match, the type of investment, and the absorptive capacity of firms and places.
Policies to make FDI work for EU SMEs and regions strongly matters. The evidence highlights the importance of embedding FDI into EU’s regional ecosystems through GVCs, particularly for SMEs that often face greater barriers to upgrading. MNEs rarely embed themselves in local economies without deliberate effort. For SMEs, the ability to benefit depends on their capacity to link into value chains, adopt new technologies, and adapt to competitive pressures. Without targeted support, the risk is that FDI boosts efficiency but fails to create broad-based opportunities for local firms or workers. Place-based strategies that combine investment promotion with skills development, infrastructure upgrading and institutional capacity building are critical to maximising impact.
Several practical policy instruments can help translate foreign investment into wider regional gains:
Local Content Units (LCUs): Flexible matchmaking services, typically housed within investment promotion agencies, that connect MNEs with suitable local suppliers. Unlike rigid local content requirements, LCUs address information gaps and open new opportunities for SMEs to participate in GVCs (Crescenzi and Harman, 2023[17]).
Linkage programmes: Dedicated initiatives that support the upgrading of domestic suppliers through a variety of targeted interventions, including training, technical assistance, and even secondments of engineers or high-skilled workers from MNE into local firms.
Supplier mapping and brokerage: Platforms that identify and connect SMEs with the sourcing needs of foreign investors. Evidence from international practice shows these services can deliver lasting improvements in SME productivity and profitability.
Regional investment promotion with aftercare: Subnational agencies can target projects that align with regional specialisations and local strengths. Through aftercare services, they can also ensure that foreign investors are embedded in regional value chains and innovation systems.
These policy tools can be particularly relevant in less developed EU regions, where this chapter finds comparatively stronger FDI–productivity associations. By combining proactive investment promotion with LCUs, supplier development, and local absorptive capacity building, regions can maximise the potential productivity gains from inward investment for SMEs. Several countries have successfully used these tools to embed MNEs in local economies, such Ireland, Singapore and Costa Rica (see Box 3.2).
Targeted programmes can help translate inward investment into broader productivity gains. Several countries have successfully embedded multinationals in local economies by supporting linkages with domestic SMEs. Three widely cited examples illustrate different approaches:
Ireland - National Linkages Programme
Established by IDA Ireland in the late 1980s to deepen the integration of foreign investors.
Focused on supplier upgrading, with training and technical assistance for SMEs to meet multinational standards.
Helped build Ireland’s capacity in electronics and pharmaceuticals by connecting local firms to global value chains.
Singapore - Local Industry Upgrading Programme (LIUP)
Launched in the 1980s to foster long-term partnerships between MNEs and local SMEs.
Featured secondments of multinational engineers into domestic firms to transfer know-how and improve quality.
Later evolved into broader supplier development schemes under Enterprise Singapore, sustaining high-value linkages.
Costa Rica – Provee Supplier Development Programme
Provides supplier mapping, certification, and brokerage to match SMEs with foreign investors.
Has facilitated over 1 300 SME–MNE linkages, particularly in the medical devices and electronics sectors, with measurable gains in SME productivity and export capacity.
These experiences show the importance of dedicated institutions and programmes. By reducing information gaps, upgrading local capabilities, and brokering trust between SMEs and multinationals, linkage initiatives can support foreign investment delivers durable benefits for host economies.
The High-Level Group on the Future of Cohesion Policy calls for placing GVC participation and FDI attraction more centrally within future investment priorities, urging the creation of regional ecosystems that can absorb and leverage global knowledge flows by coupling local firms with international networks (European Commission, 2024[26]). To operationalise this agenda Cohesion Policy instruments could include dedicated regional linkage programmes and investment-promotion measures (Crescenzi, Di Cataldo and Giua, 2021[27]) that act as matchmakers between foreign investors and local suppliers, alongside supplier-development support to upgrade SME capabilities. This approach aligns with the EU’s evolving focus on competitiveness and open strategic autonomy, recognising that resilient, diversified supply chains and innovation diffusion depend on engaging all regions - and their SMEs - in global networks.
More research is needed to fully assess the contribution of FDI to SMEs’ performance in EU countries and regions and evaluate related policy tools, including Cohesion Policy tools. To move beyond descriptive patterns, more robust evaluations are required. Quasi-experimental approaches and richer subnational datasets would help test which policy tools - such as LCUs, linkage programmes or aftercare - generate measurable productivity improvements for SMEs under different regional conditions. Joint efforts by EU countries should enable interoperable firm data and richer project-level FDI data to be linked with VAT/e-invoicing networks, trade microdata (goods and services), and employer–employee records. These should be combined with high-resolution regional industry data (that are still limited) and explicit policy tags in federated data rooms. Targeted SME surveys could effectively complement administrative sources to evaluate spillovers and upgrading pathways.
A summary of the dataset is provided in Box 3.1. This annex provides further methodological detail on data sources, cleaning, sampling, and variables construction. It also outlines how foreign direct investment (FDI) measures were derived and provides descriptive statistics.
The firm-level data used in the empirical analysis are drawn from the Orbis database (Bureau van Dijk), which provides balance sheet data and personal information for firms worldwide. The original downloaded sample of 1 466 983 German, Italian, and Romanian firms has been cleaned to consider only active industrial and services firms reporting unconsolidated financial statements.6 Firms without information on incorporation year, geographical location at the subnational level – defined according to the European Union (EU) Nomenclature des Unités Territoriales Statistiques (NUTS) – and sector of activity defined at the two‐digit level of the EU NACE Rev. 2 Classification have been removed. The sample has been cleaned also by culling firms reporting missing figures for tangible fixed assets and depreciations over the period 2014–2023 to estimate firm‐level variables for real investments in tangible fixed assets and capital stock for the years from 2015 to 2023. The resulting sample has been further polished by considering only firms reporting strictly positive figures for value added, total labour cost, intermediate inputs, and employment for at least three consecutive years during the period 2014–2023. The cleaning procedure left a sample of 484 391 firms.
As shown in the first two columns of Annex Table 3.A.1, the cleaned sample shows very little cross-country representativeness. Therefore, following (Rodríguez‐Pose et al., 2020[28]), the final estimation sample has been obtained by randomly drawing a 5% of firms stratified to reflect both absolute cross‐country representativeness and relative within‐country representativeness in terms of two‐digit NACE Rev. 2 sector, subnational regional geography defined according to the NUTS classification, and size class with respect to official figures derived from the Structural Business Statistics (SBS) provided by the European Statistical Office (Eurostat). In particular, the subnational geographical unit of analysis considered varies across countries between the level 1 (i.e., Germany) and the level 2 (i.e., Italy and Romania) of the NUTS regional classification: the underlying rationale is to consider regions with an effective devolved power to implement investment promotion policies and influence the economic performance of local firms in each specific country.7 Firm size classes are defined according to the EU Recommendation No. 2003/361 as follows: micro firms, with a number of employees strictly lower than 10; small firms, with a number of employees ranging in the interval ; medium firms, with a number of employees ranging in the interval ; and large firms, with a number of employees equal to or greater than 250.8
The randomized selection procedure has resulted in a cleaned final sample consisting of an unbalanced panel dataset of 24 220 firms observed for at least three consecutive years over the period 2014–2023—see the last two columns of Annex Table 3.A.1. The firm-level dataset includes firms: belonging to all the four size classes identified by the EU Recommendation No. 2003/361 (Annex Table 3.A.2); located in all German NUTS-1, and Italian and Romanian NUTS-2 regions (Annex Figure 3.A.1 and Annex Table 3.A.3); and active in the two-digit industrial sectors 10 to 43 and the two-digit services sectors 45 to 82 (Annex Table 3.A.4).9
Note: Randomisation based on SBS data (Eurostat) with respect to cross-country representativeness and within-country (i) sector, (ii) region, and (iii) size class distributions.
Note: (c) refers to percentage values defined on column totals; (r) refers to percentage values defined on row totals.
Source: Authors’ calculations based Orbis data.
Note: Percentage values defined on country totals.
Note: Percentage values defined on country totals.
This section outlines the key variables used in the analysis. The firm‐level unbalanced panel dataset of 24 220 firms observed over the period 2015–2023 includes the following time-varying information to be used in the empirical analysis: Total Factor Productivity (TFP), employment level, and age defined as observation year minus year of a firm’s incorporation. It also includes information on the region in which a firm is located and its two-digit NACE Rev. 2 sector of activity. Annex Table 3.A.5 reports some descriptive statistics of the (log-transformed) time-varying firm-level variables by country, while Annex Table 3.A.6 reports some descriptive statistics of the same (log-transformed) variables by firm size class.
Note: All variables are log-transformed.
Note: All variables are log-transformed.
Investment and capital stock
The cleaned sample of 24 220 firms observed for at least three consecutive years over the period 2014–2023 has been used, first, to calculate the firm-level variables for real investments and capital stock. The variable for firm-level real investments in tangible fixed assets () for firm operating in sector , located in country , and observed in year has been defined as follows:
Equation 1
where the term denotes the book value (BV) of tangible fixed assets, the term represents the book value of depreciations, and the term conveys a sector- and country-specific investments price deflator provided by Eurostat. The capital stock of firm at the beginning of year has been defined as the difference between its capital stock at the end of year () and its capital expenditure during year , with capital stock defined using the Perpetual Inventory Method (PIM) as follows:
Equation 2
where the term denotes the depreciation rate. Therefore, firm-level capital stock estimates have been obtained for the years 2015 to 2023.
Total factor productivity
The cleaned sample of 24 220 firms observed over the period 2015-2023 has then been used to estimate firm-level TFP. Specifically, the TFP of a firm has been estimated as the residual of a Cobb–Douglas production function specified as follows in logarithmic form:
Equation 3
where the term represents the mean efficiency level across firms and over years; the term denotes the value added of firm in year ; the terms and denote the capital and labour input, respectively; the term represents productivity shocks potentially observed or that can be predicted by the firm when making inputs’ decisions, and thus influencing its decision process; and the term is an independent and identically distributed component which represents productivity shocks not affecting a firm’s decision process (Olley and Pakes, 1996[29]; Van Beveren, 2007[30]). Hence, the estimated firm-level TFP can be computed solving for as follows:
Equation 4
Firm-level TFP has been estimated through the two-step semiparametric approach proposed by (Ketterer and Rodríguez‐Pose, 2018[31]), which uses intermediate inputs () to proxy for unobserved productivity and solving the simultaneity problem between input choices and productivity shocks. Firms’ TFP has been estimated by country at the industry level using deflated balance sheet data on value added, total labour cost, and intermediate inputs. Value added has been deflated with the corresponding national production price index and has been used as output variable in the production function; total labour cost has been deflated with the corresponding national wage index and has been used as labour input; intermediate inputs—defined as the sum of services, raw materials, and consumptions—have been deflated with a national intermediate consumptions index; the capital input has been defined as the capital stock computed through the PIM using a national gross fixed capital deflator. Therefore, TFP estimates have been obtained for an unbalanced panel dataset of 24 220 firms observed over the period 2015-2023.
The firm-level dataset has been then integrated with region‐specific data for the period 2015–2023 drawn from the Regio database provided by Eurostat, namely: Gross Domestic Product (GDP) defined in purchasing power standards (PPS); population; land surface; share of population aged 25–64 years with tertiary education to capture human capital endowment; unemployment rate defined as the percentage of the unemployed population aged 15–74 years; share of employment in high-tech sectors; share of employment in manufacturing sectors; and share of employment in services sectors. In particular, GDP and population figures have been used to calculate a variable for regional GDP per capita, while population and land surface figures have been used to calculate a variable for regional population density. Annex Table 3.A.7 reports some descriptive statistics of the (log-transformed) time-varying region-level variables used in the empirical analysis by country.
Note: All variables are log-transformed.
This section presents the construction of Foreign Direct Investment (FDI) exposure variables. Data on inward FDI are drawn from the fDi Markets database (Financial Times), which provides information on greenfield inward (and outward) investment projects in terms of year of realisation, source and destination region, sector defined at the four-digit level of the North American Industry Classification System (NAICS), monetary value of the project in US Dollars (USD), and number of jobs created through the investment. This wealth of information makes the fDi Markets database one of the best available sources to analyse FDI-related phenomena at the sector and subnational level over time. The analysis of the association between regional inward FDI and firm-level performance is based on three FDI variables defined in terms of the number of projects realized in a region over a five-year window (i.e., over a period to ). Therefore, given that the firm-level estimation sample includes firms observed over the period 2015–2023, only inward FDI projects realised in the period 2011–2023 have been considered to construct the three time-varying FDI variables. Annex Figure 3.A.1 reports the distribution of FDI projects realised in the period 2011–2023 by two-digit NACE Rev. 2 sector, while Annex Figure 3.A.3 displays the within-country regional distribution of FDI projects realized in the period 2011–2023. Annex Table 3.A.8 reports the distribution of FDI projects by destination country and realisation year, while Annex Table 3.A.9 reports the distribution of FDI projects by destination region. It is interesting to note how all the regions covered in the analysis experienced the realisation of FDI projects during the period considered, even though with some degree of sectoral heterogeneity: indeed, FDI in primary sectors has been realised only in few regions and mostly in Germany, while FDI in the industrial and services sectors has been realised across all the regions covered by the analysis. Moreover, it is worth noting how Germany has registered both the highest number of projects realised and the greatest internal regional variability.
The first FDI variable captures the own-sector effect of investments realised over the five-year window period , and is defined as follows:
Equation 5
where the term denotes the FDI project realised in sector and in region in year .
The second variable captures the sector specific backward-linkage dimension of inward FDI, and is defined as the weighted sum of FDI projects realised in sectors that are backward-connected to the reference sector over the period :
Equation 6
where the term is a weight parameter capturing the share of inputs that firms active in the reference sector buy from any other sector , with .10
Finally, the third FDI variable captures the sector specific forward-linkage dimension of inward FDI, and is defined as the weighted sum of FDI projects realised in sectors that are forward-connected to the reference sector over the period :
Equation 7
where the term is a weight parameter capturing the share of inputs that firms active in the reference sector sell to any other sector , with . Annex Table 3.A.10 reports some descriptive statistics of the three (log-transformed) FDI variables by country.
Source: Authors’ calculations based Orbis data.
Source: Authors’ calculations based Orbis data.
Note: Percentage values defined on column totals.
Note: Percentage values defined on country totals.
Note: All variables are log-transformed.
Several limitations should be noted:
Representativeness: Orbis data tend to overrepresent large firms and underrepresent SMEs. Stratified sampling mitigates but does not eliminate this bias.
Multi-establishment firms: One of the drawbacks of the Orbis database is that it does not allow for the identification of multi-establishment firms. This issue, in any case, is partially relaxed by the exclusion from the sample of firms reporting consolidated financial statements, as well as by the fact that approximately 66.57% of the sample is made of micro and small firms, which tend to be overwhelmingly mono‐establishment (Cainelli and Iacobucci, 2012[32]; Cainelli and Iacobucci, 2011[33]).
FDI coverage: fDi Markets records announced greenfield projects only; mergers and acquisitions (M&A) are excluded. However, the overall validity and reliability of the database have been affirmed by several empirical studies that have used it for the analysis of inward FDI (Crescenzi, Pietrobelli and Rabellotti, 2013[34]; Crescenzi and Ganau, 2024[35]) and outward FDI (Crescenzi and Iammarino, 2017[36]; Crescenzi, Ganau and Storper, 2021[37]). Furthermore, as highlighted by (Crescenzi, Ganau and Storper, 2021[37]), “scholarly papers as well as official reports by, among others, the European Commission and the United Nations Conference on Trade and Development (UNCTAD) are convergent in supporting the use of the fDi Markets database” for the regional analysis of FDI.
Regional variables: Eurostat provides harmonised indicators, but these may not capture local idiosyncrasies.
Interpretation: The resulting dataset should be read as providing stylised patterns rather than precise population-level estimates.
The firm-level variables for TFP and employment have been aggregated at the regional level to conduct a preliminary region-level analysis aimed at identifying some stylised facts. Firms’ TFP and employment figures have been aggregated at the level of the 44 regions constituting Germany, Italy, and Romania to study yearly variations over the period 2015–2023 with respect to the (log-transformed) total number of firms, the (log-transformed) total employment of firms, and the (log-transformed) average TFP of firms. The preliminary region-level analysis has been conducted by specifying the following equation:
Equation 8
where the term denotes the dependent variable for region in year ; the term denotes a variable capturing the number of inward FDI projects realised in region over a five-year window period ; the terms and denote region and year fixed effects (FE), respectively; the term denotes a vector of (log-transformed) time-varying regional control variables for GDP per capita, population density, unemployment rate, share of tertiary-educated population, share of employment in high-tech sectors, share of manufacturing employment, and share of services employment; and denotes the error term.
The results of the estimation of Equation 8 are reported in Annex Table 3.B.1. Looking at specifications (3), (6), and (9), it emerges the absence of any statistically significant effect of inward FDI independently from firms’ size class – that is, micro, small, and medium-sized enterprises (MSME) versus large firms. By contrast, a different picture emerges when considering the (log-transformed) regional average firms’ TFP in the bottom panel of Annex Table 3.B.1: the results suggest a positive and statistically significant association between inward FDI and TFP, which seems to be driven by MSMEs rather than by large firms. Overall, this preliminary evidence points to inward FDI contributing to a Schumpeterian selection mechanism in the case of MSMEs rather than to a size-related effect. By contrast, there seems to be no aggregate effect for large firms ascribable to inward FDI.
Note: Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
The cross-country firm-level analysis is based on the unbalanced panel dataset of 24 220 firms observed over the period 2015–2023, and considers two dependent variables to study the association between regional inward FDI and firm-level outcomes, namely firms’ TFP and employment. Formally, the analysis is based on the following equation:
Equation 9
where the term denotes the dependent variable – capturing either log-transformed TFP or log-transformed employment – for firm operating in sector and located in region in year ; the term denotes a vector of FDI variables capturing the own-sector, backward-linkage, and forward-linkage dimensions of inward FDI, respectively, referring to the reference sector and realised in region over a five-year window period ; the term denotes a vector of (log-transformed) time-varying firm-level control variables, namely firms’ employment (firms’ TFP) when the dependent variable if firms’ TFP (firms’ employment), and firms’ age; the term denotes a vector of (log-transformed) time-varying regional control variables for GDP per capita, population density, unemployment rate, share of tertiary-educated population, share of employment in high-tech sectors, share of manufacturing employment, and share of services employment; the terms and denote firm and year FEs, respectively; the term denotes a sector-specific time trend; and denotes the error term. It is worth noting that the results based on the estimation of Equation 9 should be interpreted as statistical associations. Indeed, such setting does not allow for a proper identification strategy leading to the estimation of a causal effect of inward FDI on firms’ TFP and employment. However, these non-causal results could be interpreted as sufficiently robust due to the inclusion of both firm-level FEs – controlling for firm-specific unobserved conditions – and a sector-level time trend – accounting for sector-specific factors – in Equation 9.
Annex Table 3.C.1 reports the main results obtained from the estimation of Equation 9 on the whole sample of firms as well as on the subsamples of micro, small, medium, MSMEs, and large firms. Moreover, as a preliminary analysis, the vector of sector-specific regional inward FDI variables has been replaced by a variable capturing the regional aggregate of inward FDI projects realised over a five-year window period . Looking at the upper panel of Annex Table 3.C.1, it clearly emerges that inward FDI matters for MSMEs’ TFP to a greater extent than for large firms’ TFP. In particular, there is evidence of MSMEs’ TFP benefitting from inward FDI – entering the regional ecosystem – realised in the own sector and, to a greater extent, in forward-connected sectors. By contrast, large firms’ TFP seems to benefit from inward FDI realised in forward-connected sectors, despite to a much lower extent than in the case of MSMEs. Looking at the bottom panel of Annex Table 3.C.1, the results suggest an overall negligible association between inward FDI and firms’ employment. Indeed, it seems that only micro and large firms experience some marginally significant benefit in terms of employment from inward FDI in backward-connected sectors. Overall, these results corroborate the preliminary regional level evidence presented previously in Annex Table 3.B.1: that is, inward FDI contributes to a Schumpeterian selection mechanism rather than to a size-related effect, and this seems to be the case particularly for MSMEs compared to large firms.
Various possible sources of heterogeneity have been investigated to disentangle the association between inward FDI and firms’ outcomes.
First, Equation 9 has been estimated separately for industrial (i.e., sectors 10 to 43) and services (i.e., sectors 45 to 82) firms. As shown in Annex Table 3.C.2, the results corroborate the main evidence of inward FDI benefitting mostly MSMEs’ TFP. However, they also suggest that this is the case especially for MSMEs active in industrial sectors, whose TFP benefits from inward FDI realised in the own sector, in backward-connected sectors, and in forward-connected sectors. By contrast, services MSMEs’ TFP seems to benefit from FDI realised in forward-connected sectors only. Looking at industrial large firms’ TFP, it seems that it benefits from inward FDI realised in forward-connected sectors only; by contrast, there is no evidence of further FDI returns in the case of industrial and services large firms. Finally, it seems that only services MSMEs benefit from inward FDI realised in backward-connected sectors in terms of employment.
Second, heterogeneity related to the industrial versus services sector of activity of MSMEs and large firms has been exploited by considering an inward FDI variable capturing the regional share of investment projects realised in capital-intensive sectors over a five-year window period .11 As shown in Annex Table 3.C.3, the results suggest that a relatively higher regional share of capital-intensive inward FDI is positively associated with industrial MSMEs’ TFP only.
Third, Equation 9 has been augmented with a series of interaction terms between each inward FDI variable and a categorical variable capturing firms’ location country to assess potential country-specific heterogeneity. Annex Table 3.C.4 reports the estimated coefficients of the three inward FDI variables and their interaction terms, with Germany sets as the reference category, while the estimated country-specific marginal effects are reported in Annex Table 3.C.5. On the one hand, the results corroborate the previous evidence of a limited association between inward FDI and firms’ employment, except for Romanian MSMEs and large firms, for which inward FDI realised in forward- and backward-connected sectors, respectively, seems to play a detrimental role. On the other hand, it emerges that inward FDI benefits the most the TFP of Italian MSMEs – and, to some extent, that of Italian large firms when inward FDI is realised in forward-connected sectors. Moreover, Romanian firms’ TFP does not seem to be affected by inward FDI, while German MSMEs’ TFP seems to benefit from (be hindered by) own-sector (backward-connected sector) inward FDI.
Fourth, Equation 9 has been modified by replacing the three variables capturing the sectoral dimension of inward FDI with the variable capturing the regional aggregate of inward FDI realised over a five-year window period , and by interacting it with the set of regional controls to assess region-specific heterogeneity in the association between inward FDI and firms’ outcomes. Annex Table 3.C.6 reports the estimated coefficients of the key variables considered in this exercise; the estimated marginal effects on firms’ TFP are reported in Annex Table 3.C.7, while those referring to firms’ employment are reported in Annex Table 3.C.8. First, it seems that inward FDI is positively associated with the TFP of large firms located in regions which are more developed, more densely populated, more high-tech sector oriented, and characterised by higher unemployment rate, as well as in regions with relatively lower human capital endowment, and specialisation in manufacturing and services sectors. Second, it seems that inward FDI matters for the TFP of only those MSMEs located in regions which are very high-developed but characterised by little human capital endowment. Third, it seems that employment benefits from inward FDI emerge for MSMEs located in low-developed and low-densely populated regions characterised by low unemployment rate but high human capital endowment. In the case of large firms’ employment, inward FDI seems to matter for firms located in regions with high population density.
Fifth, Equation 9 has been augmented by interacting each of the three variables capturing the sectoral dimension of inward FDI with each region-level control variable. However, to avoid excessive multicollinearity, each regional mediation effect has been assessed separately, that is, by estimating a separate equation for each mediation variable. The estimated coefficients of the key variables of interest are reported: in Annex Table 3.C.9 when considering MSMEs’ TFP; in Annex Table 3.C.11 when considering large firms’ TFP; in Annex Table 3.C.13 when considering MSMEs’ employment; and in Annex Table 3.C.15 when considering large firms’ employment. Annex Table 3.C.10 reports the estimated marginal effects of inward FDI on MSMEs’ TFP: on the one hand, the results corroborate the overall negligible role played by inward FDI realised in backward-connected sectors; on the other hand, it seems that own-sector FDI and FDI realised in forward-connected sectors matter for MSMEs located in less developed and manufacturing-intensive regions characterized by low population density, high unemployment rate, low human capital endowment, and low specialisation in high-tech and services sectors. Annex Table 3.C.12 reports the estimated marginal effects of inward FDI on large firms’ TFP: on the one hand, the results corroborate the overall negligible role played by own- and backward-connected sectors inward FDI; on the other hand, it seems that inward FDI realised in forward-connected sectors matters for large firms located in less developed and high-densely populated regions characterised by high unemployment rate and low human capital endowment. Annex Table 3.C.14 reports the estimated marginal effects of inward FDI on MSMEs’ employment: the results show very little heterogeneity across regions with different characteristics and, overall, tend to corroborate a negligible association between inward FDI and firms’ employment. Finally, Annex Table 3.C.16 reports the estimated marginal effects of inward FDI on large firms’ employment: similar to the evidence on MSMEs’ employment, the results show very little heterogeneity across regions with different characteristics and an overall little association between inward FDI and employment.
Note: Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: Marginal effects refer to Annex Table 3.C.4. Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: Marginal effects refer to columns (1) and (2) in Annex Table 3.C.6. Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: Marginal effects refer to columns (3) and (4) in Annex Table 3.C.6. Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: All specifications include firm-level controls, region-level controls, firm FEs, year FEs, and a sector-specific time trend. Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: Marginal effects refer to Annex Table 3.C.9. Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: All specifications include firm-level controls, region-level controls, firm FEs, year FEs, and a sector-specific time trend. Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: Marginal effects refer to Annex Table 3.C.11. Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: All specifications include firm-level controls, region-level controls, firm FEs, year FEs, and a sector-specific time trend. Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: Marginal effects refer to Annex Table 3.C.13. Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: All specifications include firm-level controls, region-level controls, firm FEs, year FEs, and a sector-specific time trend. Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
Note: Marginal effects refer to Annex Table 3.C.15. Standard errors clustered at the regional level in parentheses. All variables are log-transformed.
* p < 0.1.
** p < 0.05.
*** p < 0.01.
**** p < 0.001.
References
[23] Aitken, B. and A. Harrison (1999), “Do Domestic Firms Benefit from Direct Foreign Investment? Evidence from Venezuela”, American Economic Review, Vol. 89/3, pp. 605-618, https://doi.org/10.1257/aer.89.3.605.
[25] Alfaro, L. (2003), “Foreign Direct Investment and Growth: Does the Sector Matter?”, Harvard Business School, pp. 1-31.
[15] Alfaro-Urena, A., I. Manelici and J. Vasquez (2019), “The Effects of Joining Multinational Supply Chains: New Evidence from Firm-to-Firm Linkages”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.3376129.
[16] Amiti, M. et al. (2024), “FDI and superstar spillovers: Evidence from firm-to-firm transactions”, Journal of International Economics, Vol. 152, p. 103972, https://doi.org/10.1016/j.jinteco.2024.103972.
[3] Andrews, D., C. Criscuolo and P. Gal (2015), “Frontier Firms, Technology Diffusion and Public Policy: Micro Evidence from OECD Countries”, OECD Productivity Working Papers, No. 2, OECD Publishing, Paris, https://doi.org/10.1787/5jrql2q2jj7b-en.
[11] Barry, F., H. Gorg and E. Strob (2005), “Foreign direct investment and wages in domestic firms in Ireland: Productivity spillovers versus labour‐market crowding out”, International Journal of the Economics of Business, Vol. 12/1, pp. 67-84, https://doi.org/10.1080/1357151042000323102.
[22] Blalock, G. and P. Gertler (2008), “Welfare gains from Foreign Direct Investment through technology transfer to local suppliers”, Journal of International Economics, Vol. 74/2, pp. 402-421, https://doi.org/10.1016/j.jinteco.2007.05.011.
[8] Blomström, M. and A. Kokko (1998), “Multinational Corporations and Spillovers”, Journal of Economic Surveys, Vol. 12/3, pp. 247-277, https://doi.org/10.1111/1467-6419.00056.
[32] Cainelli, G. and D. Iacobucci (2012), “Agglomeration, Related Variety, and Vertical Integration”, Economic Geography, Vol. 88/3, pp. 255-277, https://doi.org/10.1111/j.1944-8287.2012.01156.x.
[33] Cainelli, G. and D. Iacobucci (2011), “Business groups and the boundaries of the firm”, Management Decision, Vol. 49/9, pp. 1549-1573, https://doi.org/10.1108/00251741111173989.
[27] Crescenzi, R., M. Di Cataldo and M. Giua (2021), “FDI inflows in Europe: Does investment promotion work?”, Journal of International Economics, Vol. 132, p. 103497, https://doi.org/10.1016/j.jinteco.2021.103497.
[42] Crescenzi, R., M. Di Cataldo and A. Rodríguez‐Pose (2016), “Government Quality and the Economic Returns of Transport Infrastructure Investment in European Regions”, Journal of Regional Science, Vol. 56/4, pp. 555-582, https://doi.org/10.1111/jors.12264.
[4] Crescenzi, R., A. Dyèvre and F. Neffke (2022), “Innovation Catalysts: How Multinationals Reshape the Global Geography of Innovation”, Economic Geography, Vol. 98/3, pp. 199-227, https://doi.org/10.1080/00130095.2022.2026766.
[37] Crescenzi, R., ,. Ganau and M. Storper (2021), “Does foreign investment hurt job creation at home? The geography of outward FDI and employment in the USA”, Journal of Economic Geography, Vol. 22/1, pp. 53-79, https://doi.org/10.1093/jeg/lbab016.
[35] Crescenzi, R. and R. Ganau (2024), “Inward FDI and regional performance in Europe after the Great Recession”, Cambridge Journal of Regions, Economy and Society, Vol. 18/1, pp. 167-192, https://doi.org/10.1093/cjres/rsae038.
[17] Crescenzi, R. and O. Harman (2023), Harnessing Global Value Chains for regional development, Routledge, London, https://doi.org/10.4324/9781003356141.
[36] Crescenzi, R. and S. Iammarino (2017), “Global investments and regional development trajectories: the missing links”, Regional Studies, Vol. 51/1, pp. 97-115, https://doi.org/10.1080/00343404.2016.1262016.
[34] Crescenzi, R., C. Pietrobelli and R. Rabellotti (2013), “Innovation drivers, value chains and the geography of multinational corporations in Europe”, Journal of Economic Geography, Vol. 14/6, pp. 1053-1086, https://doi.org/10.1093/jeg/lbt018.
[21] Crespo, N. and M. Fontoura (2007), “Determinant Factors of FDI Spillovers – What Do We Really Know?”, World Development, Vol. 35/3, pp. 410-425, https://doi.org/10.1016/j.worlddev.2006.04.001.
[41] de Groot, O. et al. (2023), Monetary policy and regional inequality.
[26] European Commission (2024), Forging a sustainable future together, https://doi.org/10.2776/974536.
[40] Ganau, R. and A. Rodríguez-Pose (2021), “Credit constraints in European SMEs: does regional institutional quality matter?”, Applied Economics Letters, Vol. 29/15, pp. 1388-1392, https://doi.org/10.1080/13504851.2021.1934385.
[39] Ganau, R. and A. Rodríguez‐Pose (2019), “Do high‐quality local institutions shape labour productivity in Western European manufacturing firms?”, Papers in Regional Science, Vol. 98/4, pp. 1633-1667, https://doi.org/10.1111/pirs.12435.
[14] Girma, S., H. Görg and M. Pisu (2008), “Exporting, linkages and productivity spillovers from foreign direct investment”, Canadian Journal of Economics/Revue canadienne d’économique, Vol. 41/1, pp. 320-340, https://doi.org/10.1111/j.1365-2966.2008.00465.x.
[1] Goldin, I. et al. (2024), “Why Is Productivity Slowing Down?”, Journal of Economic Literature, Vol. 62/1, pp. 196-268, https://doi.org/10.1257/jel.20221543.
[20] Gorg, H. (2004), “Much Ado about Nothing? Do Domestic Firms Really Benefit from Foreign Direct Investment?”, The World Bank Research Observer, Vol. 19/2, pp. 171-197, https://doi.org/10.1093/wbro/lkh019.
[9] Javorcik, B. (2004), “Does Foreign Direct Investment Increase the Productivity of Domestic Firms? In Search of Spillovers Through Backward Linkages”, American Economic Review, Vol. 94/3, pp. 605-627, https://doi.org/10.1257/0002828041464605.
[12] Jude, C. (2015), “Technology Spillovers from FDI. Evidence on the Intensity of Different Spillover Channels”, The World Economy, Vol. 39/12, pp. 1947-1973, https://doi.org/10.1111/twec.12335.
[31] Ketterer, T. and A. Rodríguez‐Pose (2018), “Institutions vs. ’first‐nature’ geography: What drives economic growth in Europe’s regions?”, Papers in Regional Science, Vol. 97, pp. S25-S63, https://doi.org/10.1111/pirs.12237.
[18] Lembcke, A. and L. Wildnerova (2020), “Does FDI benefit incumbent SMEs?: FDI spillovers and competition effects at the local level”, OECD Regional Development Working Papers, No. 2020/02, OECD Publishing, Paris, https://doi.org/10.1787/47763241-en.
[19] Levinsohn, J. and A. Petrin (2003), “Estimating Production Functions Using Inputs to Control for Unobservables”, Review of Economic Studies, Vol. 70/2, pp. 317-341, https://doi.org/10.1111/1467-937x.00246.
[10] Newman, C. et al. (2015), “Technology transfers, foreign investment and productivity spillovers”, European Economic Review, Vol. 76, pp. 168-187, https://doi.org/10.1016/j.euroecorev.2015.02.005.
[13] Nguyen, N., S. Sun and R. Welters (2024), “The impact of FDI on R&D investment of small and medium-sized enterprises in Vietnam: The role of institutions”, International Review of Economics & Finance, Vol. 95, p. 103519, https://doi.org/10.1016/j.iref.2024.103519.
[7] OECD (2024), Strengthening FDI and SME Linkages in Czechia, OECD Publishing, Paris, https://doi.org/10.1787/4c97d104-en.
[6] OECD (2023), Policy Toolkit for Strengthening FDI and SME Linkages, OECD Publishing, Paris, https://doi.org/10.1787/688bde9a-en.
[5] OECD (2017), Small, Medium, Strong. Trends in SME Performance and Business Conditions, OECD Publishing, Paris, https://doi.org/10.1787/9789264275683-en.
[29] Olley, G. and A. Pakes (1996), “The Dynamics of Productivity in the Telecommunications Equipment Industry”, Econometrica, Vol. 64/6, p. 1263, https://doi.org/10.2307/2171831.
[38] Rodriguez-Pose, A. and M. Di Cataldo (2014), “Quality of government and innovative performance in the regions of Europe”, Journal of Economic Geography, Vol. 15/4, pp. 673-706, https://doi.org/10.1093/jeg/lbu023.
[2] Rodríguez-Pose, A. and R. Ganau (2021), “Institutions and the productivity challenge for European regions”, Journal of Economic Geography, Vol. 22/1, pp. 1-25, https://doi.org/10.1093/jeg/lbab003.
[28] Rodríguez‐Pose, A. et al. (2020), “Credit constraints, labor productivity, and the role of regional institutions: Evidence from manufacturing firms in Europe”, Journal of Regional Science, Vol. 61/2, pp. 299-328, https://doi.org/10.1111/jors.12514.
[24] Smeets, R. (2008), “Collecting the Pieces of the FDI Knowledge Spillovers Puzzle”, The World Bank Research Observer, Vol. 23/2, pp. 107-138, https://doi.org/10.1093/wbro/lkn003.
[30] Van Beveren, I. (2007), “Total Factor Productivity Estimation: A Practical Review”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.1004429.
← 1. See Annex Table 3.A.8 for the distribution of FDI projects by country and year.
← 2. See Annex Table 3.B.1 for full regression outputs.
← 3. See Annex Table 3.C.1 for full regression outputs.
← 4. See Annex Table 3.C.2 and Annex Table 3.C.3 for full regression outputs.
← 5. See Annex Table 3.C.4 for full regression outputs.
← 6. The original download of Orbis data included 34 903 German, 818 958 Italian, and 613 122 Romanian firms.
← 7. This subnational classification has been frequently used in previous research at both regional (Rodriguez-Pose and Di Cataldo, 2014[38]; Crescenzi, Di Cataldo and Rodríguez‐Pose, 2016[42]; Ketterer and Rodríguez‐Pose, 2018[31]; Rodríguez-Pose and Ganau, 2021[2]) and firm level (Ganau and Rodríguez‐Pose, 2019[39]; Ganau and Rodríguez-Pose, 2021[40]; Rodríguez‐Pose et al., 2020[28]).
← 8. The choice of developing the empirical analysis on a sample randomly drawn from the Orbis database presents both advantages and disadvantages. Among the disadvantages, the choice has implications in terms of a loss in the number of observations and reduced significance levels. The main advantage is that it increases the representativeness of the sample with respect to the true population of firms operating in the countries analysed. This latter aspect is particularly relevant given the cross‐country nature as well as the subnational and firm size-specific focus of the analysis.
← 9. Industrial sectors include the two-digit manufacturing sectors 10 to 33; the two-digit sector 35 concerning electricity, gas, steam, and air conditioning supply; the two-digit sectors 36 to 39 concerning water supply, sewerage, waste management, and remediation activities; and the two-digit construction sectors 41 to 43.
← 10. For any sectors and , the parameter is derived from national input-output tables drawn from the Organisation for Economic Co-operation and Development (OECD).
← 11. Following (de Groot et al., 2023[41]), the set of capital-intensive sectors includes the one-digit NACE Rev. 2 sectors: mining and quarrying; electricity, gas, steam, and air conditioning supply; water supply, sewerage, waste management, and remediation activities; and construction. The remaining sectors are thus considered as labour-intensive.