A NUTS-2 European Union interregional system of Social Accounting Matrices for the year 2017: The RHOMOLO V4 dataset

García-Rodríguez, Abián et al. Working Paper A NUTS-2 European Union interregional system of Social Accounting Matrices for the year 2017: The RHOMOLO V4 dataset JRC Working Papers on Territorial Modelling and Analysis, No. 01/2023 Provided in Cooperation with: Joint Research Centre (JRC), European Commission Suggested Citation: García-Rodríguez, Abián et al. (2023) : A NUTS-2 European Union interregional system of Social Accounting Matrices for the year 2017: The RHOMOLO V4 dataset, JRC Working Papers on Territorial Modelling and Analysis, No. 01/2023, European Commission, Joint Research Centre (JRC), Seville This Version is available at: https://hdl.handle.net/10419/283086 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/ A NUTS-2 European Union interregional system of Social Accounting Matrices for the year 2017: The RHOMOLO V4 dataset JRC Working Papers on Territorial Modelling and Analysis No 01/2023 Authors: García-Rodríguez, A. Joint Research Centre 2023 JRC WORKING PAPER Lazarou, N.J. Mandras, G. Salotti, S. Thissen, M. Kalvelagen, E. A NUTS-2 European Union interregional system of Social Accounting Matrices for the year 2017: The RHOMOLO V4 dataset Abián García-Rodríguez1, Nicholas Lazarou1, Giovanni Mandras2, Simone Salotti1, Mark Thissen3, and Erwin Kalvelagen4* 1. European Commission, Joint Research Centre (JRC) Seville, Spain 2. Cassa Depositi e Prestiti, Rome, Italy 3. PBL Netherlands Environmental Assessment Agency, The Hague, The Netherlands 4. Amsterdam Optimization, Washington DC, USA Abstract. We describe the procedure used to construct a set of interregional NUTS-2 European Union Social Accounting Matrices for the year 2017, using official Eurostat national and regional account data, and auxiliary data on business and interregional trade flows derived from transport survey data. This procedure builds on the one presented by Thissen et al. (2019) for the year 2013. The new dataset is described by presenting some relevant features and the results of some characteristic simulations obtained with a new version of the RHOMOLO model, referred to as RHOMOLO V4, based on the new 2017 data. Keywords: Regional data, Input-Output, social accounting matrices, trade flows, general equilibrium. JEL Codes: C82, E16, R15, F14, F47. * The authors acknowledge funding through ZonMw project 10430-03201-0006 («The Resilient Region; Regional-Economic Impact Mitigation of Corona-related (De)escalation Policies») for the construction of the interregional IO-table. 1 1 Introduction In the European Union (EU), a significant part of the budget is used for regional policy, called cohesion policy, making it relevant for policymakers and citizens (Crucitti et al., 2022). Recent evidence on regional discontent (Rodríguez-Pose, 2017, Los et al., 2017) emphasizes the political importance of regional policymaking. The literature on regional economics highlights the importance of the spatial character of economic systems to explain regional economic phenomena. However, despite the attention drawn to issues related to regional inequality and economic development, reliable data at the regional level is often lacking and not comparable to national data in terms of quality and quantity. In particular, interregional trade flows are generally not measured by statistical offices. As a result, we have to rely on estimates (see, for instance, Fournier Gabela, 2020, for Japan, Liu et al., 2015, for China, and Généreux and Langen, 2002, for Canada). Based on the estimated trade flows among EU regions by Thissen et al. (2013a and 2013b), Thissen et al. (2019) describe the construction of a system of interregional Social Accounting Matrices (SAMs) with data from 2013 for all the NUTS-2 regions of the European Union (EU). NUTS stands for Nomenclature of Territorial Units for Statistics classification, and the NUTS-2 2016 definition is used in this analysis (see Table A1 in the appendix for the full list of regions). SAMs are comprehensive, economy-wide datasets containing information on all the transactions between the economic agents of a specific economy for a certain period of time. SAMs are normally used for the calibration of multisectoral models like computable general equilibrium (CGE) ones (Mainar- Causapé et al., 2018). Stone (1947) was the first to work on data organised in SAMs, and the work by Pyatt and Thorbecke (1976) and Pyatt and Round (1985) led to their widespread use as a formal framework for economic analysis and planning. The dataset described by Thissen et al. (2019) forms the basis for the calibration procedure of the spatial dynamic CGE RHOMOLO V3 model, whose complete mathematical description is offered by Lecca et al. (2018). The model, which has been developed by the JRC in collaboration with the Directorate-General for regional and urban policy (DG REGIO), is used for territorial ex-ante impact assessments of policies such as the cohesion and innovation policies (see Sakkas, 2018, on the European Social Fund; Di Comite et al., 2018, on the cohesion policy funds; and Christensen, 2018, on Horizon Europe). Several scientific articles also feature RHOMOLO V3 analyses (see, among others, Lecca et al., 2020; Di Pietro et al., 2021; Barbero et al., 2022a, 2022b, and 2023). In this paper, we present a revised procedure to construct a similar dataset for 2017, which is used for the calibration of the latest version (V4) of the RHOMOLO model. We take advantage of recent developments in data availability and estimation techniques. First, Eurostat now publishes official EU intercountry supply, use and Input-Output tables (EU IC-SUIOTs – also referred to as FIGARO 2 tables 1 ). These tables form an ideal starting point for producing EU regional SAMs. Second, we use an improved technique to estimate the interregional Supply and Use Tables (SUTs), which form the basis for the SAMs of all the NUTS-2 regions of the EU. The FIGARO tables were published for the first time in 2021. The goal is to have these updated and enriched annually. They link national accounts with data on business, trade and jobs for EU Member States and 18 main EU trading partners (Argentina, Australia, Brazil, Canada, China, India, Indonesia, Japan, Republic of Korea, Mexico, Norway, Russian Federation, Saudi Arabia, South Africa, Switzerland, Türkiye, the United Kingdom, and the United States). A rest-of-the- world (ROW) entity completes the FIGARO tables. A full description of the methodology used to produce the tables is provided by Remond-Tiedrez and Rueda-Cantuche (2019). The mostly parameter-free methodology proposed by Simini et al. (2012) is the starting point for the estimation of the interregional trade flows between the NUTS-2 regions of the EU. This approach is compatible with methods of data construction and overcomes the shortcomings of often-used gravity-style estimations. The process combines data from different sources (some of them survey-based) into a consistent and complete dataset of interregional accounts for production (Supply tables) and demand (Use tables). Interregional trade is derived from freight transport, business travel, and flight microdata taken from (Thissen et al., 2019). Notably, the existence of trans-shipment locations was taken into account. The resulting regional tables are consistent with the national account data as presented in the FIGARO intercountry SUTs. In a nutshell, the procedure used to produce the inter-regional SAMs for the EU comprises the following five steps: 1. Starting from the FIGARO intercountry SUTs (at the level of 64 NACE sectors and associated CPA products 2 ), we use regional data (mainly on production and consumption) to obtain regional SUTs. We aggregate several sectors due to data limitations, reducing the 64 NACE sectors to 56 categories. Table A2 in the appendix lists both the 64 and the aggregated 56 NACE sectors. 2. Using trade priors for the regional trade flows based on Thissen et al. (2019), we disaggregate the intercountry trade flows of the Figaro tables into regional trade flows within and between countries. 3. The resulting regional SUTs are combined into interregional IO (inputoutput) tables. Since we have rectangular SUTs (the number of sectors and products are not identical), we use the standard industry technology assumption to determine the IO table, according to which the distribution of inputs and Gross Value Added (GVA) of the product that is moved from 1 FIGARO stands for ‘Full International and Global Accounts for Research in Input-Output analysis’ and comprises the EU intercountry supply, use and Input-Output tables (EU IC-SUIOTs). 2 NACE (the term is derived from the French nomenclature statistique des activités économiques dans la communauté Européenne) is the classification of economic activities in the EU, where CPA stands for the Classification of Products by these Activities. 3 a column to another is assumed to follow the structure in the column from which the product is moved (see Miller and Blair, 2009, pp. 212-213). This conversion of SUTs into IO tables is done for analytical purposes (national accounts data are organised in SUTs, but the IO structure is better suited for economic modelling). 4. We aggregate the regional data of step 3 from 56 CPA sectors to 10 NACE Rev. 2 sectors and merge all the non-EU regions and countries into the residual entity ROW. Also, we redistribute the direct interregional imports of final demand categories such that the flows of final demand elements between regions (direct imports of households, government, and firms - gross fixed capital formation, GFCF) are transferred to the corresponding final demand element and sector of the destination region, and to the corresponding sector as an intermediate output of the origin region. 5. We combine the interregional IO tables with data for the secondary distribution of income to create interregional SAM tables. In the remainder of the paper, we provide more details on the procedure and we show some salient features of the new dataset. We also present the results of some basic simulations with the RHOMOLO V4 model calibrated with the 2017 data constructed here. Section 2 briefly describes the intercountry FIGARO tables, and proceeds to explain the procedure in step 1 to regionalise them. Section 3 illustrates steps 2 and 3 of the procedure, which yield interregional SUTs for the goods and services categorised according to the 64 CPA and 56 NACE sectors. Section 4 illustrates the sectoral and geographical aggregation used for the RHOMOLO model (step 4 of the above list). Section 5 deals with step 5: the creation of the interregional SAMs starting from the IO tables. Finally, section 6 shows some features of the dataset, as well as some modelling simulations obtained with basic economic shocks. Section 7 concludes. 2 Regionalization of national SUTs 2.1 The FIGARO Tables In this section, we provide a brief overview of the FIGARO data (Remond-Tiedrez and Rueda-Cantuche, 2019), the starting point of the first step of the procedure to produce inter-regional SAMs for the EU NUTS-2 regions. The FIGARO data are organised in SUTs: these are matrices by sector and product categories describing how domestic production and imports of goods and services in an economy are used by industries for intermediate consumption and final use/demand. Products are classified according to the CPA system, while industries use the NACE Rev. 2 system. These classifications are fully aligned: at each level of aggregation, the CPA shows the primary and secondary products of the corresponding industries using the NACE classification. The SUTs provide information on the structure of production and their costs and on the value added created during this production. They also show the flows of goods and services produced in the national economy, and between that 4 economy and the rest of the world. As such, they combine all the information of three accounts: goods and services, production, and generation of income. Table 1 presents all the elements of the National Accounts that can be found in FIGARO. Table 1: National accounts elements in FIGARO Code Description P1 Total Output P2 Total Intermediate Consumption by Activity D21X31 Taxes less subsidies on products D29X39 Other net taxes on production OP_NRES Purchases of non-residents in the domestic territory OP_RES Direct purchase abroad by residents B2A3G Gross operating surplus D1 Compensation of employees B1G Gross Value Added (GVA) P3_S13 Government consumption P3_S14 Household consumption P3_S15 NPISH (non-profit institutions serving households) consumption P51G Gross Fixed Capital Formation (GFCF) P5M Changes in valuables and inventories Data in FIGARO are disaggregated at the product level in the 64 CPA categories presented in Table A2 in the appendix. The level of detail of the data requires some elements to be classified as confidential. These gaps in the data are corrected during the regionalization phase using complementary regional data. The data in FIGARO cover the 45 countries listed in Table 2. Information on the EU members, the UK, and the US is built-up by linking national accounts and data on business, trade and jobs. The data for the remaining countries, which are the main trading partners of the EU, come from the Inter-Country IO (ICIO) tables of the OECD. There is also a ROW region which simply collects any remaining data. 5 Table 2: Countries in the FIGARO database Belgium BE Cyprus CY Slovakia SK Russian Federation RU Bulgaria BG Latvia LV Finland FI India IN Czechia CZ Lithuania LT Sweden SE China CN Denmark DK Luxembourg LU United Kingdom GB South Africa ZA Germany DE Hungary HU Norway NO Japan JP Estonia EE Malta MT Switzerland CH Korea, Republic of KR Ireland IE Netherlands NL Türkiye TR Indonesia ID Greece GR Austria AT United States of America US Australia AU Spain ES Poland PL Canada CA Saudi Arabia SA France FR Portugal PT Mexico MX Rest of the World ROW Croatia HR Romania RO Argentina AR Italy IT Slovenia SI Brazil BR 2.2 Regionalization of production and use of country-level SUTs Before we can start regionalising the FIGARO dataset, we have to make some minor corrections. The FIGARO dataset conforms to the official national account statistics of Eurostat and SUTs as constructed by the national Bureaus of Statistics. As a result, the data can pose problems for the regionalisation of the national tables since theoretically invalid numbers (mainly negative entries) are incidentally part of the tables. Although one can keep these negative numbers in the national tables, it makes little sense to regionalise them and spread erroneous values over the different regions. We, therefore, developed a procedure to correct for negative numbers where only positive numbers are theoretically possible. This correction procedure is based on double bookkeeping since products produced in the Supply table are used in the Use table. Thus, all corrections must be made in two places of the original tables. The erroneous negative values have been set to zero, booking the associated correction in the row and column discrepancies of the original FIGARO tables. Subsequently, all elements of the row and column correction factors, including those in the original FIGARO tables, are set to zero and are booked in the columns of stocks that absorbed all the errors in the original tables. After this preliminary step, step 1 in the procedure to construct the regional SAMs is the regionalisation of the country-level SUTs of the FIGARO dataset. 12 employment shares with country-level data for median hourly earnings by educational attainment level. The process is described in detail below. 5.1 Employment and wage shares by educational attainment We initially set the constraints regarding the working population of the regions. These are defined by the hours worked for the employed, which will ultimately vary between 35-45 hours per week depending on the regions and sectors. The associated skill shares from the LFS corresponding to these hours are then used to extract the skill-specific employed populations using the Eurostat series “Employment (thousand persons) by NUTS 3 regions [nama_10r_3empers]”. The first step is to import the hours worked and employment and fix any problematic cases. Zero values are replaced with missing values due to data paucity (respondents in the survey report that they are employed, but work zero hours). In some cases, not all skill categories are available within regions and sectors. In occurrences like this, the regional average across sectors is applied for each skill. If this is not available, the country average is applied. The same procedure is used when some sectors are missing. 4 From the hours and level of employment, we construct hours per week per region per sector. We obtain the 95 and 25 percentiles of the distribution of hours per week to be used to get rid of outliers and unreliable values. For values exceeding these percentiles, we replace them with the region values and recheck the percentiles. If necessary, country values are used to replace extreme values leading to a distribution of hours worked between 35 and 45 hours. The replacements could affect either or both the level of employment and the hours worked. This is to ensure that the updated hours worked per week per employee fall between 95 and 25 percentiles of the distribution. Then we obtain the employment shares as the share of the population working as Low, Medium or High skilled within a region-sector. 5.2. Wage shares and employment by region-sector-skill The wage shares are calculated using the Eurostat “median hourly earnings, all employees (excluding apprentices) by educational attainment level [earn_ses_pub2i]” and the employment share. These wage shares are used to alter the compensation of employees and the wages and salaries in the SAMs. Regional-sector employment is reported in the national accounts series “Employment (thousand persons) by NUTS 3 regions [nama_10r_3empers]”. These are then broken down into skills using the employment shares. 4 The Netherlands publish LFS data only at the country level. Thus, 12 copies of the country level hours and employment are created (each region representing 1/12th as there is no additional weighting applied). 13 6 Results 6.1 Data showcase: regional export analysis The regional granularity of the new dataset allows for levels of analysis that are not possible to perform with publicly available data. As an example, in this subsection we analyse the destination of exports at the regional level. In particular, what share of the exports of a given region has its own country as a final destination? Figure 1 provides an answer to this question. Figure 1: Share of trade with a region within the same country as destination Source: authors’ calculations. Figure 1 shows the within-country export intensity, meaning the share of trade flow volume that has the region as origin and a region in the same country as destination. Some regularities are immediately apparent. Capital regions tend to have a comparatively larger share of exports going to their own countries. These exports come with particular intensity from services sectors, in particular Professional, administrative and support activities (M-N sectors) and Information and communication (J sector). In these sectors, the volume of the flows going from the capital region to the other regions in the country is larger than what we could expect given the GVA share of these sectors in the capital regional economy (not shown for simplicity). The economies of scale and network associated with these sectors and the ease of selling these services in other regions within a country would explain their tendency to agglomerate in the 14 most populated regions. The map also shows many large non-capital regions that are very dependent on their national markets, making them amenable to policies aimed at supporting the internationalization of their industries. At the other end of the spectrum, we observe some regions with relatively low dependency on their national market as destination of their exports. While there are many reasons, we briefly explore two. First, Figure 2 shows the share of exports in a given region that has the German market as destination. A significant number of European regions are deeply connected to the German market. This includes most neighbouring areas to Germany in Austria, the Czech Republic, Hungary, Western Poland, Eastern Netherlands and Northeast France. In all cases, these exports are concentrated in the Manufacturing sector, either as intermediate inputs for the German manufacturing sector or as final goods for consumption: as in the case above, the size of the flows is larger than what we could expect given the GVA share of the manufacturing sector in the origin regions. Figure 2: Share of trade with a German region as a destination Source: authors’ calculations. 15 Figure 3: share of trade with outside the EU as destination Source: authors’ calculations. Second, Figure 3 shows the share of exports from a given region that have the ROW, understood as outside the EU, as destination. The regions with higher export intensity to the ROW are scattered around Europe, and a deeper analysis allows us to distinguish many different typologies. In some cases, this high intensity is simply the case of a strong relation with neighbouring countries outside the EU, like in the Baltics, Cyprus or Northern Sweden. Other regions have specialised in exporting manufactured goods, like cars for some regions in Hungary and Czechia or electronics in Southwest Ireland and Southern Netherlands. On the other hand, some regions have specialised in exporting services: financial services in Malta, Luxembourg and Cyprus; Information and communication services in Dublin or Trade related services in the Netherland or Denmark. 6.2 Model simulations We now provide a series of results from simulation exercises performed with the macroeconomic RHOMOLO model based on the latest dataset (RHOMOLO V4). Being able to produce macroeconomic consistent results with the new dataset on a well-established model like RHOMOLO provides a definitive check of the data constructed according to the procedure explained above, both of the internal consistency of the data and how it relates to the actual data observed in the real world. We produce simulation results for the following three macroeconomic 16 shocks: i) a total factor productivity (TFP) shock (purely supply-side); ii) a government consumption shock (purely demand-side); and iii) a transport cost shock (purely supply-side). The results are dependent on the calibration of the RHOMOLO model, using a set of auxiliary data which is described in detailed in Lecca et al. (2018). Of particular relevance for the shocks described below is the calibration of the interregional transport costs. The costs are estimated as the populationweighted average costs of road transport between pairs of cities within the NUTS 2 regions. These costs correspond to the generalized transport costs (GTC) in euros, capturing the distance and time-related costs of the optimal route between each pair of regions for a representative truck, also taking into account the actual geography. These costs are subsequently mapped into a matrix of iceberg transport costs. A full explanation of the methodology can be found in Persyn et al. (2022), and an application to the transport infrastructure investments of the 2014-2020 European cohesion policy is described by Persyn et al. (2023).  Shock i): 1% permanent increase of TFP The results from simulating a 1% permanent increase of TFP in all regions of the EU can be seen in Figures 4 and 5. The left panel in Figure 4 shows the deviations from the baseline in percentage over a 20-year horizon for a set of basic indicators at the EU level: GDP, household consumption, investment, employment, and the consumer price index (CPI). The right panel shows a histogram with the distribution of the 10-year deviation from baseline of each region’s GDP. Figure 4: permanent increase of TFP (a) percent deviations from baseline, (b) 10-year GDP deviation from baseline distribution Source: authors’ calculations. A TFP shock immediately increases production, as TFP is part of the production function. The increased productivity makes labour and capital more valuable. Firms hire more of both, leading to an improvement of the labour market and 17 more investment. Rising employment leads to higher wages, allowing the now wealthier households to increase their consumption, further stimulating the economy. The all-around improvements in technology enhance trade and lead to lower prices. All the EU regions benefit from the shock, with a relatively compact distribution of the positive economic effect. Among other determinants, there is a moderate negative correlation between the saving rate in a region and the size of the GDP impact, a consequence of the aforementioned second-round demand effects: in regions with a low savings rate, households will tend to consume more of the newly created wages. Also at the regional level, Figure 5 shows the GDP impact on the year following the shock. This exercise allows us to see results which are more directly linked to the dataset, less influenced by the dynamics of the model. On this very short run, one of the main determinants of the effect of increases to TFP on regional GDP is the labour share of the region. On impact, regions with higher labour shares experience larger effects on GDP, as labour demand increase more than in regions with lower shares, and capital has had no time to adjust through investment. Over time this effect reverses and regions with lower labour shares, and which are therefore more capital intensive, experience larger GDP increases. Following the shock, firms start adjusting their capital towards their desired levels by increasing investment. In more capital-intensive regions the demand for capital will increase more following the TFP shock. Consequently, all else equal, regions with lower labour shares will see larger levels of investment and subsequent larger increases to their regional GDP in the long term. 18 Figure 5: 1-year GDP deviation from baseline following a 1% TFP increase Source: authors’ calculations.  Shock ii): 1% permanent increase in government consumption The second shock is a 1% permanent increase in government consumption in all EU regions. Contrary to the previous shock, which affects the economy primarily through the supply side, this shock is fundamentally a demand shock. The increased government consumption stimulates the economy as the firms increase production to cover the new demand. The extra production requires of additional workers and investment to be met, raising both. Furthermore, the extra demand also results in an increase in prices as the economy heats up. 19 Figure 6: 1% permanent increase of Government consumption (a) per cent deviations from baseline, (b) 10-year GDP deviation from baseline distribution Source: author’s calculations As per Figure 6 (right panel), in terms of distribution, most regions appear concentrated around an increase of 0.05 per cent in GDP after 10 years, with a substantial number of regions to the right of that increase. Some negative values are also observed. Most of the negative cases can be explained by the loss of competitiveness induced by the increased prices mentioned above, combined with a relatively high dependence on exports outside the EU. On impact (see Figure 7), the regions that benefit the most are those with a larger private consumption to GDP ratio. These regions present a larger propensity to consume and, therefore, the newly created government consumption will have a larger effect in those regions relative to other regions with lower shares. 20 Figure 7: 1-year GDP deviation from baseline following a 1% Government consumption increase Source: authors’ calculations.  Shock iii): 1% permanent decrease in transportation costs Finally, we show in Figures 8 and 9 the results for a permanent 1% decrease in transportation costs in all EU regions, another supply-side shock. Both firms and consumers benefit from the shock. On the firms’ side, cheaper inputs lead to increased production, followed by higher investment and employment. Consumers now have access to cheaper imports, together with some additional wages due to the increased production. As for prices, we observe two competing effects: on one hand, the increased activity will tend to push prices up, while on the other hand, lower transportation costs translate to potential efficiency gains as producers can have access to cheaper inputs. At first, we see the former effect slightly dominating, but as time passes and firms have time to adjust, the latter dominates and prices go down. 21 Figure 8: 1% permanent decrease of transport costs (a) per cent deviations from baseline, (b) 10-year GDP deviation from baseline distribution Source: authors’ calculations. In terms of regions, the most interesting feature is the existence of some relatively extreme positive values. The impact of decrease in transportation costs has the largest effect on those regions that initially have relatively large transportation costs (islands like Madeira or Açores, mountain regions like the Aosta Valley in Italy, etc.) and in the regions whose imports come mainly from the EU. A similar picture can be seen on impact, as shown in Figure 9: regions with large relative transport costs have larger gains. However, we also observe relatively high gains in regions in and around Germany; these are regions with relatively strong trade flows as a percentage of GDP that benefit from the suddenly lower transport costs. 28 EL65 Peloponnisos ES11 Galicia ES12 Principado de Asturias ES13 Cantabria ES21 País Vasco ES22 Comunidad Foral de Navarra ES23 La Rioja ES24 Aragón ES30 Comunidad de Madrid ES41 Castilla y León ES42 Castilla-La Mancha ES43 Extremadura ES51 Cataluña ES52 Comunidad Valenciana ES53 Illes Balears ES61 Andalucía ES62 Región de Murcia ES63 Ciudad Autónoma de Ceuta ES64 Ciudad Autónoma de Melilla ES70 Canarias FI19 Länsi-Suomi FI1B Helsinki-Uusimaa FI1C Etelä-Suomi FI1D Pohjois- ja Itä-Suomi FI20 Åland FR10 Ile-de-France FRB0 Centre — Val de Loire FRC1 Bourgogne FRC2 Franche-Comté FRD1 Basse-Normandie FRD2 Haute-Normandie FRE1 Nord-Pas de Calais FRE2 Picardie FRF1 Alsace FRF2 Champagne-Ardenne FRF3 Lorraine FRG0 Pays de la Loire FRH0 Bretagne FRI1 Aquitaine FRI2 Limousin FRI3 Poitou-Charentes FRJ1 Languedoc-Roussillon FRJ2 Midi-Pyrénées FRK1 Auvergne FRK2 Rhône-Alpes FRL0 Provence-Alpes-Côte d’Azur FRM0 Corse HR03 Jadranska Hrvatska HR04 Kontinentalna Hrvatska HU11 Budapest HU12 Pest HU21 Közép-Dunántúl HU22 Nyugat-Dunántúl HU23 Dél-Dunántúl HU31 Észak-Magyarország HU32 Észak-Alföld HU33 Dél-Alföld IE04 Northern and Western IE05 Southern IE06 Eastern and Midland ITC1 Piemonte ITC2 Valle d’Aosta/Vallée d’Aoste ITC3 Liguria ITC4 Lombardia ITF1 Abruzzo ITF2 Molise ITF3 Campania ITF4 Puglia ITF5 Basilicata ITF6 Calabria ITG1 Sicilia ITG2 Sardegna ITH1 Provincia Autonoma di Bolzano/Bozen ITH2 Provincia Autonoma di Trento ITH3 Veneto ITH4 Friuli-Venezia Giulia ITH5 Emilia-Romagna ITI1 Toscana ITI2 Umbria ITI3 Marche ITI4 Lazio LT01 Sostinės regionas LT02 Vidurio ir vakarų Lietuvos regionas LU00 Luxembourg LV00 Latvija MT00 Malta NL11 Groningen NL12 Friesland (NL) NL13 Drenthe NL21 Overijssel NL22 Gelderland NL23 Flevoland NL31 Utrecht NL32 Noord-Holland NL33 Zuid-Holland NL34 Zeeland NL41 Noord-Brabant NL42 Limburg (NL) PL21 Małopolskie 29 PL22 Śląskie PL41 Wielkopolskie PL42 Zachodniopomorskie PL43 Lubuskie PL51 Dolnośląskie PL52 Opolskie PL61 Kujawsko-pomorskie PL62 Warmińsko-mazurskie PL63 Pomorskie PL71 Łódzkie PL72 Świętokrzyskie PL81 Lubelskie PL82 Podkarpackie PL84 Podlaskie PL91 Warszawski stołeczny PL92 Mazowiecki regionalny PT11 Norte PT15 Algarve PT16 Centro (PT) PT17 Área Metropolitana de Lisboa PT18 Alentejo PT20 Região Autónoma dos Açores PT30 Região Autónoma da Madeira RO11 Nord-Vest RO12 Centru RO21 Nord-Est RO22 Sud-Est RO31 Sud-Muntenia RO32 Bucureşti-Ilfov RO41 Sud-Vest Oltenia RO42 Vest SE11 Stockholm SE12 Östra Mellansverige SE21 Småland med öarna SE22 Sydsverige SE23 Västsverige SE31 Norra Mellansverige SE32 Mellersta Norrland SE33 Övre Norrland SI03 Vzhodna Slovenija SI04 Zahodna Slovenija SK01 Bratislavský kraj SK02 Západné Slovensko SK03 Stredné Slovensko SK04 Východné Slovensko 30 Table A2: NACE elements in FIGARO (64 at the national level, 56 after regionalisation) Code (64 activities) Description Code (56 activities) A01 Crop and animal production, hunting and related service activities A01 A02 Forestry and logging A02-A03 A03 Fishing and aquaculture A02-A03 B Mining and quarrying B C10T12 Manufacture of food products, beverages, and tobacco products C10T12 C13T15 Manufacture of textiles, wearing apparel, and leather and related products C13T15 C16 Manufacture of wood and of products of wood and cork, except furniture; manufacture of articles of straw and plaiting materials C16 C17 Manufacture of paper and paper products C17 C18 Printing and reproduction of recorded media C18 C19 Manufacture of coke and refined petroleum products C19 C20 Manufacture of chemicals and chemical products C20 C21 Manufacture of basic pharmaceutical products and pharmaceutical preparations C21 C22 Manufacture of rubber and plastic products C22 C23 Manufacture of other non-metallic mineral products C23 C24 Manufacture of basic metals C24 C25 Manufacture of fabricated metal products, except machinery and equipment C25 C26 Manufacture of computer, electronic and optical products C26 C27 Manufacture of electrical equipment C27 C28 Manufacture of machinery and equipment n.e.c. C28 C29 Manufacture of motor vehicles, trailers and semi-trailers C29 C30 Manufacture of other transport equipment C30 C31_32 Manufacture of furniture and other manufacturing C31_32 C33 Repair and installation of machinery and equipment C33 D35 Electricity, gas, steam and air conditioning supply D35 E36 Water collection, treatment and supply E36 E37T39 Sewerage; waste collection, treatment and disposal activities; materials recovery; remediation activities and other waste management services E37T39 F Construction F G45 Wholesale and retail trade and repair of motor vehicles and motorcycles G45 31 G46 Wholesale trade, except of motor vehicles and motorcycles G46 G47 Retail trade, except of motor vehicles and motorcycles G47 H49 Land transport and transport via pipelines H49 H50 Water transport H50 H51 Air transport H51 H52 Warehousing and support activities for transportation H52 H53 Postal and courier activities H53 I Accommodation and food service activities I J58 Publishing activities J58 J59_60 Motion picture, video and television programme production, sound recording and music publishing activities; programming and broadcasting activities J59_60 J61 Telecommunications J61 J62_63 Computer programming, consultancy and related activities; information service activities J62_63 K64 Financial service activities, except insurance and pension funding K K65 Insurance, reinsurance and pension funding, except compulsory social security K K66 Activities auxiliary to financial services and insurance activities K L68 Real estate activities L68 M69_70 Legal and accounting activities; activities of head offices; management consultancy activities M69_70 M71 Architectural and engineering activities; technical testing and analysis M71 M72 Scientific research and development M72 M73 Advertising and market research M73 M74_75 Other professional, scientific and technical activities; veterinary activities M74_75 N77 Rental and leasing activities N77 N78 Employment activities N78 N79 Travel agency, tour operator and other reservation service and related activities N79 N80T82 Security and investigation activities; services to buildings and landscape activities; office administrative, office support and other business support activities N80T82 O84 Public administration and defence; compulsory social security O84 P85 Education P85 Q86 Human health activities Q86 32 Q87_88 Residential care activities; social work activities without accommodation Q87_88 R90T92 Creative, arts and entertainment activities; libraries, archives, museums and other cultural activities; gambling and betting activities R-T R93 Sports activities and amusement and recreation activities R-T S94 Activities of membership organisations R-T S95 Repair of computers and personal and household goods R-T S96 Other personal service activities R-T T Activities of households as employers; undifferentiated goods - and services - producing activities of households for own use R-T 33 Table A3: List of NUTS-2 statistics for EU27 Data source Availability Sectoral/commo dity details Geographic al details Notes Eurostat regional accounts – GDP, households’ incomes, employment and wages 2000-2019 11 NACE Rev. 2 sectors (covering the whole economy) NUTS 1 2016 and NUTS 2 2016 regions of EU27 + NO + CH + IS + MK + TR + UK Eurostat SBS – employment and wages 2008-2019 89 NACE Rev. 2 divisions (covering sectors and private services) NUTS 1 2016 and NUTS 2 2016 regions of EU27 + UK The 89 divisions are aggregated up to the 56 sectors used in the regional dataset Eurostat - Economic accounts for agriculture (agr_r_accts) 1980-2019 Detailed A sector products NUTS 1 2016 and NUTS 2 2016 regions of EU27 + UK Eurostat - Hospital beds by NUTS 2 regions (hlth_rs_bdsrg) 1993-2020 NUTS 1 2016 and NUTS 2 2016 regions of EU27 + NO + CH + IS + MK + TR + RS Used to regionalize Q sector GVA Eurostat - Students enrolled by education level (educ_uoe_enra11) 2013-2020 NUTS 1 2016 and NUTS 2 2016 regions of EU27 + NO + CH + IS + MK + TR + RS Used to regionalize P sector GVA