The impact of economic specialization on regional economic development in the European Union: Insights for formation of smart specialization strategy

Dzemydaitė, Giedrė Article The impact of economic specialization on regional economic development in the European Union: Insights for formation of smart specialization strategy Economies Provided in Cooperation with: MDPI – Multidisciplinary Digital Publishing Institute, Basel Suggested Citation: Dzemydaitė, Giedrė (2021) : The impact of economic specialization on regional economic development in the European Union: Insights for formation of smart specialization strategy, Economies, ISSN 2227-7099, MDPI, Basel, Vol. 9, Iss. 2, pp. 1-15, https://doi.org/10.3390/economies9020076 This Version is available at: https://hdl.handle.net/10419/257234 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ economies Article The Impact of Economic Specialization on Regional Economic Development in the European Union: Insights for Formation of Smart Specialization Strategy Giedr˙ e Dzemydait˙ e   Citation: Dzemydait˙ e, Giedr˙ e. 2021. The Impact of Economic Specialization on Regional Economic Development in the European Union: Insights for Formation of Smart Specialization Strategy. Economies 9: 76. https://doi.org/10.3390/ economies9020076 Academic Editors: Luís Carvalho and Joana Costa Received: 23 March 2021 Accepted: 11 May 2021 Published: 13 May 2021 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Copyright: © 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). Faculty of Economics and Business Administration, Vilnius University, Saul˙ etekio av. 9, LT-10222 Vilnius, Lithuania; [email protected] Abstract: The smart specialization concept was implemented in the EU in 2014, stating that regions have to specify specialization areas for development of innovations. Economic specialization reveals a comparative advantage in that field. However, there are different arguments linking specialization to economic development. This study analyzes these arguments and aims to investigate the impact of economic specialization on regional economic development and to give insights into identifying prospective areas in regional economies. A panel fixed effect estimation of industry-level regional data suggests that economic specialization in broader regional employment, called relative specialization, is ambiguously associated with economic development. Our findings suggest that neither economic specialization nor economic diversity are a clear-cut solution for ensuring economic growth. Economic structure in EU regions differs, and there is no one answer for which approach is better for economic development. Specialization measures, particularly the location quotient, cannot fully capture the dynamics in the industry structure that could be essential for formation of regional development strategy. Keywords: economic structure; economic specialization; smart specialization; economic growth; EU regions; tradable sectors 1. Introduction European regional development policy has been counted for more than two decades. As stated in the Treaty of Rome, the general focus of this policy is to reduce regional differences throughout the common European market. About one-third of the EU budget was dedicated to reducing regional disparities for programming periods of 2007–2013 and 2014–2020. Due to emerging challenges and global competition, European Cohesion Policy was reformed in 2014, strengthening the innovation-led industrial policy perspective (McCann and Ortega-Argilés 2013 ;Benner 2020). The specific features of the EU context significantly influenced the logic of the changes. Industrial change focusing on innovation is one of the critical challenges of the 21st century that refers to ongoing economic and social development in all industries and fields of activity—both emerging and traditional sectors in decline. At the European level, the vast majority of socio-economic players agreed on the need for a proactive approach to industrial change from two perspectives: fostering higher value-added activities in regions and dealing with social consequences due to changes in the economy and the European labor market. Such a proactive approach to industrial change was introduced in the EU in 2014. The European Commission stated prerequisite requirements for the EU regions to receive funding from the European Regional Development Fund. Regions had to choose their specific specialization areas for innovation and to adopt research and innovation strategies for smart specialization. Smart specialization strategy reveals a place-based policy approach to regional economic development (Barca et al. 2012). Regions identified strategic areas for intervention Economies 2021,9, 76. https://doi.org/10.3390/economies9020076 https://www.mdpi.com/journal/economies Economies 2021,9, 76 2 of 15 based on the analysis of the economy’s strengths and the potential of regional innovation systems (European Commission 2015,2017a,2017b;McCann and Ortega-Argilés 2015). Place-based policies are supposed to stimulate private sector investment and economic growth in the treated place, and as such, they are sometimes difficult to appraise and evaluate (Duranton and Venables 2018). This approach to regional innovation policy spreads worldwide. Other countries and regions outside the EU (in Latin America, sub-Saharan Africa, Eastern partnership countries, and others) investigate opportunities to apply regional research and innovation strategies for smart specialization (Dosso et al. 2020;Barroeta et al. 2017;Kleibrink et al. 2017). At the core of policy formation comes the selection of smart specialization areas and raises a question of the role of economic specialization in this process. The main question behind the policymaking is how to shift regional economic development into more smart growth and how to facilitate new prospective industries and activities to flourish. The idea behind this is that regions have to use their strengths and develop sectors and activities where regional actors had a related experience. This leads to a diversification process in relation to a current specialization (Foray et al. 2009,2011; Balland et al. 2019). Diversification means that new prospective activities are emerging. Specialization means that these activities relate to the regional context and previous practice. Therefore, in the first step of policy formation comes an analysis of regional economic structure and current specialization and innovation areas. One of the available methods is an analysis of regional economic specialization (Foray et al. 2012). It is a quantitative analysis to calculate degrees of specialization of regional economies based on employment or value-added data. Specialization indexes, usually the location quotient (LQ) index, indicate critical masses of activity. Even though it does not reveal innovation-driven linkages, it still gives evidence of a regional economy where value added and exports are generated in the economy and where knowledge and competencies are concentrated (Foray 2015). Various EU regions applied the LQ index for specialization analysis as the first step of formation of a smart specialization strategy. Then, through the entrepreneurial discovery process, regional actors further discussed prospective future economic growth areas and formulated smart specialization strategies. Even though a measurement of economic specialization in the sense of LQ indexes are broadly involved in a policy formation around the EU and other countries (Foray et al. 2012;Postoiu and Dachin 2014), there is still a gap in the scientific literature for giving more evidence and critical analysis of the applicability of quantitative measures for policy formation. A recent discussion of Hassink and Gong (2019), Foray (2019), and Benner (2020) underlined a need for rigorous measurements of prospective areas and a critical judgment of the currently applied techniques in policy formation. Prioritization of prospective areas proved difficult and limited capacities of implementing agencies, especially in lagging regions (e.g., Capello and Kroll 2016;Karo et al. 2017;Gianelle et al. 2019;Trippl et al. 2020). Proper quantitative or qualitative techniques could reduce the risk of local actors’ incapacity. Most recently, Balland et al. (2019) suggested involving a new measurement of relatedness in policy formation that is expanding knowledge in this field. Kemeny and Storper (2014) critically analyzed absolute and related specialization for the USA regions and suggested that relative specialization cannot capture dynamics in the economic structure. For this reason, a critical assessment of quantitative techniques is a question for further scientific research that is not sufficiently covered for the EU regions. In this paper, we followed these thoughts and aimed to investigate the impact of economic specialization on regional economic development and give insights about identifying prospective areas in regional economies. The research question was to what extent higher specialization rates, measured by location quotient, reveal prospective areas for economic growth and to what extent such analysis could be valuable in the policy formation of smart specialization strategies. The novelty of the paper is that we match an impact evaluation with insights for formation of smart specialization strategies. Most Economies 2021,9, 76 3 of 15 studies present the economic nature of how regions evolve rather than give insights into the application of economic specialization measures in the process of policy formation, apart from some research, e.g., Kemeny and Storper (2014), who analyzed USA regional data, or Balland et al. (2019), who suggested a new framework. This paper gives insights into how to divide tradable and non-tradable sectors essential for economic specialization analysis. It supplements previous explanations (Jensen and Kletzer 2005;Moretti and Thulin 2012;Kemeny and Storper 2014). A dataset covers the most recent data of the EU regions that gives a sense of a broader EU perspective rather than a case study of a region or country, but still with economic activities distributed in comparatively disaggregated levels (48 sectors). Case studies could be found from a variety of researchers, i.e., Postoiu and Dachin (2014), Paliokaite et al. (2015), and Sotarauta and Suvinen (2019). The paper consists of four main parts. Section 2systemizes related literature and gives a general image of the impact of an economic specialization on regional economic development. It reveals the gains and drawbacks of an economic specialization to a regional economy. Section 3explains research methods and data. This section also presents an approach to the division between tradable and non-tradable sectors that supports a selection of economic sectors involved in the analysis. Section 4reveals the estimation results of the panel data model. Sections 5and 6summarize the main findings and explain policy implications. 2. Related Literature There are two sides of the coin while analyzing the economic specialization and its impact on economic development. Both specialization and diversification processes of economic structure could benefit the regional economy from different perspectives. From the most recent literature overview, new innovative activities are supposed to emerge as a diversification process in current specialization areas (Foray et al. 2011;Hassink and Gong 2019). Regions need to continuously reinvent themselves by developing new activities that mean a process of diversification (Balland et al. 2019). The explanation of specialization and diversification processes and their impact on economic development comes from the works of Jacobs (1969), Marshall (1920), Arrow (1962) , and Romer (1990), who analyzed agglomeration effects and high spatial concentration of companies in the same or different sectors. These evolved into dominant theoretical concepts, most recently analyzed by Boschma and Martin (2010), Content et al. (2019), and other researchers. Because of the limited size of a local market, economic specialization of a region forms when firms sell goods or services in foreign markets (Kemeny and Storper 2014). Higher economic specialization rates relate to higher exports and reflect a comparative advantage of companies in that field. Change of production level in exporting sectors influences regional income directly and indirectly through income-multiplying effects (Moretti and Thulin 2012;Capello 2016, p. 122). The output of tradable sectors is not limited by local income and market size. If an external demand is growing and the region’s output enjoys increasing rations of its unit price relative to imports, it benefits the economy’s overall income. The tradable sectors considerably impact overall productivity in the European Union (Friesenbichler and Glocker 2019). Increasing the size of localized activity, which means a process of economic specialization, could also enhance the productivity of firms through sharing of input suppliers, a specialized labor market and its development, and technological learning and spillover effects through actors spread across different organizations (Duranton and Puga 2004; Henderson 2003). Industrial concentration and economic specialization lead to positive externalities in the local economy that could affect economic development, employment, and income (De Groot et al. 2009). On the other hand, arguments are linking a more diversified economic structure to regional economic development. The greater diversity in the economy could generate the greater capacity to create more types of goods and services (Jacobs 1969). The spatial prox- Economies 2021,9, 76 4 of 15 imity of companies from different sectors could create positive externalities to economic development. It contributes to the exchange and combination of ideas between different sectors, which is beneficial for creating new activities. Even though knowledge is spread primarily within the individual economic sector, it could also spread across sectors. If the variety of companies and skills are related, it is supposed to give more opportunities for regional economic development (Mameli et al. 2012;Van Oort et al. 2015;Misiak and Dykas 2021). Regions are more likely to diversify into complex technologies when relatedness is high (Balland et al. 2019). An additional shortcoming of the specialization of regions is the risk of changing technologies or black swans. When a particular economic sector is affected by an unfavorable business cycle, a highly specialized territory may experience a severe economic downturn (Dietz and Garcia 2002;Šidlauskait ˙ e-Riazanova and Miškinis 2019). One example is the COVID-19 crisis that revealed an uncompromising slump of regions specialized in tourism (Šidlauskait ˙ e-Riazanova 2020;Skare et al. 2021;Zhang et al. 2021). Highly specialized regions could become a hostage of their economic structure. The question is to what extent such regions can diversify in new economic activities in general and in the case of a crisis. Different goods and activities are associated with differed technological and learning opportunities. Regions specialized in dynamic and innovative industries are better positioned to achieve sustained economic growth than others (Hausmann et al. 2007). Diversification into more complex technologies is attractive but challenging for European Union regions to accomplish (Balland et al. 2019). A critical mass of capabilities to develop collective learning and compete at a global level is needed (Iacobucci and Enrico 2016). There is a risk of a lack of local preconditions for innovation development in regions with few research institutions and high-tech clusters. These conditions form challenges for developing new specializations in more complex activities where regions do not have an existing practice. According to Capello and Kroll (2016), there are issues in developing high valueadded specializations in peripheral regions. If the regional economy is highly dependent on multinational enterprises (MNEs) that are settled in the region, a repositioning of regions in international value chains can often not be controlled by policymakers from within these regions alone. MNEs have their global corporate strategies and could not be specifically interested in becoming engaged with these regional local development strategies. MNEs remain vulnerable to more general corporate plans. Decisions to reallocate activities are out of the control of the regional government. Even though different arguments link more specialized or diversified economic structure to economic development, a variety of arguments comes for economic specialization through the lenses of comparative advantage, intra-sectoral dissemination and knowledge spillover within the sector, and localization externalities that come from the economic specialization. The extent to which economic specialization influences regional economic development remains a question for scientific debate. 3. Research Methodology 3.1. Model Specification In this paper, we investigated the relationship between economic specialization and regional economic growth in European regions. The research framework is built on a neoclassical growth theory (Solow 1952) that is commonly applied for modeling economic growth in various research (Liu 2019). Control variables in a model represent changes in capital and labor expressed by indicators of gross fixed capital formation, persons with tertiary education, or working-age population. These variables are common in an economic growth analysis (Akcigit 2017). For this study, we involved a measure of economic specialization in the model as an independent variable. There are different approaches on how to measure economic specialization. Most studies measure the level of relative specialization by location quotients following Glaeser et al.’s (1992) idea. This measure is applied in the policymaking process to evaluate prospective smart specialization areas (Foray et al. 2012). For this reason, we used a Economies 2021,9, 76 5 of 15 relative specialization indicator based on location quotient calculations. A coefficient of locations quotients (LQ ri ) incorporates the effect of regional industrial scale by the following expression (Lu et al. 2011): LQri =Eri/(∑N i=1Eri) ∑M r=1Eri/(∑N i=1∑M r=1Eri)=Sri Si (1) where LQ ri —the location quotient of industry iin region r,r= 1, . . . ,M,i= 1, . . . ,N; E—the number of employees (employment); S ri —the employment share of industry iin region r;Si—the employment share of industry iin total employment. The relative specialization of region rwas measured by the average location quotient weighted by the number of people employed: Specializationr=∑N i Eri Er LQri (2) where E ri denotes the number of employees in region rin industry i, r = 1, . . . ,M, i = 1, . . . , N;Er—the total number of employees in region r. We applied a panel fixed effect estimation for industry-level regional data based on a neoclassical growth theory. We estimated an equation with and without year fixed effects to consider time-specific shocks and their impact on estimated results. Estimated model (2) was as follows: GDPgrit =β1Specit +β2GDPit−1+β3Employit +β4GFCFit +β5EDUit +µi+ηt+εit (3) where GDPgr it —an annual growth of GDP per inhabitant in region iand period t,r= 1, . . . ,M;Spec it —regional specialization (Specialization r in Equation (3)); GDP it−1 —lagged GDP per inhabitant; Employ it —an annual growth of employed persons; GFCF it —gross fixed capital formation per inhabitant; EDU it —a percentage of the population with tertiary education; µi —unobserved individual fixed effects; ηt —time effects; εit —an error term. GDP it−1 is involved due to the dependence of regional economic growth on the level of economic development (Monfort 2008). In the EU context, European regions with lower GDP per capita obtained higher cohesion funds related to the level of development for catching up with economically developed regions. We have chosen to analyze the regional level rather than the national level. Smart specialization strategies were developed in the EU at the regional level. Additionally, for relative specialization analysis, regional data is more suitable for comparing units that are more similar in size from a territorial perspective and density of resources. Regions in the European Union are divided into administrative units according to the Nomenclature of Territorial Units for Statistics (NUTS). The NUTS classification was applied into a formal Regulation of the European Parliament and the Council in 2003 (Eurostat 2015). The NUTS harmonizes the collection of regional data, ensures comparable regional statistics for analysis, and for the comparison of the socio-economic situation and policy measures. NUTS2 is a regional level for the application of regional policies in the EU. However, industry-level data of NUTS2 was limited for the analysis. Therefore, we analyzed data of NUTS1 level of 2008–2016, assuming that economic specialization is reflected in more aggregated territorial units. A time period was selected due to the regional and industrial data availability. 3.2. Selection of Economic Sectors for Analysis Local markets are limited in size. For a region to achieve a higher specialization rate in some economic activities, it has to have at least part of the demand from foreign markets. For this reason, tradable sectors are involved in the analysis of economic specialization. Tradable sectors are those whose output is traded internationally, and it could be both goods and services (Gervais and Jensen 2019;Francois and Hoekman 2010). Economies 2021,9, 76 6 of 15 A set of studies have focused on the manufacturing sector, implicitly assuming that services are non-tradable, but this traditional assumption that goods are tradable and services are non-tradable is increasingly inadequate. Gervais and Jensen (2019) stated that service trade has grown over time and now accounts for about 20% of global international transactions, and tradable service industries account for about the same share of value added as tradable manufacturing industries in the US. The potential welfare gains from trade liberalization in the service sector seem considerable (Francois and Hoekman 2010). In a globalized economy, services form a significant part of international trade. Therefore, service sectors have to be included in the analysis of economic specialization. There is a discussion on how to separate tradable vs. non-tradable sectors. For example, Moretti and Thulin (2012) defined the non-tradable sector as locally produced and consumed services and the tradable sector as manufacturing and a part of the service sector that were exported outside the local economy. Kemeny and Storper (2014) calculated the Herfindahl index of geographical concentration to find a cut-off point in the distribution of concentration values at which tradable activities are distinguished from non-tradable in the US metropolitan areas, stating that if sectors are spatially ubiquitous, they had to be generally non-tradable (Jensen and Kletzer 2005). Herfindahl index of geographical concentration (Concj) is evaluated by the formula: Concj=∑K k=1 ejk Ej!2 (4) where e—employment in industry jand city k;Eis the total employment across all cities in industry j. Herfindahl index values vary from 0 to 1, while industries with values near 0 exhibit a uniform distribution over space, while values closer to 1 indicate sectors with highly concentrated activity in only a few locations. With no clear theoretical guidance on such a cut-off, Kemeny and Storper (2014) have chosen to eliminate sectors with a lower Herfindahl index rate (lower than 0.036) that seemed to differentiate non-tradable from tradable sectors in practice. For example, eliminated sectors were death care services, retail stores, car repairing, architectural services, etc. We used this approach to evaluate the distribution of economic activities between European regions with an expansion. While analyzing the Herfindahl index of geographical concentration for sectors in the EU, some sectors with a comparatively low Herfindahl index have high specialization values (LQ, Equation (1)). For example, almost all regions in the EU had wholesale and retail trade and repair of motor vehicles and motorcycle activities (sector code—G45). The Herfindahl index was close to 0 (0.0161) because such activities are in all regions. However, for example, in the LT0 region, the LQ index of this sector was high (LQ = 1.42), revealing that a variety of companies in LT0 were repairing and re-exporting motor vehicles and motorcycles to foreign markets and it was a significant part of their value added. The same situation occurred with accommodation and food and beverage services activities (codes of sectors—I55 and I56). All regions in the EU had such activities that meant low HI (0.0187 and 0.0185). However, some regions focused on tourism, e.g., the Crete region, which was highly specialized in these sectors with high LQ values. For this reason, we did not cut off all sectors with a comparatively small Herfindahl index in the way Kemeny and Storper (2014) did with the US data. This approach seemed not suitable for EU data because there were considerable differences in economic activities between EU regions and their economic structure. Therefore, we involved the LQ index in analysis for identifying the tradable sectors (Equation (1)). Rather than estimating the impact of specialization in the full range of all sectors, we focused on tradable sectors. We eliminated such sectors as veterinary activities, rental and leasing activities, employment activities, retail trade, water supply, etc. Moreover, we did not involve sectors such as C32—other manufacturing or M74—other professional, scientific, and technical activities because of the aggregated data of different types of economic activities under the same Economies 2021,9, 76 7 of 15 code. We involved 48 industries in the analysis according to NACE classification. NACE classification groups together economic activities that have similar input and production techniques and similar outputs. A final list of sectors involved in the analysis is presented in Appendix A. 4. Results This part first represents an overview of the level of economic specialization in the EU regions during the period analyzed and then gives insights about estimation results of economic specialization impact on economic growth. A broader discussion of the results is presented in the following part. All EU regions tended to have a weighted average specialization (Specialization r ) above 1 (Figure 1) during the period analyzed. It revealed that all regions had some areas of specialization. The intensity of regional specialization varied from around 1.09 to 1.93, with most regions with a specialization rate not above 1.49. Usually, those regions were specialized in some specific sector which distinguished them from other regions, e.g., the NL1 (Noord-Nederland) region, which was mainly focused on water transport (H50), or the PL2 (Makroregion Poludniowy) region, which was specialized in mining and quarrying (B) activities. Economies 2021, 9, x FOR PEER REVIEW 7 of 15 impact of specialization in the full range of all sectors, we focused on tradable sectors. We eliminated such sectors as veterinary activities, rental and leasing activities, employment activities, retail trade, water supply, etc. Moreover, we did not involve sectors such as C32—other manufacturing or M74—other professional, scientific, and technical activities because of the aggregated data of different types of economic activities under the same code. We involved 48 industries in the analysis according to NACE classification. NACE classification groups together economic activities that have similar input and production techniques and similar outputs. A final list of sectors involved in the analysis is presented in Appendix A. 4. Results This part first represents an overview of the level of economic specialization in the EU regions during the period analyzed and then gives insights about estimation results of economic specialization impact on economic growth. A broader discussion of the results is presented in the following part. All EU regions tended to have a weighted average specialization (Specialization r ) above 1 (Figure 1) during the period analyzed. It revealed that all regions had some areas of specialization. The intensity of regional specialization varied from around 1.09 to 1.93, with most regions with a specialization rate not above 1.49. Usually, those regions were specialized in some specific sector which distinguished them from other regions, e.g., the NL1 (Noord-Nederland) region, which was mainly focused on water transport (H50), or the PL2 (Makroregion Poludniowy) region, which was specialized in mining and quarrying (B) activities. Figure 1. Median Specialization r in NUTS1 regions of the EU in 2009–2016. Source: Eurostat, author’s calculations. Figure 1. Median Specialization r in NUTS1 regions of the EU in 2009–2016. Source: Eurostat, author’s calculations. A majority of European regions were comparatively not highly specialized and served a variety of different activities. This tendency was changing very gradually. The median specialization varied from 1.30 to 1.33 and had no clear tendency of change for the overall sample from 2009 to 2016 (Figure 1). No clear tendency reveals both specialization and diversification processes. The median specialization was at a comparatively similar level. Economies 2021,9, 76 8 of 15 There was a highly mixed picture of economic specialization changes while analyzing different regions in the European Union (Figure 2). Some regions became more specialized while others became more diversified, and a majority of the regions stayed more or less at the same level of economic specialization. For example, NL1 (Noord-Nederland), DE5 (Bremen), and EL6 (Kentriki Ellada) were among those regions where specialization increased. Meanwhile, DE8 (Mecklenburg-Vorpommern), PT1 (Continente), and LU0 (Luxembourg) were among those regions where specialization decreased. It was closely related to the economic activities that regions were serving. In some capital regions, a lower specialization was due to the growing diversity of economic activities within a territory. Economies 2021, 9, x FOR PEER REVIEW 8 of 15 A majority of European regions were comparatively not highly specialized and served a variety of different activities. This tendency was changing very gradually. The median specialization varied from 1.30 to 1.33 and had no clear tendency of change for the overall sample from 2009 to 2016 (Figure 1). No clear tendency reveals both specialization and diversification processes. The median specialization was at a comparatively similar level. There was a highly mixed picture of economic specialization changes while analyzing different regions in the European Union (Figure 2). Some regions became more specialized while others became more diversified, and a majority of the regions stayed more or less at the same level of economic specialization. For example, NL1 (Noord-Nederland), DE5 (Bremen), and EL6 (Kentriki Ellada) were among those regions where specialization increased. Meanwhile, DE8 (Mecklenburg-Vorpommern), PT1 (Continente), and LU0 (Luxembourg) were among those regions where specialization decreased. It was closely related to the economic activities that regions were serving. In some capital regions, a lower specialization was due to the growing diversity of economic activities within a territory. (a) (b) Figure 2. Specializationr in NUTS1 regions of the EU in 2009 (a) and 2016 (b). Source: Eurostat, author’s calculations. There was no clear relation of changes in the economic specialization to the regional economic growth. Some economically strong regions got more specialized and vice versa. However, more specialized regions hold a risk of technology change or losses in demand that could emerge in the long run. Significantly, at a higher risk are lagging regions with lower learning capabilities. Lu et al. (2011) for Chinese regions or Sotarauta and Suvinen Figure 2. Specializationrin NUTS1 regions of the EU in 2009 (a) and 2016 (b). Source: Eurostat, author’s calculations. There was no clear relation of changes in the economic specialization to the regional economic growth. Some economically strong regions got more specialized and vice versa. However, more specialized regions hold a risk of technology change or losses in demand that could emerge in the long run. Significantly, at a higher risk are lagging regions with lower learning capabilities. Lu et al. (2011) for Chinese regions or Sotarauta and Suvinen (2019) for Norway regions revealed that higher specialization is supposed to be related to lower diversity and lower growth perspectives. Further, we estimated the contribution of economic specialization to regional economic development by a panel fixed effect estimation (Table 1). All control variables, such as capital, employment, and education, statistically significantly contributed to regional economic growth. There was a strong link between the income level (GDP pc-1) and economic growth. Regions with lower GDP per capita had higher economic growth rates, and it reveals catching-up processes at the EU regional level. Economies 2021,9, 76 15 of 15 Russu, Corneliu. 2015. Industrial Specialization of the European Union Member Countries. Economic Insights—Trends and Challenges 67: 63–65. Šidlauskait ˙ e-Riazanova, Brigita, and Algirdas Miškinis. 2019. Aspects of the Development of Lithuanian Economic Specialisation in the Context of Globalization. Social Research 42: 59–73. [CrossRef] Šidlauskait ˙ e-Riazanova, Brigita. 2020. Assessment of the demand side pandemic impact on the EU economy in the context of globalization. Applied Economics: Systematic Research 14: 59–76. [CrossRef] Skare, Marinko, Domingo Riberio Soriano, and Małgorzata Porada-Rocho´n. 2021. Impact of COVID-19 on the travel and tourism industry. Technological Forecasting and Social Change 163. [CrossRef] Solow, Robert Merton. 1952. On the Structure of Linear Models. Econometrica 20: 29–46. [CrossRef] Sotarauta, Markku, and Nina Suvinen. 2019. Place leadership and the challenge of transformation: Policy platforms and innovation ecosystems in promotion of green growth. European Planning Studies 27: 1748–67. [CrossRef] Trippl, Michaela, Elena Žukauskait ˙ e, and Adrian Healy. 2020. Shaping smart specialisation: The role of place-specific factors in advanced, intermediate and less-developed European regions. Regional Studies 54: 1328–40. [CrossRef] Van Oort, Frank, Stefan de Geus, and Teodora Dogaru. 2015. Related variety and regional economic growth in a cross-section of European urban regions. European Planning Studies 23: 1110–27. [CrossRef] Witell, Lars, Hannah Snyder, Anders Gustafsson, Paul Fombelle, and Per Kristensson. 2016. Defining service innovation: A review and synthesis. Journal of Business Research 69: 2863–72. [CrossRef] Zhang, Hanyuan, Haiyan Song, Long Wen, and Chang Liu. 2021. Forecasting tourism recovery amid COVID-19. Annals of Tourism Research 87: 103–49. [CrossRef]