Essays on the Definition, Measurement and Spatial Distribution of Creative Industries and Creative Employment in Portugal

Essays on the Definition, Measurement and Spatial Distribution of Creative Industries and Creative Employment in Portugal Sara Cristina Santos Cruz Tese de Doutoramento em Economia Supervised by: Aurora A. C. Teixeira Björn T. Asheim 2014 ii Biographical Note Sara Cristina Santos Cruz was born on 23 May, 1978, in Coimbra, Portugal. In 2001, she concluded her degree in Economics at the Faculty of Economics of the University of Coimbra, Portugal. In 2002, she was awarded, together with the 120 best national young graduates in economics and management, the prize of High Potential Young Graduates of SONAE group, Portugal. From 2002 to 2007, she performed a management career in two creative leading groups - L’Oréal Portugal (as brand manager; department of marketing); and the publishing group Porto Editora (as sub-director of department). After the enriching passage through the world of management, the deep interest in the academic field called for the return to studies, and in 2008, she completed her Master degree in Economics of Development, on the topic of Industrial location - Industrial clusters, at the School of Economics and Management, University of Porto, Portugal. From 2007 to 2011, she was researcher at the Center for Economics and Finance (Cef.Up), School of Economics and Management, University of Porto, within the scope of an international research project entitled as “Resources, Rights and Capabilities - in search of social foundations for Europe”, where she had opportunity to develop work on the textile industries and workers of the Ave Valley, northern Portugal, in the fields of labour and regional economics. From 2011 to date, she has been committed to the Foundation for Science and Technology, Ministry of Education and Science, as a doctoral student/ researcher in Regional Economics (School of Economics and Management, University of Porto, Portugal), having developed research on the industrial location and spatial distribution of creative industries and creative employment in Portugal. During her master and doctoral programs, she published articles in co-authorship with Aurora Teixeira, in the international journals Regional Studies, European Planning Studies and Annals of Regional Science, and has also been referee of the journals Environment and Planning C, European Planning Studies and The Professional Geographer/ Association of American Geographers. iii Acknowledgements This Doctoral Thesis is the result of years of research and writing. This personal and demanding journey has taught me to explore new grounds in the face of new challenges, with constant perseverance and self-determination. I would like to express my sincere gratitude to my supervisor - and a mentor for me -, Aurora Teixeira, for her remarkable guidance, encouragement and companionship over the years. Much of the motivation for this thesis came from her enlightening comments and advices, which encouraged me to go further and better in all the steps of this research. It has been really a privilege to share from her scientific knowledge and her human values during this entire journey. I would also like to thank my co-supervisor, Björn Asheim, for his kindness and prompt availability; Anabela Carneiro, for her valuable suggestions when dealing with the data; Paulo Guimarães, for his helpful recommendations; and Armindo Carvalho, for the useful advices on statistical methods of spatial analysis. My sincere thanks to the School of Economics and Management, University of Porto (FEP-UP) and all its staff, a very special ‘home’ to me along this decade, where I met integrity, inspiration and values that will always be part of my life. I also acknowledge the courtesy of the Gabinete de Estratégia e Estudos, Ministry of Economy of Portugal, for permitting the access to data which turned this study possible, as well as to the Foundation for Science and Technology, Ministry of Education and Science, for allowing the financial support through a doctoral grant (SFRH/BD/69571/2010). In the course of these years, I thank all those who have accompanied me, both in professional life and at home, from my colleagues to my dearest friends. They have all contributed to this effort in one way or another. To my family, whom I thank for their beloved presence. Words cannot express how I am grateful for their constant support, encouragement and profound conviction in all the circumstances of my life. This doctoral thesis is dedicated to them. iv Abstract Over the past decade, the academic and political debate on industrial location has gradually come to highlight the geography of knowledge-intensive and creative activities as drivers of regional growth. Following the original UK government’s report on the mapping of creative industries in 1998, and Richard Florida’s study on the ‘creative class’ in 2002, a considerable amount of case studies on creative clusters, cultural quarters and creative cities has been put forward in several regions of the developed world. Within the empirics of location, there has been an increasing interest in the analysis of the spatial distribution of creative industries and their importance in urban growth. These industries have a tendency to co-locate and their uneven spatial patterns are explained by territorial factors or location determinants. Despite all the novelties, so far, literature has hardly achieved common agreement on what defines and constitutes the Creative economy. In this context, the present Doctoral thesis - Essay 1 and Essay 2 - is first dedicated to the systematization of the growing corpus of literature on creative industries and creative occupations, by providing a thorough survey in terms of existing definitions and taxonomies. Here, it is undertaken an extensive literature review related with the different methodological approaches on the measurement of the Creative economy, both in terms of creative industry sectors and of creative occupations. Then, in Essay 3, it is developed a measurement approach that properly defines the Creative economy in Portugal, involving both creative industries (industry sectors) and creative occupations (employment). The data and information provided by that measurement approach and extracted from the Linked Employer-Employee databases (Gabinete de Estratégia e Estudos, Ministry of Economy of Portugal), allows an exploratory analysis of the geographical patterns of creative industries and creative employment in all the 308 municipalities of Portugal, using the software of spatial analysis ArcGIS 10.1 ®. Finally, in order to understand the reasons why creative industries locate in particular regions, in Essay 4, it is carried out the analysis of these firms’ location determinants, in all the Portuguese municipalities, using a recent Discrete Choice Model approach on the modelling of their location behaviour. v Resumo Ao longo da última década, o debate académico e político sobre localização industrial tem vindo gradualmente a destacar a geografia das atividades criativas e baseadas no conhecimento como motores de crescimento regional. Na sequência do relatório governamental do Reino Unido sobre o mapeamento das indústrias criativas, em 1998, e do estudo de Richard Florida sobre a ‘classe criativa’, em 2002, uma quantidade considerável de estudos sobre clusters criativos, bairros culturais e cidades criativas tem sido apresentada em várias regiões do mundo desenvolvido. Na literatura empírica sobre localização, tem havido um interesse crescente na análise da distribuição espacial das indústrias criativas e da sua importância no crescimento urbano. Essas indústrias revelam tendência a concentrarem-se geograficamente e os seus padrões de distribuição irregulares são explicados por fatores territoriais ou determinantes de localização. Apesar de todos os desenvolvimentos até à data, a literatura dificilmente tem alcançado consenso sobre o que define e constitui a Economia Criativa. Neste contexto, a presente tese de Doutoramento - Essay 1 e Essay 2 - é primeiramente dedicada à sistematização do crescente corpus de literatura sobre indústrias e ocupações criativas, fornecendo um estudo aprofundado em termos de definições e taxonomias existentes. Aqui, é realizada uma extensa revisão da literatura relacionada com as diferentes abordagens metodológicas sobre a mensuração da economia criativa, tanto em termos de setores industriais criativos como de ocupações criativas. Em seguida, no Essay 3, é desenvolvida uma metodologia de mensuração com vista a definir apropriadamente a economia criativa em Portugal, envolvendo tanto as indústrias criativas (setores industriais) como as ocupações criativas (emprego). Com base nessa metodologia e nos dados extraídos dos Quadros de Pessoal (Gabinete de Estratégia e Estudos, Ministério da Economia, Portugal), é levada a cabo uma análise exploratória dos padrões geográficos das indústrias criativas e do emprego criativo nos 308 concelhos de Portugal, utilizando o software de análise espacial ArcGIS 10.1 ®. Finalmente, a fim de compreender as razões pelas quais as indústrias criativas se localizam em regiões específicas, no Essay 4, é levado a cabo o estudo dos determinantes de localização dessas empresas, utilizando uma recente abordagem com base nos Modelos de Escolha Discreta para modelizar o seu comportamento de localização. vi Contents ESSAY 1 ....................................................................................................................................... 1 The magnitude of creative industries in Portugal: what do the distinct industry-based approaches tell us? ........................................................................................................................................... 2 Abstract ..................................................................................................................................... 2 1. Introduction ........................................................................................................................... 3 2. Approaches to the measurement of Cultural and Creative industries: a brief review ........... 4 3. Methodological considerations ........................................................................................... 10 4. Estimating the weight of Portuguese creative industries..................................................... 11 4.1. According to the main industry-based approaches in literature ................................... 11 4.2. A proposal based on ‘core’ creative industries ............................................................ 15 5. Conclusions ......................................................................................................................... 18 Acknowledgements ................................................................................................................. 20 References ............................................................................................................................... 20 Annex 1 ................................................................................................................................... 25 ESSAY 2 ..................................................................................................................................... 36 Assessing the magnitude of Creative employment: A comprehensive mapping and estimation of existing methodologies ............................................................................................................... 37 Abstract ................................................................................................................................... 37 1. Introduction ......................................................................................................................... 38 2. Measuring the creative employment: a review of the main methodologies ........................ 39 2.1. The industrial perspective: conventional industry-based approaches .......................... 40 2.2. The sociological perspective: occupational-based approaches .................................... 41 2.3. The combined industryand occupation-based approaches ......................................... 42 3. Mapping the distinct methodologies used in literature ....................................................... 47 3.1. Industrial perspective: conventional approach ............................................................. 47 3.2. Sociological perspective: occupation-based approaches .............................................. 49 3.2.1. Florida’s original proposal .................................................................................... 49 3.2.2. Proposals following Florida’s ............................................................................... 51 3.2.3. Refinements of Florida’s proposal ........................................................................ 52 3.3. The combined industry and occupation-based approach ............................................. 54 3.3.1. The creative trident ............................................................................................... 54 3.3.2. The 2010 DCMS proposal .................................................................................... 57 vii 4. Computing the magnitude of the creative employment according to the existing methodological approaches ..................................................................................................... 61 5. Concluding remarks ............................................................................................................ 65 Acknowledgments ................................................................................................................... 66 References ............................................................................................................................... 66 Annex 2 ................................................................................................................................... 71 ESSAY 3 ..................................................................................................................................... 74 The neglected heterogeneity of spatial agglomeration and co-location patterns of creative employment: Evidence from Portugal ........................................................................................ 75 Abstract ................................................................................................................................... 75 1. Introduction ......................................................................................................................... 76 2. Empirical literature on the location of creative industries and occupations: a brief review 78 3. Methodology ....................................................................................................................... 87 4. Agglomeration and co-location of core creative employment in Portugal: results ............. 91 4.1. Agglomeration of creative employment in each core creative group........................... 91 4.2. Co-location of core creative employment .................................................................... 94 4.3. Characterization of core creative clusters based on regional indicators....................... 98 5. Conclusions ......................................................................................................................... 99 Acknowledgements ............................................................................................................... 102 References ............................................................................................................................. 102 ESSAY 4 ................................................................................................................................... 108 The determinants of spatial location of creative industries start-ups: Evidence from Portugal using a discrete choice model approach .................................................................................... 109 Abstract ................................................................................................................................. 109 1. Introduction ....................................................................................................................... 110 2. Empirical literature on the determinants of industrial location ......................................... 111 Agglomeration Economies ................................................................................................ 112 Talent/ Human Capital ...................................................................................................... 116 Tolerance ........................................................................................................................... 118 Technology ........................................................................................................................ 119 Inter-territorial spillovers .................................................................................................. 121 3. Methodology ..................................................................................................................... 123 3.1. Data considerations .................................................................................................... 123 3.2. Location determinants: variables selected and respective indicators ......................... 124 viii 3.3. A description of the selected modelling approach: Discrete Choice Model .............. 128 3.4. A description of the selected econometric estimation: Conditional Logit Model ...... 129 4. Empirical results ............................................................................................................... 131 4.1. Results for creative firms as a whole ......................................................................... 131 4.2. Empirical results by creative industry sector ............................................................. 141 5. Conclusions ....................................................................................................................... 144 Acknowledgements ............................................................................................................... 146 References ............................................................................................................................. 146 Appendix ............................................................................................................................... 152 ix List of Figures Figure 2. 1: The boundaries of the creative employment according to the main measurement perspectives .............................................................................................. 46 Figure 2. 2: The magnitude of the creative employment in Portugal according to the main measurement perspectives and approaches ............................................................ 63 Figure 3. 1: Agglomeration patterns of Portuguese core creative groups ...................... 92 Figure 3. 2: Location patterns of core creative clusters, mainland Portugal and islands 97 Figure 4. 1: Number of new creative establishments in Portugal by municipality (our database; n=369 establishments/ j=308 municipalities), in 2009 ................................. 132 5 Similarly, UNCTAD (2004, 2008) organizes the creative sector in terms of “upstream activities” (cultural activities in strict sense, such as the performing and visual arts), and “downstream” market-driven industries (e.g., advertising, publishing or media related activities). Under this approach, cultural activities represent a segment of the entire universe of creative industries. From all the frameworks presented, it is possible to draw distinct approaches according to each template’s characteristics and rationale: i) Economic/ Industrial approach (e.g., DCMS template; WIPO Copyright model), based on the fact that creative industries use creativity as an input and protect their output with copyright/ intellectual property rights, earning profits therefrom; ii) Cultural Content perspective (e.g., Symbolic Model; Concentric Circles Model), stressing the intrinsic value of culture and popular arts as the major argument to group creative industries; and iii) Upstream-Downstream branches of activity approach (e.g., Heng et al., 2003; Scott, 2004; UNCTAD, 2008), distinguishing between upstream and downstream industries in the creative economy. It is clear from Table 1.1 that the extent of the creative sector is vast and diversified, comprising a range of industries that goes from purely aesthetic or cultural fields (e.g., visual and performing arts, cultural heritage) to highly knowledge-intensive segments (e.g., digital, technological, service-based activities), most of them revealing strong interdependencies among each other (UNCTAD, 2008). Moreover, all the approaches seem quite arbitrary and subjective in their selection and listing of industries. Each model posits arbitrarily different valuations in terms of core and peripheral, included or excluded industries, according to their interpretation of creative industries. Each template has distinctive characteristics which reveal advantages but also limitations (see a synthesis in Table 1.2). Under the Economic/ Industrial approach, creative industries are the set of cultural and copyright industries that use creativity in their production process and generate output protected by intellectual rights (DCMS, 1998, 2001). In this perspective, the DCMS model presents advantages as a supporting template for policy-making and governmental decision. However, the arbitrariness of its “eclectic list” (Cunningham, 2002: 54) and the lack of compatibility with available classification systems (Higgs and Cunningham, 2008), impose some limitations to the measurement of the creative economy. 6 The WIPO Copyright model focuses on intellectual property/ copyright as a representation of the creativity incorporated in goods (WIPO, 2003). This involves industries directly or indirectly related with the creation, manufacturing/ production, broadcasting and distribution of copyrighted goods. An additional set of “interdependent” and “partial copyright industries” includes activities where intellectual property does not play a major role in their production processes. In this context, creative industries are directly entailed in the intellectual property of their output. The broad criteria of the WIPO model are not free from critiques. In this line of reasoning, not only cultural and creative activities reliant on copyright, but all the industries that create or commercialize patents/ intellectual rights should have to be included (e.g., “pharmaceuticals, electronics, engineering, chemicals”) (Hesmondhalg, 2008: 560). One advantage of this model is that it takes into account the linkages between the digital economy (ICTs) and the diffusion of cultural/ creative outputs which have wide effects on the creative economy. Major drawbacks stem from difficulties in assessing such impacts, which have proved to be hard to estimate or preview (Handke, 2006), and in quantifying the creative economy, given the wide-ranging extent of sectors considered as creative industries, i.e., all the industries that are based on copyright (Thorsby, 2008a). Another major limitation resides in the assessment of the copyright factor † associated with each partial and interdependent copyright industry in this approach (WIPO, 2003; Chow and Leo, 2005). † The copyright factor (or weighting) is “the percentage indicating the portion of a particular activity/industry that can be attributed to copyright-based activities” (WIPO, 2003: 85). 7 Table 1. 1: Mapping creative industries – industry-based approaches and respective templates Economic/ Industrial approach Cultural Content approach Upstream-Downstream Branches of Activity approach INDUSTRIES DCMS (UK)* WIPO Copyright Symbolic Model Concentric Circles Model Heng et al. (2003), Scott (2004) UNCTAD (2004) Performing Arts Core Core Core Production Activities Arts Visual Arts/Graphic Arts Core Music Core Musical Instruments Interdependent Literature Core Arts & Antiques Market Core Heritage Wider cultural Heritage Museums/Galleries Other core Creative Arts Peripheral Production Activities (local dense networks) Architecture Core Partial Related Fashion Core Borderline Design Core Functional creations Crafts Core Clothing/Footwear Partial Photography Core Film Core Core Core Other core Distribution Activities (global networks of distribution) Media Video Core Wider cultural TV and Radio Core Publishing Core Advertising Core Related New Media [functional creations] Internet Software/Digital Contents Core Core Borderline Computer Media Core Core Wider cultural Collecting Societies Core Sport Borderline Recording (sound) Interdependent Wider cultural Paper Photocopiers/Photographic Equipment Consumer Electronics Borderline Household Goods Partial Toys Sources: UNCTAD (2008); DCMS (2010). We used different shades of grey, ranging from the darkest, indicating the core cultural and creative activities, to the lightest, indicating more peripheral cultural and creative activities. 8 Table 1. 2: Mapping Creative Industries - advantages and drawbacks of each industry-based approach in literature and in our mapping with international/ national industry codes Approach Model/ Template Sectors considered/ Characteristics Advantages Disadvantages Economic/ Industrial perspective Focus on the level of creativity/ copyright component in final goods DCMS Model (DCMS, 1998, 2001, 2010) Creative industries include 13 sectors: Advertising, Architecture, Arts and Antiques, Crafts, Design, Designer Fashion, Video, Film and Photography, Music and Visual & Performing Arts, Publishing, Software, Computer Games and Electronic Publishing, TV and Radio Simple to use and workable. Based on a supporting template for policy-making and governmental decision. Selection of a restrictive number of creative sectors. Arbitrary exclusion of certain activities from the listing (e.g., Heritage, Museum, Recreation). Difficulty in separating the creative from non-creative component of industry codes related with activities that are not entirely ‘creative’. The Crafts sector cannot be captured by means of industry codes. Portions of codes taken – some degree of arbitrariness. WIPO copyright model (World Intellectual Property Organization, 2003) Copyright-based industries are discriminated in terms of: - Core Copyright-based Industries - Interdependent Copyrightbased Industries - Partial Copyright-based Industries More objective methodology on the selection of copyrightbased activities, since criteria lie on copyright goods and intellectually protected contents. Set of broad and all-inclusive industry codes involving the wholesale, retail sale and rental activities. Difficult to assess the creative or copyright-based part of each of the industry code considered. Difficulties in obtaining an appropriate copyright factor for Interdependent and Partial copyright-based industries. Cultural Content approach Focus on Activities that produce culture/ creative content vs. Activities using creativity as input to diffuse it through distribution networks Cultural Concentric circles model (KEA European Affairs, 2006) Creative and symbolic contents generated in the Core Cultural centre and transmitted through a succession of concentric circles. The four levels: i) Core Cultural Centre ii) Layer 1: Wider Core Cultural activities iii) Layer 2: Creative activities iv) Layer 3: Related Industries (ancillary services, equipment, supply services which facilitate the production and diffusion of cultural and creative contents). Emphasis on fine arts and on Cultural production. Importance of Fine Arts/ Culture as the epicentre of the creative economy. Representation with concentric layers, useful in policy analysis. Selection process reveals limitations: no consensus on defining / delimiting cultural and creative industries  no precise way of deciding which activities should be considered in the Cultural and those that should belong to the Creative sector. Mapping the Core of Cultural and Arts activities is strongly limited to the industry classification system, Some industry codes cannot be disaggregated into more detailed level  no way of separating production from distribution activities. Branches of activity approach UpstreamDownstream activities approach (Heng et al., 2003; Scott, 2004; UNCTAD, 2004, 2008) This approach distinguishes between: i) Creation activities Software production; Advertising production; TV& Radio; Publishing; Design; Architecture; Arts& Antiques Market; Performing, Visual arts & Music; Museums; Film & Video; Photography ii) Distribution + Ancillary Activities Software distribution; TV & Radio broadcasting; Publishing related services; Performing arts & Music distribution; Film & Video distribution; Photography related services Simple to use. Distinction between Creation/ Production activities and Distribution/ broadcasting activities. It facilitates the analysis of the interdependencies between creation and distribution activities. Limitations of industry classification codes in mapping either creation or distribution industries. Some industry codes, even at their maximum breakdown, include both creation and distribution activities  no way of separating production from distribution activities. Difficulties in quantifying linkages and interdependencies (e.g., spillovers, externalities, flows that surpass national borders) throughout the value-chain. 9 The Cultural Content approach includes both the Symbolic model, which envisages Fine Arts at the core of cultural and creative industries, and the Concentric Circles model, stressing that creative goods as symbolic contents (sound, text and image) are generated at the centre - core creative arts - and then transmitted through a succession of levels - concentric circles (Thorsby, 2008a,b; UNCTAD, 2008). A major feature of these templates, which can be considered as an advantage, is that they are reliant on a more narrow/ selective process of grouping creative industries, restricted to those that produce culture in a strict sense (Thorsby, 2008b). One critique of this perspective is that cultural/ creative contents can be resources and outcomes not only of purely cultural activities, but also of the entire creative economy (e.g., software, digital media, design, advertising). In this context, creative industries are considered as a broader arts economy (Potts, 2009). The Branches of Activity (Upstream-Downstream activities) approach categorizes the creative economy in terms of “upstream activities”, i.e., core cultural activities, and “downstream activities”, i.e., commercial and distribution industries, dedicated to the diffusion and commercialization of cultural contents (e.g., Heng et al., 2003; Scott, 2004; UNCTAD, 2008: 13). The strength of this perspective lies in the importance of tracing the linkages and interdependencies among all the industries that compose the value-chain, differentiating upstream from downstream segments (Scott, 2004). However, this advantage becomes a drawback when it comes to quantitatively measuring those interdependencies throughout the value-chain, such as dynamic spillovers, externalities and the flows that surpass national borders, as links are often established between local and transnational companies (Scott, 2004; Vang and Chaminade, 2007). Whereas in Cultural Concentric Circles or in Upstream-Downstream activities approaches there is some consensus in distinguishing activities that produce culture/ creative contents (e.g., literature, music, design) from those which use creativity as an input and diffuse it through broadcast and distribution networks (e.g., advertising, publishing, film, video, TV, radio), in the DCMS model or in the WIPO Copyright perspective, the focus is instead on the level of creativity/ copyrighted component that is incorporated in goods as the main factor for distinguishing the creative core (Table 1.2). 10 3. Methodological considerations Besides the intense debate surrounding the definition and delimitation of Creative Industries (CIs), estimations of their weight in the economy, usually in terms of employment, have been often performed using disparate and non-comparable datasets, involving information on distinct regions or countries, even when the same approach is used. In order to have a more precise idea of the differences between the existing methodologies, it is necessary to depart from a single dataset and map all the proposals according to their industry-based approach, using a comparable scheme of industry classifications. For this purpose, we undertook an extensive mapping of the approaches in literature – DCMS model; WIPO template of copyright-based industries; Concentric Circles model; UpstreamDownstream activities model – to measure the creative industries, as they were presented in Section 2. We used codes from the International Standard Industrial Classification of economic activities, ISIC (Revision 3.1 and the latest Revision 4), and the corresponding codes for the Portuguese economic activities, based on the most recent industrial nomenclature (Classificação das Actividades Económicas - Revision 3, CAE - Rev. 3). In order to be as accurate as possible in this mapping and the respective estimation of all the approaches analyzed, we used detailed 5-digit industry codes, the maximum breakdown of Portuguese industry classification. 1 Then, estimations of the dimension of CIs were carried out, using each mapped industry-based approach. This empirical exercise allowed identifying the distinguishing features of each approach and taking accurate comparisons among them, departing from the same database. To estimate the weight of creative industries as a percentage of national employment, we used data for Portugal, extracted from the Matched Employer-Employee Databases of GEE/ ME, Gabinete de Estratégia e Estudos, Ministry of Economy, Portugal. The data used is the latest available, from 2009, and covers all the industries and establishments operating in the national territory - mainland Portugal and Autonomous Regions - except for Public Administration servants and the self-employed. According to this dataset, the total employed population in 2009, in all the activity sectors, was 1 In Tables A1.1-A1.5, the details of these mappings are presented. 11 3.128.126 workers. All the figures obtained for each 5-digit industrial code have been extracted using the software STATA®. 2 All other methodological details and technical limitations of the data used are thoroughly described in Cruz and Teixeira (2013). 4. Estimating the weight of Portuguese creative industries 4.1. According to the main industry-based approaches in literature Estimations according to each of the mapped methodologies were accomplished using a unique database, so that all the information could be properly compared. The data on Portugal was extracted from the official employment datasets for 2009, and the results are presented in Table 1.3. Using the DCMS industry-based approach, it is estimated that Portuguese employment in the creative sectors (reference year 2009) accounts for approximately 2.5% of total national employment. As previously mentioned, this approach relies on a selective list of 13 creative sectors, inspired in the original DCMS methodology (cf. table A1.1). Furthermore, in cases where the sectors also comprised activities outside the creative economy (e.g., manufacturing activities), only a portion of the respective industrial code was considered, so as to capture only the creative activities. This perspective revealed to be restrictive both in the selecting process and when applying portions of industry codes to extract only the creative component. When analysing our data using the DCMS approach, the estimations led to very modest results. The estimate of 2.5% suggested that the use of such a selective approach on the creative core and the application of quite arbitrary portions of industrial codes could be underestimating the effective size of CIs in Portugal. We believe that this approach is more suited to the specific context of creative sectors in the country where it was first implemented (UK). The WIPO copyright approach stands at the opposite extreme. This approach is developed by an international organization and the criteria applied appear to be more objective and broader than that of DCMS. The methodological issues can be easily 2 The mapping exercise revealed to be a complex, time-consuming task, since a large number of empirical studies did not disclose their methodological procedures, taxonomy and their respective industry codes. Despite the suitable compatibility between ISIC - Rev. 4 and the Portuguese CAE - Rev. 3, the respective conversion was also a challenging task, since, in many cases, one ISIC code corresponded to several Portuguese 5-digit industry codes. In this case, to prevent ambiguity, the thorough interpretation of each creative sector’s context and knowledge on the details of each ISIC and Portuguese CAE code revealed to be crucial. 12 adopted by any country with a set of developed rules on the protection of intellectual property and copyrights. The WIPO approach is reliant on a broader definition of CIs which is based on copyright-based industries (cf. table A1.2). When using the WIPO approach, our estimates led to a national employment share of 4.6% in Portuguese copyright-based industries – Core, Partial and Interdependent. The weight of Core Copyright-based industries was 3.9% of total employment. The relative weight of Interdependent Copyright-based industries in total employment was 0.3%; and the relative share of Partial Copyright-based industries was 0.4% (cf. Table 1.3). One aspect stands out from the results obtained for the cases of Interdependent and Partial Copyright industries: their size in terms of relative weight in total employment appears confined to very modest values. These seemingly paradoxical results derive from the fact that although Portugal has a large number of workers in the apparel, textile and footwear industries, the proportion of those working in activities related with goods subject to copyright is very small. The copyright factors that were applied to Partial and Interdependent copyright industries, according to the WIPO methodology and to the available empirical studies, are thus responsible for the results obtained for those industries. Despite the need of appropriate copyright factors to apply to the Interdependent and Partial Copyright-based industries in our country, when seeking to capture only those activities related with copyrighted goods (and which can only be obtained through extensive business surveys), the WIPO approach proved to be more objective on the calculation of the potential size of creative industries in Portugal. Using the cultural approach of the Concentric Circles model (cf. Table A1.3), we obtained an estimate for the relative weight of Cultural and Creative industries in national employment of 3.7% (cf. Table 1.3). According to the methodology followed (KEA, 2006), the Core Cultural centre - composed by the fine arts and cultural/ artistic activities, such as ‘Visual arts’, ‘Performing arts’, ‘Photography’, ‘Heritage, museums and antique market’ activities - only represented 0.4% of total employment in Portugal. This is a modest result and should only be interpreted as indicative, since our industry nomenclature - CAE - Rev. 3, even at its maximum disaggregation of 5-digit codes, was not able to capture the majority of the activities involved in Culture and Fine Arts. 13 Table 1. 3: Estimating Creative Industries according to the existing industry-based approaches Industrybased Approaches Employment share of Creative Industries (relative weight in total economy employment) Portugal (2009) Studies from other countries using each methodology DCMS Model Core Creative industries (13 sectors) 2.5% UK (2009): 4.99%; UK (2010): 5.14%(a) Scotland (2007): 3.0%(b) WIPO copyright model Creative industries - Total Copyright-based Industries, where: 4.6% Hungary (2002): 6.0% Romania (2005): 3.7% Bulgaria (2005): 4.3% (c)  Core Copyright industries 3.9%  Partial Copyright industries 0.4%  Interdependent Copyright industries 0.3% Concentric circles model Total Cultural and Creative Industries where 3.7% Australia (2001): 3.6% Canada (2001): 4.0% New Zealand (2001): 4.1% UK (2001): 7.5% USA (2004): 3.8% (d)  Cultural Industries [Core Centre (0.4%) + Wider Core Cultural (1.4%)] 1.8%  Cultural and Creative Industries [Core Centre + Layer 1 + Layer 2] 2.6%  Cultural Industries + Creative Industries + Supporting related Industries [Core Centre + Layer 1 + Layer 2 + Layer 3] 3.7% UpstreamDownstream branches of activity Creative industries where: 4.1%  Creation/ production activities 2.3%  Distribution/ broadcasting activities 1.8% Notes: (a) DCMS (2011) does not take into account ‘Software Consultancy’ and ‘Business and domestic software development’: the estimates obtained reflect this fact; (b) Scottish Government Social Research (2009); (c) According to the figures presented in the WIPO reports (2005, 2008) and Tchalakov et al. (2007); (d)These figures were obtained using the Concentric Circles approach (Thorsby, 2008b: 155). In fact, this limitation is transversal to all the industry classification systems and a weakness of all industry-based approaches in literature: the extreme difficulty in capturing and discriminating cultural and artistic activities. This conclusion is corroborated when we analyse the contribution of Wider Core Cultural activities (Layer 1), i.e., ‘Film and video’, ‘TV and radio’, ‘Software and computer games publishing’, ‘Music’ and ‘Literature and press’ in total employment, which amounted to almost 1.5% of the national workforce in 2009. This means that the industrial classification system in use, CAE - Rev. 3, revealed a greater ability to capture the activities that were included in the first layer of activities, in contrast to those in the Core cultural centre. Yet, similarly to the Core centre, difficulties in assessing Creative activities (Layer 2), 14 such as ‘Design and Fashion’, ‘Architecture and engineering’ and ‘Advertising’, arose when we used industrial codes and assumed portions to only capture the creative component of ‘Fashion design’ and of ‘Engineering services’, given the limitations of the industry classification system. The estimates obtained led to a relative weight of 0.8% in total Portuguese employment. At last, the vast category of industries supporting cultural and creative industries (Layer 3), which ranges from ancillary services to the supply of equipment and resource materials, including the ICT sector, represented 1.1% of total employment. Once more, better estimates obtained in this segment derived from the fact that the sectors here included had more-detailed descriptions in terms of the industrial nomenclature used. Despite the relevance of the Concentric Circles Model and its cultural approach to creative industries, the extreme difficulty of industry classification codes to describe and assess artistic/ cultural activities imposes limitations to a fair measurement of cultural and creative industries in Portugal. Finally, based on the Upstream-Downstream model of creative activities (cf. Table A1.4), the estimations suggest that all the industries involved both in Creation and Distribution activities contributed to 4.1% of national employment, with Creation and Production Industries accounting for 2.3% of total employment, and Distribution and Ancillary activities for 1.8% (cf. Table 1.3). Despite the relative simplicity of putting this approach into practice, difficulties arose when it was necessary to separate exclusively Creation activities from their associated Distribution/ Broadcasting activities, even when using detailed 5-digit industry codes. 3 The methodologies detailed above to estimate the weight of creative industries, although providing useful information on a diversity of practical procedures, have proved to be limited in assessing the importance of CIs in Portugal (cf. Table 1.2). The DCMS approach is too selective and particularly designed to describe the creative economy of the UK. The WIPO approach reveals higher objectivity in the criteria used, but the industry categories of commercialization and supporting services are too broad; moreover, copyright factors applied to Partial and Interdependent Copyright industries are difficult to assess and have effects on the results obtained. The Cultural Concentric 3 For instance, it was not possible to disaggregate Radio and Television production activities from their respective broadcasting services or to distinguish Photographic production activities from their related services. 21 Department for Culture, Media and Sport (DCMS) (1998), Creative Industries Mapping Document, London: DCMS. Department for Culture, Media and Sport (DCMS) (2001), Creative Industries Mapping Document 2001, London: DCMS. Department for Culture, Media and Sport (DCMS) (2010), Creative Industries Economic Estimates Technical Note - February 2010, London: DCMS, https://www.gov.uk/government/publications/creative-industries-economicestimates-february-2010 [accessed September 2014]. Department for Culture, Media and Sport (DCMS) (2011), Creative Industries Economic Estimates Full Statistical Release - 8 December 2011, London: UK DCMS, https://www.gov.uk/government/publications/creative-industrieseconomic-estimates-december-2011 [accessed September 2014]. Florida, R. (2002), The rise of the Creative Class - and How it’s Transforming Work, Leisure, Community and Everyday Life, New York: Basic Books. Florida, R. (2005a), The Flight of the Creative Class: The New Global Competition for Talent, New York: Harper Collins Publishers. Florida, R. (2005b), Cities and the Creative Class, New York: Routledge. Hall, P. (2000), “Creative cities and economic development”, Urban Studies, Vol. 37, Nº4, pp. 639-649. Handke, C. (2006), Surveying innovation in the Creative industries, Berlin: Humboldt University Berlin/ Rotterdam: Erasmus University Rotterdam. Heng, T., Choo, A., Ho, T. (2003), “Economic Contributions of Singapore’s Creative Industries”, Economic Survey of Singapore First Quarter 2003, Singapore Ministry of Trade and Industry. Hesmondhalgh, D. (2008), “Cultural and Creative Industries”, in Bennet, T. and Frow, J. (eds.), The Sage Handbook of Cultural Analysis, pp. 552-569, USA: Sage Publications. Higgs, P., Cunningham, S. (2008), “Creative Industries Mapping: Where have we come from and where are we going?”, Creative Industries Journal, Vol. 1, Nº1, pp. 730. 22 Higgs, P., Cunningham, S., Bakhshi, H. (2008), Beyond creative industries: Mapping the creative economy in the UK, London: NESTA. Jones, P., Comfort, D., Eastwood, I., Hillier, D. (2004), “Creative Industries: Economic contributions, management challenges and support Initiatives”, Management Research News, Vol. 27, Nº11, pp. 134-41. KEA, European Affairs (2006), The Economy of Culture in Europe, study prepared for the European Commission (Directorate-General for Education and Culture), Brussels, European Commission, http://www.keanet.eu/ ecoculture/studynew.pdf [accessed September 2014]. Landry, C. (2003), The creative city: A toolkit for urban innovators, London: Earthscan. Markusen, A., Wassall, G., DeNatale, D., Cohen, R. (2008), “Defining the Creative Economy: industry and occupational approaches”, Economic Development Quarterly, Vol. 22, Nº1, pp. 24-45. Oakley, K. (2004), “Not So Cool Britannia The Role of the Creative Industries in Economic development”, International Journal of Cultural Studies, Vol. 7, Nº1, pp. 67-77. Potts, J. (2009), “Art & innovation: An evolutionary economic view of the creative industries”, UNESCO Observatory e-journal – Multi-disciplinary research in the arts, http://web.education.unimelb.edu.au/UNESCO/pdfs/ejournals/artinnovation.pdf, [accessed September 2014]. Power, D. (2002), “"Cultural Industries" in Sweden: An assessment of their place in the Swedish economy”, Economic Geography, Vol. 78, Nº2, pp. 103-127. Pratt, A. (2004), “Mapping the cultural industries: Regionalization; the example of South East England”, in Power, D. and Scott, A. (eds), Cultural industries and the production of culture, pp. 16-36, New York: Routledge. Pratt, A., Norris, D., Marler, L. (2009), “Research frontiers for the creative class”, http://nciia.org/conf08/assets/pub/pratt.pdf [accessed September 2014]. Scott, A. (2003), The cultural economy of cities, CA, Thousand Oaks: Sage. 23 Scott, A. (2004), “Cultural-products industries and urban economic development: prospects for growth and market contestation in global context”, Urban Affairs Review, Vol. 39, Nº4, pp. 461-490. Scottish Government Social Research (2009), Creative industries, creative workers and the creative economy: a review of selected recent literature. Scotland: Scottish Government, http://www.scotland.gov.uk/Resource/Doc/289922/0088836.pdf [accessed September 2014]. Storper, M., Scott, A. (2009), “Rethinking human capital, creativity and urban growth”, Journal of Economic Geography, Vol. 9, pp. 147-167. Tchalakov, I., Borisova, V., Keskinova, D., Damyanov, G., Arkova, R., Andreeva, T., Kalchev, J., Todorov, T. (2007), “The Economic Contribution of Copyright-Based Industries in Bulgaria”, Brussels, European Commission, http://www.wipo.int/export/sites/www/copyright/en/performance/pdf/econ_contri bution_cr_bg.pdf [accessed September 2014]. Tepper, S. (2002), “Creative assets and the changing economy”, Journal of Arts Management, Law, and Society, Vol. 32, Nº2, pp. 159-168. Thorsby, D. (2008a), “Modelling the cultural industries”, International Journal of Cultural Policy, Vol. 14, Nº3, pp. 217-232. Thorsby, D. (2008b), “The concentric circles model of the cultural industries”, Cultural Trends, Vol. 17, Nº3, pp. 147-164. UNCTAD, United Nations Conference on Trade and Development (2004), Creative Industries and Development (document TD (XI)/BP/13). Geneva: United Nations, http://unctad.org/en/Docs/tdxibpd13_en.pdf [accessed September 2014]. UNCTAD, United Nations Conference on Trade and Development (2008), Creative Economy Report 2008: The challenge of assessing the creative economy towards informed policy-making. Geneva: United Nations, http://www.unctadxii.org/en/Media/News-Archive/The-Creative-EconomyReport-the-Challenge-of-Assessing-the-Creative-Economy---Towards-InformedPolicymaking/ [accessed September 2014]. 24 World Intellectual Property Organization (2003), Guide on Surveying the Economic Contribution of the Copyright Based Industries, WIPO publication No 893(E), Geneva. World Intellectual Property Organization (WIPO) (2005), “The Economic Contribution of Copyright-Based Industries in Hungary – the 2005 Report”, http://www.wipo.int/export/sites/www/copyright/en/performance/pdf/econ_contri bution_cr_hu.pdf [accessed September 2014]. World Intellectual Property Organization (WIPO) (2008), “The Economic Contribution of Copyright-Based Industries in Romania – 2008”, http://www.wipo.int/export/sites/www/copyright/en/performance/pdf/econ_contri bution_cr_ro.pdf [accessed September 2014]. 25 Annex 1 Table A1. 1: Mapping the DCMS industry-based model of Core Creative Industries, using ISIC and the correspondent Portuguese industry/ SIC codes Core Creative Sectors UK 2003 SIC codes Proportion of code taken ISIC codes – Rev. 4 Portuguese SIC codes - CAE - Rev. 3 - 5 digits 1. Advertising 74.40 100% 7310; 7320 73110; 73120; 73200 2. Architecture 74.20 25% 7110; 7120 71110; 71120; 71200 3. Arts and Antiques 52.48/ 9; 52.50 5% 4774; 4791 47790 4. Crafts* "Majority of businesses too small to be picked up in business surveys" (source: DCMS, 2010: 2) - - 23411; 23412; 23413; 23414; 23110; 23120; 23131; 23132; 23140; 23190; 16291; 16292; 32110; 32121; 32122; 32123; 32130 5. Design "No codes match this sector" (source: DCMS, 2010: 2) - - 74100 6. Designer Fashion 17.71; 17.72; 18.10; 18.22; 18.23; 18.24; 18.30; 19.30 0.5% 1410; 1420; 1430; 1512; 1520; 14110; 14120; 14131; 14132; 14133; 14140; 14190; 14200; 14310; 14390; 15120; 15201; 15202 74.87 2.5% 7410 74100 7. Video, Film & Photography 92.11; 92.12; 92.13; 22.32 74.81 100% 100% 25% 25% 5911; 5912; 5913 5914 1820 7420 59110; 59120; 59130 59140 18200 74200 9&10. Music and the Visual & Performing Arts 22.14; 22.31 92.31; 92.32; 92.34 92.72 25% 100% 25% 5920 9000; 7990 9321; 9329 59200 90010; 90020; 90030; 90040; 79900 9321; 9329 11. Publishing 22.11; 22.12; 22.13 22.15 92.40 100% 50% 100% 5811; 5813 5819 6391; 6399 58110; 58130; 58140 58190 63910; 63990 8&12. Software, Computer Games & Electronic Publishing 22.33 72.21 72.22 25% 100% 100% 1820 5820 6201; 6202 18200 58210; 58290 62010; 62020 13. Radio & TV 92.20 100% 6010; 6020 60100; 60200 Notes: The selection of codes and proportions taken was based on DCMS (2010) “Creative Industries Economic Estimates - 10 February 2010", available online at: https://www.gov.uk/government/publications/creative-industries-economic-estimatesfebruary-2010 [accessed September 2014]. Correspondence Tables between ISIC - Rev. 3.1 and ISIC - Rev. 4 available at: http://unstats.un.org/unsd/cr/registry/regso.asp?Ci=60 [accessed September 2014]. The Portuguese nomenclature CAE Rev. 3 (Classificação das Actividades Económicas) has direct correspondence with ISIC - Rev. 4. Codes eventually repeated in the mapping only were considered once in the estimations. * It was considered here some industry codes of traditional manufacturing related with Crafts activities (ceramics; pottery; hand-painting decoration; glass; woodcrafts; jewelry); the proportion used was 5%. 26 Table A1. 2: Mapping the WIPO Copyright Model, using ISIC and the correspondent Portuguese industry/ SIC codes - CORE Copyright Industries Sector Description ISIC codes – Rev. 3.1 Copyright factor associated Portuguese SIC codes - CAE - Rev. 3 - 5 digits Literature Press - Authors, writers, translators - Artistic and literary creation - Newspapers, magazines/ periodicals, books publishing - News agencies - Pre-press, printing, and post-press of books, magazines, newspapers - Wholesale and retail of press and literature - Libraries/ Archives 9214; 7499; 2212; 9220; 2221; 2219; 2222; 5139; 5239; 9231 100% 90030; 74300; 58110/20/30/40/90; 18110/20/30/40; 63910; 63990; 46492; 47610; 47620; 91011; 91012 Music, Theatrical Productions Operas - Composers, lyricists, choreographers, directors - Artistic and literary creation - Publishing of music - Manufacturing of recorded music - Wholesale and retail of recorded music (sale and rental) - Agents/ ticket agencies 9214; 9219; 9249; 2213; 2230; 5233; 7130; 5139; 9214; 7414; 9214 100% 90010; 90020; 90030; 90040; 59200; 18200; 46430; 47630; 77220; 74900; 79900 Motion Picture and Video - Directors, actors - Motion picture and video production and distribution - Motion picture exhibition - Video rentals and sales, video on demand - Ancillary services 9214; 9211; 9212; 7130; 2230 100% 59110/20/30/40; 77220; 18200 Radio and Television - Radio and television broadcasting - Independent producers (not related with broadcasting) - Cable Television (systems and channels) - Satellite television - Ancillary services 9213; 7499; 6420; 9213 100% 60100; 60200; 61100; 61300 Photography - Photographic Activities (Studios and commercial photography) - Photo Agencies and Libraries 7494; 2222; 7499; 9231 100% 74200; 74900; Software and Databases - Programming, development and design, manufacturing - Wholesale and retail pre-packaged software (business programs, video games, educational programs, etc.) - Database processing and publishing 7221; 7229; 5151; 5239; 7240; 7230 100% 58210; 58290; 62010/20/30/90; 46510; 47410; 63110; 63120 Visual and Graphic Arts - Artists - Art galleries and other wholesale and retail - Picture framing and other allied services - Graphic design 9214; 7494; 9214; 7499 100% 90030; 47784; 74100 Advertising - Agencies, buying services 7430; 7413 100% 73110; 73120; 73200 Copyright Collecting Societies - Activities of professional organizations 9112 100% 94120 Source: World Intellectual Property Organization (WIPO) (2003), “Guide on Surveying the Economic Contribution of the Copyright-Based Industries”. Notes: The selection of codes and the copyright factor associated was based on the WIPO (2003). Correspondence Tables between ISIC - Rev. 3.1 and ISIC - Rev. 4 available at: http://unstats.un.org/unsd/cr/registry/regso.asp?Ci=60 [accessed September 2014]. The Portuguese nomenclature CAE Rev.3 (Classificação das Actividades Económicas) is compatible with ISIC - Rev. 4. Codes eventually repeated in the mapping only were considered once in the estimations. 27 Table A1.2 (cont.): Mapping the WIPO Copyright Model, using ISIC and the correspondent Portuguese industry/ SIC codes - INTERDEPENDENT Copyright Industries Sector Description ISIC codes – Rev. 3.1 Copyright factor associated * Portuguese SIC codes - CAE - Rev. 3 - 5 digits TV sets, Radios, VCRs, CD players, DVD players, Cassette players, Electronic Game Equipment, and related Manufacture of television and radio receivers, sound or video recording; Wholesale; Retail Sale; Renting of personal and household appliances 3230; 5139; 5233; 7130 35% 26400; 46430; 47430; 77220 Computers and Equipment Manufacture of office/ accounting/ computing machinery; Wholesale of computers/ computer peripheral equipment/software; Renting of office machinery/ equipment (including computers) 3000; 5151; 7123 35% 26200; 46510; 47410; 77330 Musical Instruments Manufacture of musical instruments; Wholesale; Retail Sale of household goods, articles and equipment 3692; 5139; 5233 20% 32200; 46494; 47593 Photographic and Cinematographic Equipment Manufacture of optical instruments and photographic equipment; Wholesale; Retail Sale; Renting of other machinery and equipment 3320; 5139; 5239; 7129 30% 26702; 27400; 47782; 77390 Photocopiers Manufacture of office, accounting and computing machinery; Wholesale; Retail Sale other machinery and equipment 3000; 5159 30% 28230; 46660; 47781; 77330 Blank Recording Material Manufacture of other chemical products; Wholesale of electronic and telecommunications parts and equipment; Retail sale of household appliances, articles and equipment 2429; 5152; 5233 25% 20594; 26800; 46520; 47630 Paper Manufacture of pulp, paper and paperboard; Wholesale of other intermediate products, waste and scrap; Other retail sale in specialized stores 2101; 2109; 5149; 5239 25% 17110; 17120; 17230; 46762; 47620 Notes: *Copyright factors not available for Portugal. The information was based on the WIPO (2003) guide, on the WIPO (2005, 2008) reports, on the report by Tchalakov et al. (2007) and on Chow and Leo (2005). Codes eventually repeated in the mapping only were considered once in the estimations. 28 Table A1.2 (cont.): Mapping the WIPO Copyright Model, using ISIC and the correspondent Portuguese industry/ SIC codes - PARTIAL Copyright Industries Sector Description ISIC codes – Rev. 3.1 Copyri ght factor associa ted* Portuguese SIC codes - CAE - Rev. 3 – 5 digits Apparel, textiles and footwear Manufacture of wearing apparel; Manufacture of made-up textile articles; Manufacture of footwear; Wholesale of textiles, clothing and footwear; Retail sale of textiles, clothing, footwear and leather goods 1810; 1721; 5131; 5232; 1920; 5131; 5232 0,5% 14110; 14120; 14131/2/3; 14140; 14190; 14200; 14310; 14390; 15120; 13910; 13920; 13961; 13962; 13991; 13992; 13993; 46410; 47510; 15201; 15202; 46421/2; 47711/2; 47721/2 Jewellery and coins Manufacture of jewellery and related goods; Wholesale of other household goods; Other retail sale in specialized stores 3691; 5139; 5239 25% 32110; 32121; 32122; 32130; 46480; 46494; 47770; 47784 Other Crafts Activities of other membership organizations; Other retail sale in specialized stores 9199; 5239 40% 94991; 47784 Furniture Manufacture of furniture; Wholesale of other household goods; Retail sale; Renting of personal and household goods 3610; 5139; 7130 5% 31010/20/30; 31091; 31092; 31093; 31094; 46470; 46650; 47591 Household goods, China and Glass Manufacture of glass and glass products; Manufacture of knitted and crocheted fabrics and articles; Manufacture of other products of wood; Manufacture of other fabricated metal products; Wholesale of other household goods; Retail sale of household 2610; 173; 2029; 2899; 5139; 5233 0,5% 23110; 23120; 23131; 23132; 23140; 23190; 13920; 46410; 47510; 16291; 16292; 46494; 47593; 23411; 23412; 23413; 23414; 46441; 47592; 25710; 25991; 22292; 27510; 46430; 47540 Wall coverings and Carpets Manufacture of carpets and rugs; Manufacture of other articles of paper and paperboard; Other retail sale in specialized stores 1722; 2109; 5239 2% 13930; 17240; 46470; 46732; 47530 Toys and Games Manufacture of games and toys; Wholesale of other household goods; Other retail sale in specialized stores 3694; 5139; 5239 40% 32400; 46493; 47650 Architecture, Engineering, Surveying Architectural and engineering activities and related technical consultancy 7421 10% 71110; 71120; 71200 Interior Design Other business activities 7499 - Already considered in Core Copyright Industries. The Portuguese industry code 74100 - Design activities cannot be disaggregated into more detail. Museums Museums activities and preservation of historical sites and buildings 9232 50% 91020; 91030 Notes: * Copyright factors not available for Portugal. The information was based on the WIPO (2003) guide, on the WIPO (2005, 2008) reports, on the report by Tchalakov et al. (2007) and on Chow and Leo (2005). Codes eventually repeated in the mapping only were considered once in the estimations. 29 Table A1. 3: Mapping the Cultural Concentric Circles Model, using ISIC and the correspondent Portuguese industry/ SIC codes - CORE Creative Centre – Cultural Fine Arts Sector Description ISIC codes – Rev. 3.1 Portuguese SIC codes - CAE - Rev. 3 - 5 digits Visual Arts [crafts, painting, sculpture, photography] Crafts: “ranges in most categories in manufacturing and retail” (KEA, 2006: 309) Paintings and Sculpture: Artistic and literary creation and interpretation; Operation of arts facilities and museums; Other business activities; Exhibition halls; Other retail sale in non-specialized stores; Other retail sale in specialized stores Photography: Photographic Activities No industry codes describing Crafts 9214; 7499; 7010; 5219; 5239 7494; 9220 *47784; 23411; 23412; 23413; 23414; 23110; 23120; 23131; 23132; 23140; 23190; 16291; 16292; 32110; 32121; 32122; 32123; 32130 90030; 94120; 90040; 91020; 91030; 74900; 47784 74200; 63910; 63990 Performing Arts (including festivals) Theatre Artistic and literary creation and interpretation Dance Artistic and literary creation and interpretation Circus Other entertainment activities 9214 9214 9219 90010; 90020 Heritage Museums and Libraries Arts & Antiques Market Museums activities and preservation of historical sites and buildings Library and Archives activities Arts and Antiques Market 9232 9231 5240 91020; 91030 91011; 91012 47790 Source: KEA European Affairs (2006). Notes: * Here, it was considered some industry codes of traditional manufacturing related with Crafts activities (ceramics; pottery; hand-painting decoration; glass; woodcrafts; jewelry); the proportion used was 5%. Codes eventually repeated in the mapping only were considered once in the estimations. 30 Table A1.3 (cont.): Mapping the Cultural Concentric Circles Model, using ISIC and the correspondent Portuguese industry/ SIC codes Layer 1 - WIDER Core Cultural Industries Sector Description ISIC codes – Rev. 3.1 Portuguese SIC codes - CAE - Rev. 3 - 5 digits Film and Video Production of films/ videos (including commercials, activities of studios); Distribution of videos and DVDs; Reproduction of recorded media; Exhibition/ Projection of movies; Wholesale of video tapes and DVDs; Retail sale of video tapes and DVDs; Video Sale through rental of videos and DVDs 9211; 2230; 9212; 5139; 5233; 7130 59110; 59120; 59130; 18200; 59140; 46430; 47630; 77220 Television and Radio National radio and television broadcasting companies; Other radio and television broadcasters; Independent producers (not related with the broadcasting); Cable Television (systems and channels); Satellite Television 9213; 7499; 6420 60100; 60200; 61100; 61300 Software Publishing including Games Development, production, supply and documentation of ready-made (noncustomized) software, including games 7221 58210; 58290; 62010 Music Artistic and literary creation and interpretation; Printing and publishing of music; Production/manufacturing of recorded music; Wholesale and retail of recorded music (sale and rental) 9214; 2213; 2230; 5139; 5233; 7130 90030; 59200; 18200; 46430; 47630; 77220 Literature and Press Book publishing; Newspapers publishing; Magazines/periodicals; Wholesale and retail sale of press and literature (book stores, newsstands, etc.); Retail sale via mail order houses/ Internet 2211; 2212; 5139; 5239; 5251; 7240 58110; 58120; 58130; 58140; 58190; 46492; 47610; 47620; 47910 Source: KEA European Affairs (2006). Notes: Codes eventually repeated in the mapping only were considered once in the estimations. 37 Assessing the magnitude of Creative employment: A comprehensive mapping and estimation of existing methodologies * Abstract The present study surveys and maps the existing methodological approaches for measuring the creative employment. Based on a unique matched employer-employee dataset which encompasses over 3 million Portuguese workers, we found that the magnitude of the creative class varies considerably between approaches, ranging from 2.5%, using the conventional industry-based taxonomy and 30.8%, using Florida’s occupational proposal. The disparities are justified on the basis of the departure definition of what creative employment is and from operationalization issues regarding which industries and occupations to be included. Interestingly, when we focus on ‘core’ creative employment, the figures conveyed by the distinct approaches are strikingly similar (around 6%) suggesting that, at least in what core creative employment is concerned, the distinct approaches converge. The diversity of approaches and measurements are not necessarily a bad thing in itself, but has to be adequately acknowledged in order to accomplish adequate public policy guidance. Keywords: Creative employment; Occupations; Industries; Measurement; Portugal JEL-codes: L80; C81 * Published in the European Planning Studies (http://dx.doi.org/10.1080/09654313.2013.822475), Taylor & Francis, August 2013. 38 “If there is no agreement on how to define and measure the creative class, there is little prospect that it will provide useful public policy guidance. If no one knows how the creative class is constituted … there are likely to be no effective policy levers.” (Sands and Reese, 2008: 6) 1. Introduction The literature on the creative class and industries is relatively recent and consists of an array of publications which range from theoretical and policy-based articles (Pratt et al., 2009; Heinze and Hoose, 2013) to empirical studies on the estimation of creative employment in national and regional economies (Florida et al., 2008; Asheim and Hansen, 2009; Mellander et al., 2010). Since Florida’s (2002) seminal contribution, several studies and government reports have been published world-wide on the analysis of creative workers, their dimension (KEA European Affairs, 2006; Cunningham and Higgs, 2009), spatial, sector and knowledge-based distribution (Gabe, 2006; Clifton, 2008; Mellander, 2009), the determinants of their location preferences (Hansen and Niedomysl, 2009), and their effect on economic growth (Florida et al., 2008). Despite the reasonable amount of literature produced on the topic, several challenges remain for anyone undertaking empirical and quantitative analyses of creative activities (Çetindamar and Günsel, 2012; Lazzeretti et al., 2012; Lysgård, 2012). Fuzzy and allembracing definitions of which occupations should be included in the creative class (McGranahan and Wojan, 2007; Markusen et al., 2008), lack of objectivity in the criteria to select who is creative or not (Boschma and Fritsch, 2009; Clark, 2009), limitations of data used, and problems of highly aggregated occupational code categories (Higgs et al., 2008) seem to jeopardize an accurate analysis. Hornidge (2011) suggests that it is useful to frame ‘creative industries’ as a boundary concept, defined by the different actors who use it in varying ways, and underlines that a common identity and common structure uniting these different definitions are still in the process of being constructed. Intrinsically a theoretical construct, ‘creative industries’ must be operationalized before it can be used to direct and evaluate local policies (Reese et al., 2010). However, the diversity of methodological proposals for estimating the creative employment and the 39 use of distinct datasets tend to hamper a rigours analysis and account of the magnitude of that creative employment. Based on distinct datasets, existing estimates of the weight of creative industry/ class range from a meagre 2.1%, for the UK, in 2008 (Clark, 2009), to a stunning 52.4%, for the Netherlands, in 2001 (Clifton and Cooke, 2009). We aim at assessing the magnitude of creative employment by estimating its weight for all relevant existing methodologies - conventional approaches (DCMS traditional industry-based); occupation-based approaches (Florida’s approach, occupation-based approaches following Florida’s and the refinements of Florida’s taxonomy); and the combined industry and occupation-based approaches (the creative trident approach and the 2010 DCMS methodology) - using a unique and comparable micro dataset (including over 3 million workers) from the official employment datasets of Portugal (reference year: 2009). This allows a comparable quantification and properly discussion of the distinct figures provided by each methodology. In the next sections (2 and 3), we describe and map the most relevant existing measurement approaches. In Section 4 we estimate the size of the creative employment according to each approach, using Portugal data for the year 2009 as the reference case. Finally, in Section 5, we outline the most relevant contributions and policy implications of the present study. 2. Measuring the creative employment: a review of the main methodologies The empirical literature on the measurement analysis of creative employment can be distinguished into two main conceptual perspectives: one, more economic and industrial-based, centred on (creative) industries and the other, more sociologicaldriven, based on (creative) people. The first conventional measures employed in empirical studies on the creative economy have been developed with an industrial perspective, based on the conception of creativity as a productive process which generates wealth by the exploitation of intellectual property rights. In parallel with the industrial methodologies, a sociological occupational perspective on the creative employment – focused on what people do and their professional occupation – has emerged, associated with the concept of ‘creative class’ (see Florida, 2002, 2004). 40 The following sections describe the approaches developed within each perspective (Sections 2.1 and 2.2), as well as the approaches which combine industries and occupations (Section 2.3). 2.1. The industrial perspective: conventional industry-based approaches In terms of measurement, these approaches make use of the Standard Industrial Classification (SIC) system in order to estimate the size of creative industries. Here, creative employment is computed “by allocating all jobs in earmarked creative establishments - actual physical locations of production and service - into nested industries defined by major product” (Markusen et al., 2008: 29), and summing up of all the workers in all the creative industries. This first generation of methods emerged with the UK Creative Industries Mapping Document (DCMS, 1998, 2001), focused on capturing empirical information about specialized industries in each sector of the creative economy, for governmental purposes. The creative employment, in this case, is simply measured by the existing employment in each ‘creative’ sector, considering both direct and indirect/ support activities in the process (DCMS, 1998, 2001). Despite the relevance of the approach, drawbacks in delimiting the creative sectors led to difficulties in the measurement of creative activities, restricting the potential dimension of these industries. Indeed, the industry-based approach has been criticised by several authors (e.g., Pratt, 2004; Markusen et al., 2008; Granger and Hamilton, 2010). It has been stated that the results provided lead to an underestimation of creative employment, since they include the total number of employees working within those considered as creative industries, but overlook the creative employment outside those industries. Besides, there are limitations of the SIC systems in use. Even the most recent SIC codes seem to be inadequate when it comes to capturing information on the creative industries. The SIC classification mostly relies on narrow coding which does not provide detailed information on each sector, even when codes are disaggregated at their maximum levels. This limits a refined analysis of each activity sector and does not provide a sufficient detail for an accurate treatment of creative activities, tending to mitigate or aggregate them into broad categories (Granger and Hamilton, 2010). 41 Moreover, creative processes are being developed across all the sectors of the economy, but SIC codes hardly capture those activities. This is particularly true for the Design and Digital Media sectors, which are often intertwined with other activity sectors, some of them outside the creative core, such as the categories of product development, industrial design and fashion design, which mostly operate within the manufacturing sectors. This is also the case of Architecture, Crafts, Visual and Performing arts, whose activities often take place outside the creative core, within the manufacturing and services sectors. 2.2. The sociological perspective: occupational-based approaches Here, the Standard Occupational Classification (SOC) codes are used for the empirical estimation of creative employment, “[which] is divided into nested occupational groups based on skill content and work process”, giving particular emphasis to what “workers do rather than what they make” (Markusen et al., 2008: 29). This line of research went beyond the industrial approaches by focusing on occupations instead of the aggregate employment of specialized industry sectors (Higgs and Cunningham, 2007). Unlike industry-based methodologies (e.g., DCMS, 1998, 2001), mostly centred on a restricted number of creative industries, occupational approaches broadened the dimension of creative employment by accounting for the occupations considered as creative in all the economic activities. This type of measurement methodology allows for a detailed analysis of the creative workforce and the occupational structure over time, across regions and countries. For instance, Gabe (2006) used a shift-share model to study the evolution of creative workforce in urban areas of the United States (US), between 1990 and 2000, whereas McGranahan and Wojan (2007) developed a detailed analysis of creative categories in order to assess the occupational structure of US nonmetropolitan counties (cf. Table 2.1). One frequent drawback pointed to occupational-based approaches is that activities considered as creative are often associated with those involving higher educational levels (Markusen et al., 2008) to the detriment of others (e.g., craft work) that are also creative but associated with less formal education. In particular, as stressed by Glaeser (2005), by using census occupational data and grouping creative workers into high skilled categories, Florida’s (2002) criteria led to biases in the measurement of creative 42 occupations. It was further uncovered that each occupational category code covered a diversity of detailed professions with their categorization as creative involving a high degree of arbitrariness (McGranahan and Wojan, 2007). On this issue, McGranahan and Wojan (2007) proposed a refinement of Florida’s occupational groups based on a ranking of the creativity required by each given activity. This procedure conferred greater objectivity on the scrutiny of creative occupations, producing more robust estimations of creative employment than Florida’s (2002) study. Occupational approaches also overlook or neglect self-employed workers; since official source data mostly contain information on firms employing creative workers, they do not account for the self-employed, while their contribution to the creative economy appears to be significant (Van Steen and Pellenbarg, 2012). This problem is particularly relevant in the case of bohemians, for whom freelance works represent a significant part of their activity (Fritsch and Stuetzer, 2009). Finally, occupational-based approaches fail to permit the discrimination between the type of industries where creative workers operate and their industrial affiliation, since here, SIC codes are not taken into account. 2.3. The combined industryand occupation-based approaches Limitations of the two above approaches called for the development of a methodology making a combined use of the Standard Industrial and Occupational Classification (SIC/SOC) codes. The type of information gathered in this combined approach provides data on industries where creative workers are operating, and allows the identification of creative individuals working in non-creative sectors of activity, as well as of noncreative/ support labour existing in creative industries. Higgs et al. (2008) proposed the ‘creative trident’ approach to map the creative economy, employing both industry and occupational codes (see Table 2.1). More recently, studies drawing on the DCMS industry-based approach (DCMS, 2006, 2010a,b) have enlarged their analysis of core creative sectors by using both industry and occupational codes. According to these studies and some other authors (e.g., Barbour and Markusen, 2007), combined industry and occupational-based approaches provide a richer account of the occupational distribution within industries. Table 2. 1: Creative employment - a synthesis of empirical results in literature Methodological Approach Characteristics Author(s) | Study Methodology Followed Empirical results - Relative weight of creative employment in total workforce :: INDUSTRIAL PERSPECTIVE:: Conventional, industrybased approach Under these approaches, mostly drawn from the DCMS framework, estimates of creative employment are restricted to Core specialized creative sectors. This leads to more modest estimations of creative employment than found with other approaches, particularly those following Florida’s (2002, 2004) definition. [Use of SIC codes] DCMS (1998, 2001), UK Creative Industries Mapping Document Creative employment is measured by the total employment in each of the thirteen core creative sectors, considering both direct and indirect or supporting/non-creative activities in the process. UK (1998): 5% DCMS (2001), Creative Industries Mapping Document 2001 Creative employment is measured using a method closely following the DCMS (2001) framework - industry-based approach UK (2001): 7% Boix et al. (2010), “The geography of creative industries in Europe: Comparing France, Great Britain, Italy and Spain” France: 4.5%; Great Britain: 5.7%; Italy: 5.6%; Spain: 4.1% Curran and Van Egeraat (2010), “Defining and Valuing Dublin’s Creative Industries” Ireland (2006): 6.8% White (2010), “Creative industries in a rural region: Creative West. The creative sector in the Western Region of Ireland” Western Region of Ireland (2008): 3% :: SOCIOLOGICAL PERSPECTIVE:: Occupational-based approach Under these approaches, estimates of creative employment cover all the creative occupations across all the industry sectors of the economy. This leads to a much broader perspective of the creative class, particularly because it includes all the creative professionals, a vast category that is present in almost all activity sectors. [Use of SOC codes] Florida (2002), The rise of the Creative Class – and How it’s Transforming Work, Leisure, Community and Everyday Life Creative Employment is determined on the basis of Florida’s (2002, 2004) definition of creative class: Super Creative Core; Creative Professionals; and Bohemians (see Section 2). US (1999): 30.0%, of which: Super Creative Core: 11.7%; Creative Class: 18.3% Florida (2005), The Flight of the Creative Class: The New Global Competition for Talent BROAD definition (including technicians) (2002): UK: 33.8%; Germany: 40.2%; Norway: 41.6%; Denmark: 41.8%; Finland: 41.0%; Sweden: 42.4%; Netherlands: 47.0%; United States: 27.3%; Canada: 38.1% NARROW definition (excluding technicians)(2002): UK: 20.1%; Germany: 20.1%; Norway: 18.8%; Denmark: 213%; Finland: 24.7%; Sweden: 22.9%; Netherlands: 29.5%; United States: 23.6%; Canada: 25.0% Clifton (2008), “The ‘creative class’ in the UK: an initial analysis” Estimation of Creative Employment in England and Wales (2001), following Florida’s (2002, 2004) definition of creative class. England and Wales Total (2001): 37.3% Clifton and Cooke (2009), “Creative knowledge workers and location in Europe and North America: a comparative review” Estimation of Creative Employment in Europe, following Florida’s (2004) creative class concept, although considering a “small number of occupations” as creative professionals (Clifton and Cooke, 2009: 79). (2001): UK: 36.3%; Germany: 33.3%, Norway: 18.6%; Denmark: 27.6%; Finland: 33.4%; Sweden: 29.8%; The Netherlands: 52.4% 43 Boschma and Fritsch (2009), “Creative Class and Regional Growth: Empirical Evidence from Seven European Countries” Estimation of Creative Employment in 7 European countries (Denmark, England and Wales, Finland, Germany, the Netherlands, Norway, and Sweden), following Florida’s (2002, 2004) definition of creative class. 7 developed European countries (2002): 37.7%, of which: Creative Core: 26%, Creative Professionals: 70%; Bohemians: 4% 44 (…) Methodological Approach Characteristics Author(s) | Study Methodology Followed Empirical results - Relative weight of creative employment in total workforce Fritsch and Stuetzer (2009), “The geography of creative people in Germany” Estimation of Creative Employment in West Germany, following Florida’s (2002) definition of creative class. West Germany (2004): 36.8% Mellander (2009), “Creative and Knowledge Industries: An Occupational Distribution Approach” Estimation of Creative employment, by studying the occupational structure within industries (private sector) in Sweden, and following Florida’s (2002) definition of creative class. Sweden (2001): 36.8% Mellander et al. (2010), “Occupational and Industrial Distribution in Denmark: A comparison study with the United States, Canada and Sweden” Estimation of Creative employment, by studying the occupational structure within industries in Denmark, in comparison with the United States, Canada and Sweden, and closely following Florida’s (2002) definition of creative class. Denmark(2007): 39.5%; USA (2005): 35.1%; Canada (2006): 30.9%; Sweden (2005): 43% :: SOCIOLOGICAL PERSPECTIVE:: Occupational-based approach Refinements of Florida Here, refinements of Florida’s (2002) taxonomy are developed to restrict creative occupations to those that the authors believe are actually creative. [Use of SOC codes] Gabe (2006), “Growth of Creative Occupations in U.S. Metropolitan Areas: A Shift-Share Analysis” Recasting of Florida’s (2002) concept, restricting the analysis of creative employment to six categories: “management; computer and mathematical; architecture and engineering; life, physical, and social science; education, training, and library; and arts, design, entertainment, sports, and media occupations”. USA urban (1999): 18.1% McGranahan and Wojan (2007), “Recasting the Creative Class to Examine Growth Processes in Rural and Urban Counties” Recasting of Florida’s (2002) measure, by the exclusion of occupational categories from the summary groups of ‘Business’, ‘Educational’ and ‘Legal’ occupations and by excluding the whole summary category of ‘Healthcare’ occupations. Urban USA (2003): 30.9% Rural USA (2003): 19.4% :: COMBINED INDUSTRY and OCCUPATIONBASED APPROACHES [Mostly drawn upon the DCMS framework] Under these approaches, estimates of creative employment are calculated by all the occupations (creative occupations + noncreative/support occupations) in Core creative sectors (specialist and support mode) + All the creative occupations in non-creative sectors of activity (embedded creative employment) [Use of SIC and SOC codes] Higgs et al. (2008), Beyond creative industries: Mapping the creative economy in the UK (coord. Higgs, P., Cunningham, S. and Bakhshi, H.) - The selection of ‘core creative sectors’ is mostly drawn from the DCMS framework; - For creative employment, the authors develop the Creative Trident approach: CREATIVE employment = specialist and support creative occupations in the ‘specialized creative sectors’ - the Core Creative industries, or those dedicated to the ‘pre-creation’ and ‘creation’ stages of the process + All the creative occupations in non-creative sectors of activity (embedded creative employment), namely, in sectors such as ‘manufacturing’ , ‘real estate’, ‘business activities’, ‘wholesale and retail trade’, and ‘financial intermediation’. UK (2001): 7.1% 45 (…) Methodological Approach Characteristics Author(s) | Study Methodology Followed Empirical results - Relative weight of creative employment in total workforce Clark (2009), “Crunching creativity: an attempt to measure creative employment” Use of original DCMS framework with 2003 SIC codes (less specified industry categories). UK (2008): 5.5% Use of a SIC SOC matrix with UK 2007 SIC codes formulation, which provide a more detailed specification of each industry’s grouping category. UK (2008): 2.1% DCMS (2010a), Creative Industries Economic Estimates (Experimental Statistics) - December 2010 DCMS framework combined with occupational data based on SOC system - Use of combined industry and occupational approach to measure the creative employment in the industry sectors of the UK. Creative employment is measured by: “Employment in the Creative Industries” + “Employment in creative occupations in businesses outside the Creative Industries” (DCMS, 2010a) UK (2010): 7.8% 46 Industries’ employment structures diverge significantly from region to region and changes in regional labour structures and in the economic dynamics of industries may gain from a combined industry and occupational approach, for a better interpretation of occupational mobility across sectors over time (Barbour and Markusen, 2007; Currid and Stolarick, 2010). In this vein, such an approach is useful for regional policy implementation and management. Despite the advantages of using these approaches, they are not free from limitations. Restrictions of source information and of nomenclatures in use, such as highly aggregated data particularly on industries, long time intervals between each data upgrading process, limited knowledge on the self-employment, as well as difficulties in matching SIC with SOC codes and in capturing the creative component, are some of the major shortcomings reported by authors using combined industry and occupationalbased approaches (Higgs and Cunningham, 2007; Higgs et al., 2008). Summing up, extant empirical studies on the measurement of creative employment show that the methodologies based in the industrial perspective, such as the DCMS traditional approach, generally lead to more restricted figures of the creative employment, as they only consider the number of workers in the core of creative industries. In contrast, the sociological perspective, including the occupational-based approaches of Florida and those following Florida’s taxonomy, produces broadened results since they envisage the ‘creative class’ as a wide group of professional categories considered as creative, regardless of the economic activity sector. The empirical studies based on combined industry and occupational-based approaches evidence larger figures than those based on the industry perspective, as they also take into account the creative employment in the non-creative activity sectors, but inferior to that obtained by exclusively occupational-based approaches (see Table 2.1 and Figure 2.1). Figure 2. 1: The boundaries of the creative employment according to the main measurement perspectives Conventional perspective – Industry approach Combined Industry – occupational approaches Sociological perspective - Occupational approaches 53 We mapped this refinement approach of McGranahan and Wojan (2007) by excluding all those that were regarded by the authors as less creative occupations in the summary categories fully accounted by Florida (2002). Hence, in ‘Management occupations’, we removed all the occupations related to ‘farmers and farm managers’ (see Table A2.2, in Annex 2). From ‘Healthcare practitioners and technical occupations’, all the categories were excluded. In ‘Education, training, and library occupations’, only ‘post-secondary teachers’ and ‘librarians, curators and archivists’ were included. In ‘Business and financial operations’, only ‘accountants and auditors’ were considered. In ‘Legal occupations’, only ‘lawyers’ were included. From ‘Life, physical and social science occupations’, we excluded all the associated technicians. The summary category of ‘Computer and mathematical occupations’ was taken into account in full. The summary group of ‘Architecture and engineering occupations’ was also fully included in the recast measure. All the occupations related to ‘Arts, design, entertainment, sports, and media’ activities were wholly accounted. And finally, in ‘High-end Sales’, all the occupational codes related with ‘sales representatives’ and with the residual category of ‘other sales and related occupations, including supervisors’ were included. Since the code descriptions used by the authors on their recasting - US SOC 2000 - and the occupational nomenclatures that we used - ISCO-08 and CPP2010 - did not match exactly, the codes to be considered in our mapping were selected according to our interpretation of McGranahan and Wojan’s (2007) refinement criteria, based on the O*NET database of occupations. 15 By the same token, the descriptions of major category groups considered may differ slightly from those presented in McGranahan and Wojan (2007), but all the codes included properly describe the refined measure developed by these authors. 16 Another refining approach of Florida’s original proposal was developed by Gabe (2006), who focused on Florida’s ‘Super Creative Core’, adding up to this latter category all the management occupations. Thus, on mapping this approach we included all the detailed occupational codes which make up the summary categories of ‘Computer and mathematical occupations’, ‘Architecture and engineering occupations’, ‘Life, physical and social science occupations’, ‘Education, training and library 15 Available online at: http://www.onetonline.org/find/descriptor/browse [accessed September 2014]. 16 In this assessment, we undertook a detailed analysis on the categories that were recast by McGranahan and Wojan (2007: 201) and the structure of the US SOC 2000 codes of the U.S. Bureau of Labour Statistics, using the information available online at: http://www.bls.gov/soc/2000/socstruc.pdf [accessed September 2014]. 54 occupations’, ‘Arts, design, entertainment, sports and media occupations’, ‘Media and communication equipment workers’, and all ‘Management occupations’ (see Table A2.3, in Annex 2). All the categories excluding the latter (‘Management occupations) coincide with Florida’s (2002) ‘Super Creative Core’. Although relying upon more objective criteria in the selection of creative occupations, based on the O*NET occupational database, given that they only suggest a recasting of the summary categories present in Florida’s definition, these refinement proposals continue to conflate human capital with creativity. The occupational groups considered in these proposals had already been subject to criticism (see Glaeser, 2005) and the authors did not go beyond those categories in their refinement approaches. Indeed, ‘Jewellers’, ‘hand sewers and seamstresses’, ‘fabric and apparel patternmakers’, ‘precious metal workers’, ‘painting, coating, and decorating workers’, ‘potters’, ‘prepress technicians’, and other skilled workers in a vast array of manufacturing sectors (e.g., printing sector, wood, glass, ceramics, furniture, textiles), including occupations that also require creative thinking, continue to be absent from these refinement proposals. 3.3. The combined industry and occupation-based approach 3.3.1. The creative trident The creative trident method, presented by Higgs et al. (2008), proposes to measure creative employment by taking into account three types of creative workers: i) ‘Specialist creative workers’, employed in the creative occupations operating in the creative industrial sectors; ii) ‘Support workers’, non-creative occupations engaged in support activities, such as management, administrative, technical, in the creative sectors; and iii) ‘Embedded creative workers’, comprising individuals in creative occupations in non-creative sectors. According to this methodology, the sum of these three types of employment, in the selected creative occupations and industry sectors, gives the total creative employment in the economy. This methodological proposal was mapped using the details provided by Higgs et al. (2008) in the technical Annex of their report. To achieve the best possible accuracy in this mapping, we used the most recent industry codes - CAE - Rev. 3 - at their maximum detail, compatible with the latest international ISIC - Rev. 4 codes, in order to describe all the industry sectors that best corresponded to the creative industries defined 55 by Higgs et al. (2008). To define the core creative sectors, Higgs et al. (2008: 27) took as a departure point the Frontier Economics (2007) framework and selected all those industries directly involved in “the pre-creation and creation stages of the value chain”, which they called the “creative core”. Although the creative trident approach differs from the recent industry and occupational-based approach of DCMS basically at the level of improvements included, the selected creative sectors were aligned “with the 13 sectors that make up the official DCMS measure of the creative industries” (Higgs et al., 2008: 19), which permits direct comparisons between these two approaches. The core creative sectors covered the following segments: ‘Advertising and Marketing’; ‘Architecture’, ‘Visual Arts and Design’; ‘Film, TV, Radio and Photography’; ‘Music and Performing Arts’; ‘Publishing’; and ‘Computer Software’ (cf. Table 2.4). 17 The set of creative occupations has been mapped as corresponding to all workers whose primary purpose was the engagement in creative functions and who were directly involved in the production and creation stages. In their definition, Higgs et al. (2008: 28) included: i) “those engaged in producing primary creative output - for example, writers, musicians, visual artists, film, television and video makers, sculptors and craftspeople”; ii) “those engaged in interpretive activity - for example, performers interpreting works of drama, dance, music, etc. in a wide variety of media from live performance to digital transmission via the Internet”; and iii) “those supplying creative services in support of artistic and cultural production - for example, book editors, lighting designers, music producers, etc.”. We mapped all the occupational codes according to the nomenclature UK SOC 2000, followed by Higgs et al. (2008: 60) in their technical Annex, and using the corresponding codes of the latest international ISCO-08 system and of Portuguese most recent occupational nomenclature CPP 2010 (cf. Table 2.4). During the mapping exercise, even though a suitable correspondence was found between the different industrial nomenclatures used, it was difficult to thoroughly describe the creative activities in some of the codes, particularly those related to all17 Higgs et al. (2008) excluded some industry sectors and some occupations considered by the DCMS industry and occupational-based approach as being creative. They also added other industries and professions to their definition of Creative Core that were not considered by the DCMS industry and occupational-based approach. For further details see Higgs et al. (2008: 27-30). 56 inclusive or residual categories such as ‘Other entertainment activities’ or ‘Recreational, cultural and sporting activities not otherwise specified’. Table 2. 4: Combined industryand occupational-based approach - the Creative Trident Creative Sectors UK 2003 SIC codes ISIC Rev. 4 codes Portuguese CAE - Rev 3 codes - 4 digits SOC2000 - occupational UK codes ISCO - 08 codes - 4 digits Portuguese Occupational Codes (Portuguese CPP 2010) - 4 digits 1. Advertising Advertising (744) 7310; 7320 7311; 7312; 7320 Advertising and public relations managers (1134); Marketing associate professionals (3543) 1221; 1222; 2431; 4227 1221; 1222; 2431; 4227 2. Visual Arts, Design and Architecture Manufacture of jewelry and related articles (362) 3211; 3212 3211; 3212; 3213 Artists (3411); Goldsmiths (5495); Hand craft occupations (5499); Glass and ceramics makers, decorators and finishers (5491); Furniture makers/ craft woodworkers (5492) 2651; 7311; 7313; 7314; 7315; 7316; 7317; 7521; 7522; 7318; 7319; 7531 2651; 7311; 7313; 7314; 7315; 7316; 7317; 7521; 7522; 7318; 7319; 7531 Design (no UK SIC code) 7410 7410 Graphic designers (3421); Product, clothing designers (3422) 2163; 2166; 3432 2163; 2166; 3432 Architecture (74201) 7110 7111 Architects (2431); Town planners (2432); Architectural technologists and town planning technicians (3121); Design and development engineers (2126); Draughts persons (3122) 2161; 2162; 2164; 2165; 3118 2161; 2162; 2164; 2165; 3118 3. Film, TV, Radio and Photography Motion Picture and Video activities (921); Radio and TV activities (922) 5911; 5912; 6010; 6020; 7420 5911; 5912; 6010; 6020; 7420 Arts officers, producers and directors (3416); Broadcasting associate professionals (3432); Photographers and audiovisual equipment operators (3434) 2654; 2656; 3521; 3435; 3431 2654; 2656; 3521; 3435; 3431 4. Music and the Performing Arts Recreational, cultural and sporting activities (920); Other entertainment activities (923) 5920; 9000; 9321; 9329 5920; 9001; 9002; 9003; 9004; 9321; 9329 Musicians (3415); Actors, entertainers (3413); Dancers and choreographers (3414) 2652; 2653; 2655 2652; 2653; 2655 5. Publishing Publishing (221); News agencies (924); Library, archives, museums and other cultural activities (925) 5811; 5812; 5813; 5819; 7490; 9101; 9102; 9103; 6391; 6399 5811; 5812; 5813; 5814; 5819; 7430; 9101; 9102; 9103; 9104; 6391; 6399 Authors, writers (3412); Journalists, newspaper and periodical editors (3431); Originators, compositors and print preparers (5421); Librarians (2451); Library assistants/clerks (4135); Archivists and curators (2452) 2641; 2642; 2643; 7321; 2621; 2622; 3433 2641; 2642; 2643; 7321; 2621; 2622; 3433 6. Computer Software 7220 Computer Software consultancy (‘72 Computer and related activities’) 6201; 6202; 6209 6201; 6202; 6203; 6209 Software professionals (2132); IT strategy and planning professionals (2131) 2511; 2512; 2513; 2514; 2519; 2521; 2522; 2523; 2529; 3511; 3512; 3513; 3514 2511; 2512; 2513; 2514; 2519; 2521; 2522; 2523; 2529; 3511; 3512; 3513; 3514 Note: The selection of codes is of the responsibility of the present paper’s authors, according to their interpretation of Higgs et al. (2008: 59-61) selection of industrial (UK SIC 2003) and occupational (UK SOC 2000) codes. The respective occupational codes were converted into the recent versions of ISCO-08 and the Portuguese CPP2010. 57 Estimations of this SIC-SOC approach were carried out by considering the whole proportion (100%) of employment in each industry and occupational code. The procedure for estimating the creative employment encompassed the inclusion of all ‘Specialist’ and ‘Support’ workers in each defined creative sector, plus the ‘Embedded creative workers’, i.e., those in the selected creative occupations, but operating in all the non-creative sectors of the economy. 3.3.2. The 2010 DCMS proposal In a similar way to the creative trident approach, besides the total employment in the selected creative industries, all the creative workers operating outside the defined core creative sectors are taken into account in the 2010 DCMS methodological proposal (DCMS, 2010a). The selection of creative sectors followed the original DCMS framework, which lists the following segments: ‘Advertising and Marketing’; ‘Architecture’; ‘Arts and Antiques’; ‘Crafts’; ‘Design’; ‘Designer Fashion’; ‘Video, Film, and Photography’; ‘Radio and TV’; ‘Music and the Visual and Performing Arts’; ‘Publishing’; and ‘Software and Electronic Publishing’ (Table 2.5). In this mapping, we use the latest international ISIC - Rev. 4 codes and the corresponding national industry codes CAE - Rev. 3 to describe all the industry sectors that best match the core creative industries defined by DCMS (2010a). According to DCMS (2010a), when industry sectors that were considered as creative also comprised non-creative activities only a portion of the code was accounted in the estimations. This was the case of ‘Photographic activities’, where only 25% of the code was considered, and the case of the vast number of manufacturing codes on ‘Textiles and apparel’, where a portion of only 0.5% was taken to describe Fashion Design activities. The proportion considered represents an attempt to extract the share of creative employment in those industry sectors. The industry code describing Design activities was, in accordance with DCMS (2010a), divided in three major segments: 4.5% of the code was included in the ‘Architecture’ segment, 89.7% was integrated in the ‘Design’ segment, and the remaining 5.8% was incorporated into ‘Designer Fashion’. This partition allowed for a better differentiation of the design activities and did not affect the overall result since the code as a whole is considered in the total calculation of the creative employment in all the creative 58 industries. Worthy of note is the ‘Crafts’ sector, where, according to DCMS (2010a), no industry codes were considered on the basis that the SIC system could hardly describe handicraft and craftwork activities. Here, using the SOC nomenclature, a set of creative occupations was defined as to extract the number of handicraft workers across the sectors of the economy (see Table 2.5). Then the estimation for the total employment in creative industries was given by the sum of all the workers operating in the defined creative sectors. In order to estimate the number of creative workers outside the core creative sectors, DCMS (2010a) presented a selection of creative occupations using the UK SOC 2000 codes that best fitted those professional activities, in each creative sector. On mapping these occupations, we used the latest international ISCO-08 codes and the corresponding national occupational codes of the CPP 2010. Following DCMS (2010a), in the cases of skilled workers operating in the manufacturing sectors, such as ‘labourers in building and woodworking trades’, a portion of 5% of the respective occupational codes was included in the estimations. This portion is intended to capture the share of creative workers inside those vast occupational categories. In the case of ‘Product, clothing and related designers’, a portion of 93.9% of the respective occupational codes was considered in the segment of ‘Design’ and the other 6.1% was included in ‘Designer fashion’. In the overall estimate of total creative employment, product and garment designers were fully accounted. The DCMS (2010a) approach has brought some necessary updates and adjustments to its original framework. By making use of occupational codes, this approach provided a broadened account of creative employment since it now takes into account the creative workers operating inside and outside the creative core industries. Moreover, it considers crafts occupations in the analysis and also presents a clearer differentiation between the creative sectors (e.g., Design vs. Designer Fashion) through the partition of industry and occupational codes. The estimation of creative employment through this approach, considering its details on codes, partitions and portions taken (which are somehow ad hoc and do not account for changes in the industrial and occupational structure), turned out to be anything but simple during the programming task for the extraction of data by code. 59 Table 2. 5: The 2010 DCMS proposal: combined Industry-Occupational approach Core Creative Sectors UK 2007 SIC codes Portion of SIC codes ISIC Rev.4 codes Portug uese CAE - Rev 3 codes SOC2000 - occupational UK codes ISCO - 08 codes - 4 digits Portuguese Occupational Codes (Portuguese CPP 2010) - 4 digits 1. Advertising and Marketing Advertising (73.11); Media Representation (73.12) 100% 7310; 7320 7311; 7312; 7320 Advertising and public relations managers (1134); Marketing associate professionals (3543); Public Relations Officers (3433) 1221; 1222; 2431; 2432; 4227 1221; 1222; 2431; 2432; 4227 2. Architecture Architectural activities (71.11); Design activities (74.10) 100% 4.5% 7110; 7410 7111; 7410 Architects (2431); Town planners (2432); Architectural technologists and town planning technicians (3121) 2161; 2162; 2164; 2165 2161; 2162; 2164; 2165 3. Arts and Antiques Retail sale in commercial art galleries (47.78/1); Retail sale of antiques including antique books, in stores (47.79/1); 100% 4774 47790 “No SOC codes match this sector” (DCMS, 2010a: 23). 4. Crafts “Majority of businesses too small to be picked up in business surveys” (DCMS, 2010a: 20). Floral arrangers/ florists (5496); Hand craft occupations n.e.c. (5499); Musical instrument makers and tuners (5494); Goldsmiths (5495); Glass and ceramics makers, decorators (5491); Glass and Ceramics process operatives (8112); Furniture makers, other craft woodworkers (5492); Laborers in Building and Woodworking trades (9121) (5% of SOC); Pattern makers (5493) 6113; 7311; 7312; 7313; 7314; 7315; 7316; 7317; 7521; 7522; 7523 (5% of SOC); 7318; 7319; 7531; 7532 6113; 7311; 7312; 7313; 7314; 7315; 7316; 7317; 7521; 7522; 7523 (5% of SOC); 7318; 7319; 7531; 7532 5. Design Design activities (74.10) 89.7% 7410 7410 Artists (3411); Product, Clothing and related designers (3422) (93.9% of SOC); Graphic designers (3421); Design and Development engineers (2126) 2651; 2163 (93.9% of SOC); 2166; 3432 2651; 2163 (93.9% of); 2166; 3432 6. Designer Fashion Clothing manufacturing UK SIC 2007 codes (14.11, 14.12, 14.13, 14.14, 14.19, 14.20, 14.31, 14.39, 15.12, 15.20) 0.5% 1410; 1420; 1430; 1512; 1520 1411; 1412; 1413; 1414; 1419; 1420; 1431; 1439; 1512; 1520 Product, Clothing and related designers (3422) (6.1% of SOC); Weavers and Knitters (5411) 2163 (6.1% of SOC); 7533 2163 (6.1% of SOC); 7533 74.10 5.8% 7410 7410 7. Video, Film and Photography Motion picture and video production activities (59.11; 59.12); Motion picture and video distribution activities (59.13); Motion picture projection activities (59.14) 100% 5911; 5912; 5913; 5914 5911; 5912; 5913; 5914 Photographers and audiovisual equipment operators (3434) 3431;352 1; 3435 3431; 3521; 3435 Photographic activities 25% 7420 7420 60 (74.20); Reproduction of video recording (18.20) 10% 1820 1820 13. TV and Radio Radio broadcasting (60.10); Television programming/ broadcasting activities (60.20) 100% 6010; 6020 6010; 6020 Broadcasting associate professionals (3432); TV, Video and Audio engineers (5244) 3522; 3521 3522; 3521 9&10. Music and the Visual & Performing Arts Sound recording and music publishing activities (59.20); 100% 5920 5920 Musicians (3415); Actors, entertainers (3413); Dancers and choreographers (3414); Authors, writers (3412); Arts officers, producers and directors (3416) 2652; 2655; 2653; 2641; 2654; 2656 2652; 2655; 2653; 2641; 2654; 2656 Reproduction of sound recording (18.20); 10% Performing arts (90.01); Support activities to performing arts (90.02); Artistic creation (90.03); Operation of arts facilities (90.04) 100% 9000 9001; 9002; 9003; 9004 11. Publishing Book Publishing (58.11); Publishing of newspapers (58.13); Publishing of journals and periodicals (58.14); Other publishing activities (58.19); News agency activities (63.91) 100% 5811; 5812; 5813; 5819; 7490; 6391; 6399 5811; 5812; 5813; 5814; 5819; 7430; 6391; 6399 Journalists, newspaper and periodical editors (3431); Originators, compositors and print preparers (5421); Printers (5422); Bookbinders and Print finishers (5423); Screen Printers (5424) 2642; 2643; 7321; 7322; 7323 2642; 2643; 7321; 7322; 7323 8&12. Software & Electronic Publishing 8&12. Digital & Entertainment Media Business and domestic software development (62.01/2); Computer consultancy activities (62.02); Other software publishing (58.29); Publishing of computer games (58.21); Readymade interactive leisure and entertainment software development (62.01/1) 100% 5820; 6201; 6202; 6209 5821; 5829; 6201; 6202; 6203; 6209 Information and Communication Technology managers (1136); IT strategy and planning professionals (2131) 2511; 2512; 2513; 2514; 2519; 2521; 2522; 2523; 2529; 1330 2511; 2512; 2513; 2514; 2519; 2521; 2522; 2523; 2529; 1330 Note: The selection of codes is of the responsibility of the present paper’s authors, according to their interpretation of DCMS (2010a: 18, 24) selection of industrial (UK SIC 2007) and occupational (UK SOC 2000) codes. DCMS (2010a) "Creative Industries Economic Estimates – December 2010 (Experimental statistics) - Full Statistical Release", available online at: https://www.gov.uk/government/publications/creative-industries-economic-estimates-december-2010-experimental-statistics [accessed September 2014]. Despite the challenges that the combination of data on industries and occupations brought to the mapping exercise and the respective estimations, this approach proposes a richer perspective of the creative employment by extending the analysis beyond the core creative sectors to include the creative employment existing across all the noncreative sectors of the economy. 61 4. Computing the magnitude of the creative employment according to the existing methodological approaches The data was extracted from Quadros de Pessoal, the Matched Employer-Employee Databases of the GEE/ ME 18 , Ministry of Economy of Portugal, for 2009 (the latest available at the time of this study). It covers all the employment in industries and establishments operating in the national territory with at least one employee. It excludes Public Administration and Domestic services and does not account for selfemployment. According to the latest information available (2009), the total employment in the private, structured sector was 3.128.126 workers. Before proceeding with the estimations, two points are worth mentioning regarding the exclusion from the analysis of self-employed and public servants (government employees who work in any of the departments of a state or territory government). Some studies report (e.g., Van Steen and Pellenbarg, 2012) that self-employment contributes significantly to creative employment, most notably in the most developed countries, as many of self-employed are freelance workers in sectors such as construction, consultancy, and culture, sports and recreation. In these latter countries, however, the share of self-employed workers in the total is much lower than in less developed countries. According to the OECD, in 2010, that share ranged from under 8% in the United States, and Norway to well over 30% in Greece, Mexico, and Turkey. In Portugal that figure was approximately 20%, with more than 80% of self-employed concentrated in the primary and tertiary sectors. 19 Noticeable, according to this data, there is a trend, since 1990, for a decrease in the share of self-employed workers in the generality of countries regardless of their development level. The exclusion of self-employment from the analysis is regrettable and important. However, as we are estimating the magnitude of the creative employment for one single country (Portugal), this exclusion does not substantially bias the analysis. That would not be the case if the analysis involved cross-country comparisons. 18 Courtesy of the GEE/ ME - Gabinete de Estratégia e Estudos of the Ministry of Economy of Portugal, October - December 2011. The GEE/ ME is not responsible for the results and interpretation contained in this study. These are of the authors’ full responsibility. 19 Data gathered from the OECD Fact Book 2011-2012: http://www.oecd-ilibrary.org/sites/factbook2011-en/07/01/04/index.html?itemId=/content/chapter/factbook-2011-61-en and from Eurofound 2009: http://www.eurofound.europa.eu/comparative/tn0801018s/pt0801019q.htm [accessed September 2014]. 62 Some bias has also to be acknowledged by the fact that we are excluding from the analysis public servants. Such exclusion is likely to substantially (and negatively) impact on the magnitude of creative employment, particularly when we use Florida’s original proposal, which encompasses a large amount of occupations (e.g., Legislators, Administration professionals, Health professionals, Regulatory government associate professionals), which in some countries, namely in Portugal, are performed within the public sector sphere. However, when we focus the analysis of the magnitude of the creative employment on the (super) creative core, this bias is negligible. All the estimated figures have been extracted using STATA 11® statistical analysis software. The stage at which we proceed to the estimates was also a challenge to this research work, given the limitations of the SOC system that was used to extract the data available for the year 2009 from the employment datasets. 20 The conversion of all the CPP2010 occupational codes into the previous version of CNP94 was based on the instructions in the official report by INE (2010: 460-474) on the Portuguese Classification of Occupations 2010. The codes and descriptions using the previous nomenclature - CNP94, at 6-digit level, were extracted, code by code, from the Statistics Portugal (INE) official website. The estimates of the Portuguese creative employment, using each approach described and mapped in Section 3, are summarized in Figure 2.2. Occupational approaches based purely on the analysis of occupational/ SOC categories and following Florida’s (2002) taxonomy, led to more inflated results than those obtained by using simple industry-based/ SIC or combined industry-occupational/ SICSOC approaches. Accordingly, the Portuguese creative employment ranges between 20 At the time the estimations were undertaken - from October to December 2011 - the nomenclature in use to extract 2009 data was still the previous version of occupational codes corresponding to the CNP94 (Classificação Nacional de Profissões - 1994). Besides facing the already known difficulties related to more obsolete classification systems - the lack of information/SOC codes on the different categories of Designers, or the unavailability of occupational codes which were non-existent or not relevant at the time of that previous revision (e.g., Graphic designer, Interior designer, Survey and market research interviewer) - this constraint also required the exhaustive and time-consuming task of converting all the CPP2010 codes at 5 digits that were used in the mapping into the previous CNP94 codes at the maximum detail level of 6 digits, in order to capture the most precise information possible. Indeed, in order to achieve the best correspondence possible between the latest occupational revision CPP2010 and the previous nomenclature for occupations CNP94, it was necessary to look into the detail of 6-digit codes, in every single case. 69 Glaeser, E. (2005), “Review of Richard Florida’s the rise of the creative class”, Regional Science and Urban Economics, Vol. 35, Nº5, pp. 593-596. Granger, R., Hamilton, C. (2010), “Re-spatializing the creative industries: A relational examination of underground scenes and professional and organizational lock-in”, Creative Industries Journal, Vol. 3, Nº1, pp. 47-60. Hansen, H., Niedomysl, T. (2009), “Migration of the creative class: Evidence from Sweden”, Journal of Economic Geography, Vol. 9, Nº2, pp. 191-206. Heinze, R., Hoose, F. (2013), “The creative economy: Vision or illusion in the structural change?”, European Planning Studies, Vol. 21, Nº4, pp. 516-535. Higgs, P., Cunningham, S. (2007), Australia’s Creative Economy: Mapping Methodologies, Brisbane: ARC Centre of Excellence for Creative Industries, Innovation (CCI), http://eprints.qut.edu.au/6228/ [accessed September 2014]. Higgs, P., Cunningham, S., Bakhshi, H. (2008), Beyond Creative Industries: Mapping the Creative Economy in the UK, London: NESTA. Hornidge, A. (2011), “Creative industries: Economic programme and boundary concept”, Journal of Southeast Asian Studies, Vol. 42, Nº 2, pp. 253-279. Instituto Nacional de Estatística (INE) (2010), Classificação Portuguesa das Profissões 2010 (Portuguese Classification of Occupations 2010), Lisbon: INE, http://www.ine.pt/xportal/xmain?xpid=INE&xpgid=ine_publicacoes&PUBLICACOESpu b_boui=107961853&PUBLICACOESmodo=2 [accessed September 2014]. KEA European Affairs (2006), The Economy of Culture in Europe, Study Prepared for the European Commission, Brussels: Directorate-General for Education and Culture. Lazzeretti, L., Capone, F., Boix, R. (2012), “Reasons for clustering of creative industries in Italy and Spain”, European Planning Studies, Vol. 20, Nº8, pp. 12431262. Lysgård, H. (2012), “Creativity, culture and urban Strategies: A fallacy in cultural urban strategies”, European Planning Studies, Vol. 20, Nº8, pp. 1281-1300. Markusen, A. (2006), “Urban development and the politics of a creative class: Evidence from the study of artists”, Environment and Planning A, Vol. 38, Nº10, pp. 19211940. 70 Markusen, A., Wassall, G., DeNatale, D., Cohen, R. (2008), “Defining the creative economy: Industry and occupational approaches”, Economic Development Quarterly, Vol. 22, Nº1, pp. 24-45. McGranahan, D., Wojan, T. (2007), “Recasting the creative class to examine growth processes in rural and urban countries”, Regional Studies, Vol. 41, Nº2, pp. 197216. Mellander, C. (2009), “Creative and knowledge industries: An occupational distribution approach”, Economic Development Quarterly, Vol. 23, Nº4, pp. 294-305. Mellander, C., Stolarick, K., King, K. (2010), “What you do, not who you work for: A comparison with the United States, Canada and Sweden”, Martin Prosperity Research Working Paper Series, September, http://martinprosperity.org/ 2010/10/04/occupational-and-industrial-distribution-in-denmark-a-comparisonwith-the-united-states-canada-and-sweden/ [accessed September 2014]. Pratt, A. (2004), “Mapping the cultural industries: Regionalization; the example of South East England”, in D. Power and A. Scott (eds.), Cultural Industries and the Production of Culture, pp. 16-36, New York: Routledge. Pratt, J., Norris, D., Marler, L. (2009), “Research frontiers for the creative class”, Louisiana Tech University, http://nciia.org/conf08/assets/pub/pratt.pdf [accessed September 2014]. Reese, L., Faist, J., Sands, G. (2010), “Measuring the creative class: Do we know it when we see it?”, Journal of Urban Affairs, Vol. 32, Nº3, pp. 345-366. Sands, G., Reese, L. (2008), “Cultivating the creative class: And what about Nanaimo?”, Economic Development Quarterly, Vol. 22, Nº1, pp. 8-23. Van Steen, P., Pellenbarg, P. (2012), “Creative class and self-employment in the Dutch labour market”, Tijdschrift voor Economische en Sociale Geografie, Vol. 103, Nº5, pp. 634-637. White, P. (2010), “Creative industries in a rural region. Creative West: The creative sector in the Western Region of Ireland”, Creative Industries Journal, Vol. 3, Nº1, pp. 79-88. 71 Annex 2 Table A2. 1: Taxonomy following Florida’s ‘Creative Class’ - Occupational categories Creative Class category groups Occupational Categories Descriptions Occupational ISCO-08 Codes (summary categories) / Portuguese Standard Occupational codes CPP - 2010 (summary categories)* Super Creative Core . Computer and mathematical occupations; . Architecture and engineering occupations; . Life, physical and social science occupations; . Education, training and library occupations; . Arts, design, entertainment, sports and media occupations . Computing professionals (25); . Mathematicians, Statisticians and related professionals (212); . Architects, Engineers and related professionals (214; 215; 216); . Life Science professionals (213); . Physicists, Chemists and related professionals (211); . Social Science and related professionals (263); . University and higher education teachers (231); . Vocational, technological and artistic education teachers (232); . Secondary and basic education teachers (233); . Primary school and early childhood teachers (234); . Other teaching professionals (235); . Archivists, museum curators and related information professionals (262) + Bohemians . Authors, journalists and linguists (264); . Creative and performing artists (265); . Product and garment designers (2163); . Graphic and multimedia designers (2166); . Musicians, singers and composers (2652); . Dancers and choreographers (2653); . Film, stage and related directors and producers (2654); . Actors (2655); . Announcers on radio, television and other media (2656); . Creative and performing artists not elsewhere classified (2659); . Advertising and marketing professionals (2431); . Public relations professionals (2432); . Artistic, Entertainment and Sports associate professionals (342; 343); . Telecommunications and broadcasting technicians (352); . Fashion and other models (5241). Creative Professionals . Management occupations; . Business and financial operations occupations; . Legal occupations; . Healthcare practitioners and technical occupations; . High-end sales and sales management; . Administrative associate professionals . Legislators, senior officials and managers (1); . Finance professionals (241); . Administration professionals (242); . Financial and mathematical associate professionals (331); . Sales and purchasing agents and brokers (332); . Business services agents (333); . Legal professionals (261); . Health professionals (except nursing) (221; 223; 224; 225; 226); . Nursing and midwifery professionals (222); . Nursing and midwifery associate professionals (322); . Life science technicians and related associate professionals (314); . Medical and pharmaceutical technicians and health associate professionals(321; 323; 324; 325); . Physical and engineering sciences technicians (311; 312; 313; 315); . Information and communications technology operations and user support technicians (351); . Regulatory government associate professionals (335); . Finance and sales associate professionals (2433; 2434); . Administrative, legal, social and specialized secretaries and related professionals (334; 3411; 3412) Sources: Adapted from Boschma and Fritsch (2009). The selection of codes is from the responsibility of this article’s authors as a result of their interpretation on the category groups and respective descriptions. Note: * The detailed mapping at a 5-digit level can be provided upon request to the authors. 72 Table A2. 2: Refinements of Florida’s proposal by McGranahan and Wojan (2007) Creative Class category groups Occupational Categories Descriptions Occupational ISCO-08 Codes (summary categories) / Portuguese Standard Occupational codes CPP - 2010 (summary categories)* Super Creative Core . Computer and mathematical occupations; . Architecture and engineering occupations; . Life, physical and social science occupations; . Higher education and library occupations; . Arts, design, entertainment, sports and media occupations. . Computing professionals (25); . Mathematicians, Statisticians and related professionals (212); . Architects, Engineers and related professionals (216; 214; 215); . Life Science professionals (213); . Physicists, Chemists and related professionals (211); . Social Science and related professionals (263); . University and higher education teachers (231); . Vocational, technological and artistic education teachers (232) - ELIMINATED . Secondary and basic education teachers (233) - ELIMINATED . Primary school and early childhood teachers (234) - ELIMINATED . Other teaching professionals (235) - ELIMINATED . Archivists, museum curators and related information professionals (262). + Bohemians . Authors, journalists and linguists (264); . Creative and performing artists (265); . Product and garment designers (2163); . Graphic and multimedia designers (2166); . Musicians, singers and composers (2652); . Dancers and choreographers (2653); . Film, stage and related directors and producers (2654); . Actors (2655); . Announcers on radio, television and other media (2656); . Creative and performing artists n.e.c. (2659); . Advertising and marketing professionals (2431); . Public relations professionals (2432); . Artistic, Entertainment and Sports associate professionals (342; 343); . Telecommunications and broadcasting technicians (352); . Fashion and other models (5241). Creative Professionals . Management occupations; . Business and financial operations occupations; . Legal occupations; . Drafters, engineering and mapping associate professionals; . Supervising managers and process control technicians; .Finance and sales associate professionals . Legislators, senior officials and managers (1); . Finance professionals (241); . Administration professionals (242) - ELIMINATED . Financial and mathematical associate professionals (331) - ELIMINATED . Sales and purchasing agents and brokers (332); . Business services agents (333) - ELIMINATED . Legal professionals (261); . Physical, engineering and mapping technicians, and drafters (311); . Supervising managers and process control technicians (312; 313; 315); . Information and communications technology operations and user support technicians (351); . Health professionals (except nursing) (221; 223; 224; 225; 226) – ELIMINATED . Nursing and midwifery professionals (222) - ELIMINATED . Nursing and midwifery associate professionals (322) - ELIMINATED . Life science technicians and related associate professionals (314) - ELIMINATED . Medical and pharmaceutical technicians and health associate professionals(321; 323; 324; 325) - ELIMINATED . Regulatory government associate professionals (3359) - ELIMINATED . Finance and sales associate professionals (2433; 2434). Source: The selection of codes is of the responsibility of the present paper’s authors, according to their interpretation of McGranahan and Wojan’s (2007: 205) refinement approach, based on the US O*NET database of occupations, available online at: http://www.onetcodeconnector.org/find/family/code?s=11 [accessed September 2014]. 73 Table A2. 3: Refinements of Florida’s proposal by Gabe (2006) Creative Class category groups Occupational Categories Descriptions Occupational ISCO-08 Codes (summary categories) / Portuguese Standard Occupational codes CPP - 2010 (summary categories)* Creative Core . Computer specialists and mathematical science occupations . Computing professionals (25); . Mathematicians, Statisticians and related professionals (212) . Architects, surveyors, and cartographers; Engineers; . Architects, Engineers and related professionals (216; 214; 215) . Life, Physical, Social scientists and related workers . Life Science professionals (213); . Physicists, Chemists and related professionals (211); . Social Science and related professionals (263) . Post-secondary teachers . Primary, secondary, and special education school teachers . Other teachers and instructors . Librarians, curators, and archivists . University and higher education teachers (231); . Vocational, technological and artistic education teachers (232) . Secondary and basic education teachers (233) ; . Primary school and early childhood teachers (234) ; . Other teaching professionals (235); . Archivists, museum curators and related information professionals (262) . Art and design workers . Entertainers and performers, sports, and related workers; Media and communication workers Authors, journalists and linguists (264); . Creative and performing artists (265); . Product and garment designers (2163); . Graphic and multimedia designers (2166); . Musicians, singers and composers (2652); . Dancers and choreographers (2653); . Film, stage and related directors and producers (2654); . Actors (2655); . Announcers on radio, television and other media (2656); . Creative and performing artists not elsewhere classified (2659); . Advertising and marketing professionals (2431); . Public relations professionals (2432); . Artistic, Entertainment and Sports associate professionals (342; 343); . Fashion and other models (5241). . Media and communication equipment workers . Information and communications technology operations and user support technicians (351); . Telecommunications and broadcasting technicians (352) . Top Executives/ Advertising, marketing, promotions, public relations, and sales managers/ Operations specialties managers/ Other management occupations . Legislators, senior officials and managers (1); . Finance professionals (241) - ELIMINATED . Administration professionals (242) - ELIMINATED . Financial and mathematical associate professionals (331) - ELIMINATED . Sales and purchasing agents and brokers (332) - ELIMINATED . Business services agents (333) - ELIMINATED . Legal professionals (261) - ELIMINATED . Physical, engineering and mapping technicians, and drafters (311) - ELIMINATED . Supervising managers and process control technicians (312; 313; 315) - ELIMINATED . Health professionals (except nursing) (221; 223; 224; 225; 226) – ELIMINATED . Nursing and midwifery professionals (222) - ELIMINATED . Nursing and midwifery associate professionals (322) - ELIMINATED . Life science technicians and related associate professionals (314) - ELIMINATED . Medical and pharmaceutical technicians and health associate professionals(321; 323; 324; 325) - ELIMINATED . Regulatory government associate professionals (3359) - ELIMINATED . Finance and sales associate professionals (2433; 2434) - ELIMINATED Note: The selection of codes is of the responsibility of the present paper’s authors, according to their interpretation of Gabe's (2006: 398, 400-401) refinement approach, based on the US O*NET database of occupations, available online at: http://www.onetcodeconnector.org/find/family/code?s=11 [accessed September 2014]. 74 ESSAY 3 _________________________________________________________ The neglected heterogeneity of spatial agglomeration and co-location patterns of creative employment: Evidence from Portugal ____________________________________________________________ 75 The neglected heterogeneity of spatial agglomeration and co-location patterns of creative employment: Evidence from Portugal * Abstract Empirical literature on the geographic location of creative activities has been traditionally based on the spatial analysis of industries, often disregarding the creative employment that lies outside the necessarily limited boundaries of creative industries. As an extension to the most recent methodologies using industry and occupational data on industrial cluster analysis, this paper analyses agglomeration and co-location patterns of core creative activities, considering both ‘embedded’ (creative professionals working outside the creative sectors) and ‘specialized’ (creative and support professionals working in the creative sectors) creative employment. Using location quotients and principal component factor and cluster analyses, applied to all 308 Portuguese municipalities, we found that the geographical agglomeration and co-location patterns of core creative groups differ substantially. The typical arguments sustained by literature - the tendency of creative industries/ employment to agglomerate and co-locate in large metropolises - are only supported in the case of creative activities that are based on knowledge-intensive services subject to Intellectual Property Rights, namely ‘Advertising/ Marketing’, ‘Publishing’, ‘TV/ Radio’, and ‘Software/ Digital Media’, densely concentrated and co-located in developed, large urban centres, with high levels of human capital. These arguments do not hold for the traditional creative activities of ‘Architecture’, ‘Design/ Visual Arts’ and ‘Crafts’, which, although co-located, appear mostly dispersed with small concentrations around intermediate urban centres. ‘Teaching/ training/ research’ present quite dispersed geographical patterns with some clusterization around municipalities with tertiary education institutions. ‘Film/ video/ photography’ and ‘Music/ Performing arts’ show some dispersion throughout the Portuguese territory with concentration around small urban centres and in rural areas. It is evident that, from agglomeration to co-location patterns, creative employment reveals heterogeneous characteristics across creative groups. Keywords: Spatial economics; Industrial location; Creative Industries; Portugal. JEL codes: C01, R12, R30. * To be published in the Annals of Regional Science, Springer, 2014. 76 1. Introduction The rising interest in the creative economy has encouraged several authors both in political and academic spheres to focus on creative industries and cultural activities (DCMS, 2001; Pratt, 2006; Higgs et al., 2008; UNCTAD, 2008) and to assess their effects on regional and national development (Capone, 2008; Miguel-Molina et al., 2012). According to several empirical studies, creative industries and creative occupations have a tendency to co-locate geographically (Capone, 2008; Lazzeretti et al., 2008, 2012) and are often associated to urban development and the growth of cities (Florida, 2002a). The uneven spatial patterns and the co-location behaviour of creative firms and creative workers are explained by territorial factors. Highly cited studies (e.g., Florida, 2002a, 2004) have shown that creative industries and workers tend to concentrate in metropolitan centres in order to take advantage of urbanization economies. The latter are provided by product differentiation, technological diversity, the geographic concentration of people, cultural diversity, and the diffusion of knowledge and innovation (Jacobs, 1969; Lorenzen and Frederiksen, 2008; Lazzeretti et al., 2008). Despite the acknowledged role of creativity in the development of regions, the literature on the economics of location regarding creative activities is relatively scarce and recent (Boix et al., 2013). 23 Methodologies are gradually being developed and the studies are often limited by the quality of data available in each country or region (Currid-Halkett and Stolarick, 2011; Boix et al., 2013). In the empirical literature, there is a primary corpus of research related to industrybased studies on the geographical location of creative industries/ creative industrial clusters (e.g., Lazzeretti et al., 2008, 2012; De Propris et al., 2009; Miguel-Molina et al., 2012; Bertacchini and Borrione, 2013; Boix et al., 2013; Lazzeretti, 2013). A second strand is concerned with the geography of creative occupations and creative workers (e.g., Florida, 2002a; Florida et al., 2008; Markusen et al., 2008; Boschma and Fritsch, 23 To have an idea, a search in the Scopus database with the keywords ‘creative industries’ or ‘creative occupations’ yielded 554 articles using these keywords in the fields ‘title’, ‘abstract’ and ‘keywords’. Adding the keyword ‘location’ to the search only returns 37 articles (and 54 articles if the word ‘geography’ is added), 6% (9%) of the unrestricted search on creative industries and occupations. 77 2009; Hansen et al., 2009; Mellander, 2009; Clifton and Cooke, 2010; Fritsch and Stuetzer, 2012). The use of either industry-based or occupational approaches leads to differing estimations of creative employment, the most common proxy to analyse the geographical patterns of the creative economy (Markusen et al., 2008; Bertacchini and Borrione, 2013). Besides, studies based on the Standard Industrial Classification (SIC) restrict the analysis to the total employment in creative industry sectors, considering all the workers (creative and non-creative) in the same production process of the final product (the creative good), overlooking creative employment in all the non-creative activity sectors. Occupational-based methodologies, using the Standard Occupation Classification (SOC) codes, provide an inter-sectorial depiction of the creative occupational structure across the economy, but disregard the value-chain and the productive process of creative goods, where occupations, creative and non-creative, may be fundamental. Recently, a third research path associated with methodologies that combine industry and occupational data (SIC/SOC) on the industrial analysis of creative/ cultural/ knowledgebased sectors (e.g., Barbour and Markusen, 2007; Markusen et al., 2008; Higgs et al., 2008; Currid and Stolarick, 2010a,b; DCMS, 2010, 2011, 2014; Currid-Halkett and Stolarick, 2011) has raised increasing interest as a way to overcome limitations of industry-based or occupational approaches and to provide an expanded analysis of local employment structures in the industrial spectrum across regions. As an extension to these recent methodological perspectives, this paper provides a detailed analysis of creative employment, at a highly disaggregated regional level, using a combined industry and occupational-based approach, which accounts for creative employment across all industry sectors - creative and non-creative. It also aims at analysing the potentially disparate geographical pattern of the several sub-groups of creative employment. The paper seeks therefore to answer the following questions:  Do core creative industries and creative occupations tend to agglomerate?  What are the main characteristics of the locations where creative employment tends to cluster - large metropolitan hubs, small urban centres, or rural areas? 78  Do the location patterns of core creative activities differ substantially among creative groups? Do more traditional creative sectors, such as Crafts, Design and Visual arts, tend to co-locate differently from those based on intellectual property, such as Advertising and Marketing, Software and Digital media? The analysis is carried out focusing on ten core creative groups: ‘Advertising and marketing’, ‘Architecture’, ‘Design and visual arts’, ‘Crafts’, ‘Film, video and photography’, ‘TV and radio’, ‘Music and the performing arts’, ‘Publishing’, ‘Software and digital media’, ‘Teaching, training and research’, in all the Portuguese territorial units (308 municipalities). The data used was extracted from the microeconomic Matched Employer-Employee Datasets, official databases from the Portuguese government, and each value was accurately obtained by programming the respective SIC and SOC code, using STATA 12.0®. This procedure avoided any potential overlapping of data. The structure of the paper is as follows. Section 2 presents a brief review of the empirical literature on the location of creative industries and occupations. Section 3 outlines the main aspects of the methodology followed. In Section 4, the analysis of spatial patterns of agglomeration and co-location of core creative employment in Portugal is presented, and main results are discussed. Section 5 puts forward the study’s major conclusions. 2. Empirical literature on the location of creative industries and occupations: a brief review Over the past decade, the academic and political debate on industrial location has gradually come to highlight the geography of knowledge-intensive services and the clustering of ‘soft innovation’ and creative activities as drivers of regional growth (UNCTAD, 2008; Stoneman, 2009). Following the original study by DCMS (1998, 2001) on the mapping of creative industries in the UK, a considerable amount of case studies on creative clusters, cultural quarters or creative cities has been put forward in several regions of the developed world (e.g., Scott, 2000; Wiesand and Söndermann, 2005; Wu, 2005; Pratt, 2006; Roodhouse, 2006). These studies emphasize the importance that the clustering of creative activities has on producing agglomeration and urbanization economies (Jacobs, 1969), which contribute to the economic growth of 85 (…) FOCUS on CREATIVE OCCUPATIONS :: Location Analysis using SOC codes and Florida’s (2002a, 2004) taxonomy on the ‘creative class’ Author(s) Regions Industry Sectors Occupations Location measures | Indicators Main empirical results Agglomeration Co-location Boschma and Fritsch (2009) . Europe 503 regions (NUTS 3) in: Denmark, England/ Wales, Finland, Germany, the Netherlands, Norway, Sweden - ‘Creative class’: Super creative core, Creative professionals and Bohemians. . Regional share of the creative class; . Gini coefficient Finland, Norway, Denmark, Sweden: more spatially concentrated. Germany, the Netherlands, and England/ Wales: ‘creative class’ more dispersed. High spatial correlation of the shares of high-technology employment// creative core// creative professionals// employees with a tertiary degree. Clifton and Cooke (2009) . Europe (UK, Sweden, Denmark, Norway, Finland, the Netherlands, Germany) - NUTS 3. . North American large metropolitan areas - ‘Creative class’ as a whole, particularizing, then, for the Super Creative Core and Bohemians. . LQ UK, Netherlands: ‘creative class’ more spatially concentrated. Norway, Denmark, Finland, Sweden: ‘creative class’ less evenly distributed than in Germany. Germany: more evenly distributed. Bohemian index, openness and the public provision index - most significant location factors. - Andersen et al. (2010) . Denmark, Finland, Norway, Sweden - 263 functional city regions (at the level of NUTS 4 and equivalent regional units) - ‘Creative class’ as a whole. . LQ Small and large Nordic city Regions: location of ‘creative class’ related to Openness. Medium Nordic city regions: location of ‘creative class’ related to the presence of Bohemians. - Fritsch and Stuetzer (2009, 2012) . Germany (German districts) - ‘Creative class’ major category groups: Super creative core; Creative Professionals; Employed Bohemians and Freelance artists. . LQ; . Population share Berlin: Bohemians and freelance artists. Share of employed bohemians is high in cities. German medium-sized cities: highest share of the Creative Core. - 86 (…) FOCUS on CREATIVE INDUSTRIES and OCCUPATIONS :: Location analysis using SIC and SOC codes Author(s) Regions Industry Sectors Occupations Location measures | Indicators Main empirical results Agglomeration Co-location Barbour and Markusen (2007) . California, USA (eleven metropolitan areas) . All activity sectors (innovation/ information/ research-intensive vs mature/ market-oriented industries) . All occupational categories except those in forestry, farming and fishing industries. . Occupational employment by industry, by region . Employment shares; . LQ Metropolitan areas of California: higher concentration of ‘managerial/ professional’ and ‘clerical’ workers; lower shares of ‘service’, ‘manual’, ‘precision’ and ‘sales’ workers, when compared to the national share; Diversified occupational-industry structures across the eleven metro areas. - San Francisco Bay: high concentration of.high-tech/ research occupations (computer professionals, selected engineers and natural scientists; - San Jose metro area / Sillicon Valley: higher concentration of Computer/IT specialists. Innovative/ research oriented industries/ occupations: overrepresented in California regions when compared to the national distribution. Mature industries: the occuaptional mix is more similar to the national structure (e.g., services). - Higgs et al. (2008) . United Kingdom (national level) .Core Creative Industries following DCMS (1998, 2001). . “Creative occupations are a selection of occupations which produce creative goods or services, drawn from the UK SOC codes” (Hiigs et al., 2008: 19). . Employment shares (in and outside the core of creative industries) - - Currid and Stolarick (2010a) . Los Angeles metropolitan área . Information Systems (IS)/ Information Technology (IT) industries. . IS/ IT occupations. . Employment shares; . LQ Los Angeles: higher share of employment in ‘Network systems and Data communications Analysts’, in most Management occupations and in Designers, when compared to the US IS/ IT industry employment. Los Angeles: lower share of ‘Computer software engineers and Computer scientists and systems analysts’. - Currid and Stolarick (2010b) . Los Angeles and New York City metropolitan statistical areas . Cultural Industries: Publishing industries; Motion Picture/ Video Industries; Broadcasting; Performing Arts/ Sports; Museums; Amusement/ Gambling/ Recreation industries (authors’ selection). . Cultural Occupations: Arts, Design, Entertainment, Sports, Media and Museumrelated occupations (authors’ selection). . Employment shares; . LQ Clustering of artistic/ cultural industries in NY and LA, but differentiated specialization patterns: - New York: Fashion and Arts-related industries; - Los Angeles: Film and Fashion industries. Different occupational structures: - New York: higher concentration of musicians, fashion designers, writers and artists. - Los Angeles: dancers, actors and multimedia artists. - DCMS (2010, 2011, 2014) . United Kingdom (national scale) .Core Creative Industries following DCMS (2001). . Creative occupations inside and outside the core of creative industries and noncreative occupations in creative industries. . Employment shares - - CurridHalkett and Stolarick (2011) . USA (30 largest metropolitan areas) Artistic Industries following Currid and Stolarick (2010b). Artistic Occupations following Currid and Stolarick (2010b). . Employment shares; . LQ Artistic activities tend to concentrate in large cities; But artistic industries in the top 30 metros are comprised of the same occupations as that of the nation. Overall, artistic industries and artistic occupations do not colocate. 87 Others discriminate between ‘traditional’ and ‘non-traditional’ creative sectors (e.g., Lazzeretti et al., 2008, 2012; Boix et al., 2013), arguing that each set of industries has distinguishing features and location patterns. Recently, Bertacchini and Borrione (2013: 141) distinguish among “content and service-oriented creative industries, craft-based creative industries and industrial design activities”. Summing up, the empirical literature on the location of creative activities is conspicuously divided into: i) studies on industries, using SIC codes to process regional data on industry sectors; ii) studies on occupations, using SOC codes to examine occupational structures across regions and countries; and iii) studies on industries and occupations, employing SIC and SOC codes to analyse the occupational structure by industry, across regions or/ and at a national scale. The present paper appears as an extension to these recent studies on industries and occupations, aiming to provide deeper insights on the location patterns of creative industries and creative employment, at a high level of regional disaggregation. 3. Methodology In order to analyse the agglomeration and co-location patterns of creative employment, ten core creative groups were considered - ‘Advertising and Marketing’, ‘Architecture’, ‘Design and Visual arts’, ‘Crafts’, ‘Film, video and photography’, ‘TV and Radio’, ‘Music and the Performing arts’, ‘Publishing’, ‘Software and Digital media’, ‘Teaching, training and research’ - and were obtained by using both industry and occupational data. The mapping methodology used here is described in detail in Cruz and Teixeira (2013), and is summarized in Table 3.2. Data on industry sectors and on occupations was extracted from Quadros de Pessoal, Matched Employer-Employee Databases from GEE/ ME (Gabinete de Estratégia e Estudos/ Ministry of Economy, Portugal), for the most recent year available at the time of this study, 2009. 24 All the figures have been thoroughly extracted using STATA 24 According to the latest data available (2009), national employment in the private, structured sector totalled 3.128.126 workers. It covers all employment in industries and establishments operating in the national territory with at least one employee, excluding Public Administration and self-employment. Cruz and Teixeira (2013) discuss the implications of such exclusions in the estimation of core creative employment. 88 12.0®, which yielded valid, non-overlapping information for all (308) Portuguese territorial units, at the regional level of the municipality. Agglomeration and co-location patterns are analyzed in terms of core creative employment, comprising ‘embedded’ creative employment, which includes creative professionals employed in all the sectors of the economy considered as non-creative, and ‘specialized/ industrial’ creative employment, which encompasses all the professionals working in the creative industry sectors. To assess agglomeration, the location quotient (LQ) was used as the basis indicator, given its treatability and suitability as a measure of industrial concentration in a region (Lazzeretti et al., 2008, 2012; Miguel-Molina et al., 2012). The LQ is computed as follows: EmploymentTotalNational EmploymentCreativeNational EmploymentTotal EmploymentCreative LQ i j ij ij  , where i is each group of core creative employment (i=1,…, 10) and j stands for each municipality j (j=1,…, 308). In the analysis of co-location patterns, a Principal Component/ Factor Analysis and Cluster (hierarchical and non-hierarchical K-means) analyses were conducted on the LQs (used as independent variables) of each of the ten core creative groups, using the SPSS® software. These procedures served to establish groups of municipalities according to common factors of specialization in the ten core creative groups, and to reduce the 308 municipalities to a specific number of homogeneous clusters. Finally, in order to better describe the clusters obtained, a set of indicators was gathered which were identified with four types of factors commonly associated to the agglomeration and co-location of creative activities, in literature: 1) Talent/ Human Capital (Florida, 2002a, 2004, 2005; Florida et al., 2008; Boschma and Fritsch, 2009; Clifton and Cooke, 2010; Lazzeretti et al., 2012); 2) Tolerance/ Openness (Florida, 2002a, 2004; Boschma and Fritsch, 2009; Clifton and Cooke, 2010; Fritsch and Stuetzer, 2012; Lazzeretti et al., 2012); 3) Urban agglomeration and cultural amenities (Florida, 2002b; Boschma and Fritsch, 2009; Clifton and Cooke, 2010); and 4) Urban and regional development (Florida et al., 2008; Clifton and Cooke, 2010; MiguelMolina, 2012). Table 3.3 details the indicators selected and their respective sources. 89 Table 3. 2: Mapping core creative employment using Industry and Occupational codes Core Creative Sectors Industry sectors Portuguese CAE – Rev. 3 Industry codes (SIC) Creative Occupations categories Portuguese CNP94 Ocupacional Nomenclature (SOC) 1. Advertising and Marketing Advertising; Market research/ public opinion polling 7311; 7312; 7320 Sales/marketing managers; Public relations managers and professionals; Advertising/ marketing professionals; Survey and market researchers 1233; 1234; 2419; 341505; 341510; 419090 2. Publishing Publishing of books, periodicals/ others; Translation/interpretation activities; Libraries/archives/ museum activities; Information service activities (news agencies) 5811; 5812; 5813; 5814; 5819; 7430; 9101; 9102; 9103; 9104; 6391; 6399 Writers/ journalists; Philologists/ translators/ interpreters; Graphic arts composers; Archivists/ museum curators; Librarians 2451; 2444; 7341; 2431; 2432; 343115 3. Architecture Architectural activities 7111 Building, landscape, town planning Architects; Cartographers/ surveyors; Draughts persons 2141; 2148; 3118 4. Design and Visual Arts Design activities 7410 Visual artists; Designers; Decorators 2452; 3471 5. Crafts No SIC codes match this category - Technicians of precision instruments; Jewelers/cutters; Potters; Glass makers/ molders/polishers; Decorative painters; Cutters/engravers of glass and ceramics; Handicraft workers in wood/basketry; Woodworkers; Handcrafters in fabric/leather; Handloom weavers; Tailors/ dressmakers/ furriers/ hatters 3115; 7311; 7312; 7313; 7321; 7322; 7323; 7324; 7331; 7424; 7422; 7332; 7432; 7433; 7434 6. Film, Video and Photography Motion picture, video and television production, postproduction, distribution and projection activities; Photographic activities 5911; 5912; 5913; 5914; 7420 Film Directors/ Producers; Assistants of scene/ film production; Photographers/equipment technicians for the recording of image and sound; Photographic developing / printing professionals; Cultural Promoters 2455; 3131; 343120;514920; 514945; 7344 7. TV and Radio Radio activities; Television activities 6010; 6020 Speakers/ announcers of radio/television /entertainment activities; TV Producers; Technicians of audio broadcasting (radio/television/ telecommunications) 3472; 121040; 311410; 311490; 313205; 313290 8. Music/ Entertainment and the Performing Arts Sound recording/music publishing activities; Performing arts; Support activities to performing arts; Artistic and literary creation; Operation of arts facilities; Amusement/ recreation activities 5920; 9001; 9002; 9003; 9004; 9321/9 Actors; Composers/musicians/singers; Dancers; Choreographers; Restaurant/ Cafeteria Chefs 245510; 2453; 3473; 2454; 514950; 512105; 512205 9. Software and Digital Media Software publishing; Computer programming/ consultancy; Data processing/hosting/Web portals 5821; 5829; 6201; 6202; 6203; 6209; 6311; 6312 Computer systems professionals; Computing programmers; Directors of computing /IT; Computing/ IT technicians 2131; 3121; 1236; 3122 10. Teaching, training and research Research on natural sciences, engineering, social sciences and humanities 7211; 7219; 7220 Physicists/ Chemists; Mathematicians/ Statisticians; Life science professionals; Secondary/ Higher education teachers; Social/ Human sciences professionals 211; 212; 221; 23; 244 90 Table 3. 3: Indicators associated with creative activities’ literature, that were selected to describe the clusters of municipalities Group/ type Indicator Indicator computation Source Talent/ Human Capital [Florida (2002a, 2004, 2005), Florida et al. (2008), Boschma and Fritsch (2009), Clifton and Cooke (2010), Lazzeretti et al. (2012)] Proportion of population with completed tertiary education Resident population with 21 and more years old with complete tertiary education, in total resident population with 21 and more years old INE, National Statistics. Census 2001. Gross enrolment rate in upper secondary education Pupils enrolled on upper secondary education in total resident population aged between 15 and 17 years old INE, National Statistics. Reference period: 2010-2011 Proportion of professionals socially more valued Proportion of employed population in the occupational categories of ‘Representatives of legislative/executive bodies/officers/directors/executive managers’ or of ‘Specialists of intellectual and scientific activities’ in total employed population INE, National Statistics. Reference period: 2011. Tolerance/ Openness [Florida (2002a, 2004), Boschma and Fritsch (2009), Clifton and Cooke (2010), Fritsch and Stuetzer (2012), Lazzeretti et al. (2012)] Foreign population Number of foreign individuals who have applied for resident status per 100 inhabitants INE, National Statistics. Reference period: 2007. Social inequality ratio Calculation based on the weight of each socioeconomic group in the municipality’s population INE, National Statistics. Reference period: 2001. Total (regional) Attraction Rate Proportion of resident population that 5 years before inhabited in another territorial unit or another country in total resident population in the territorial unit INE, National Statistics. Reference period: 2011 Urban agglomeration and cultural amenities [Florida (2002b), Boschma and Fritsch (2009), Clifton and Cooke (2010)] Population’s density Number of individuals per square kilometer INE, National Statistics. Reference period: 2011. Firms’ density Number of firms per square kilometer INE, National Statistics. Reference period: 2010. Museums/ zoological/ botanic gardens/ aquariums Number by geographic localization INE, National Statistics. Reference period: 2011. Rooms/ spaces of live shows and performances Number by geographic localization INE, National Statistics. Reference period: 2011. Urban and regional development [Florida et al. (2008), Clifton and Cooke (2010), MiguelMolina (2012)] Employment polarization index Employed population in the territorial unit/ Employed resident population in the territorial unit INE, National Statistics. Census 2011. Proportion of purchasing power by geographic localization Index of Purchasing Power (Portugal=100) weighted by each municipality’s population weight (municipality’s population/ national population) INE, National Statistics. Reference period: 2009. Average monthly earnings (euros) Average monthly amount in Euros (per worker) by geographic localization INE, National Statistics. Reference period: 2009. 91 4. Agglomeration and co-location of core creative employment in Portugal: results 4.1. Agglomeration of creative employment in each core creative group The analysis of the location quotient for each creative group and its spatial visualization indicates that the geographical patterns of core creative employment differ substantially among the ten groups of core creative sectors considered (see Figure 3.1). Specifically, ‘Advertising and marketing’, ‘Publishing’, ‘TV and radio’, and ‘Software and digital media’ tend to agglomerate around the largest/ most important urban centres, notably the largest Portuguese cities, Lisbon (the capital) and Porto, plus Oeiras, a highly populated municipality near Lisbon. In contrast, ‘Teaching, training and research’ present quite dispersed geographical patterns around municipalities with tertiary education (university and polytechnic) institutions (e.g., Bragança, Porto, Coimbra, Viseu, Lisboa, Évora, Beja, Faro). ‘Film, video and photography’ and ‘Music and the performing arts’ present similar geographical patterns showing dispersion throughout the Portuguese territory with some concentration around small urban centres. ‘Music/ Performing Arts’ are mostly found in touristic locations located in coastal areas, whereas the independent production of ‘Film/ Video/ Photography’ is dispersed across inland municipalities, where public festivities and social events play an important role. Dispersion is also a characteristic of Architecture’, ‘Design and visual arts’ and ‘Crafts’, although they present some agglomeration in intermediate urban centres. Architecture has a long tradition in the school of Porto (e.g., modern architecture - Souto Moura and Siza Vieira). Design is mainly related to fashion/ industrial design activities in the textile manufacturing industries concentrated in the North of mainland Portugal. Handicraft activities are related to traditional arts and crafts particularly enrooted in the North-Centre of the country, located near sources of raw materials or where craft activities have long-standing tradition (e.g., textiles, ceramics, glass, woodcrafts, basketry): jewellery/ goldsmiths in Gondomar; woodcrafts in Paços de Ferreira; embroidery/ weaving in Fafe; tinsmithing/ ceramics/ porcelain, in Batalha; glass crafts, in Marinha Grande. 92 Figure 3. 1: Agglomeration patterns of Portuguese core creative groups Advertising and Marketing Publishing Architecture Design and Visual Arts Crafts Highly specialized municipalities: 15 Highly specialized municipalities: 15 Highly specialized municipalities: 15 Highly specialized municipalities: 15 Highly specialized municipalities: 15 Specialized municipalities: 5 Specialized municipalities: 15 Specialized municipalities: 44 Specialized municipalities: 34 Specialized municipalities: 54 Non-specialized municipalities: 288 Non-specialized municipalities: 278 Non-specialized municipalities: 249 Non-specialized municipalities: 259 Non-specialized municipalities: 239 Concentrated in large urban centres. Dispersed across intermediate urban centres. Oeiras (3.43); Lisboa (2.28); Matosinhos (1.45); Sintra (1.28); Porto (1.22) Oeiras (2.89); Lisboa (2.60); Porto (2.40); Amadora (2.24); Sintra (1.75) Oliveira de Azeméis (1.93); Águeda (1.69); Porto (1.64); Ovar (1.52); Oeiras (1.52) Barcelos (3.42); Guimarães (2.99); Vila Nova de Famalicão (2.52); Águeda (2.22); Marinha Grande (2.15) Batalha (8.79); Gondomar (7.37); Fafe (5.37); Paços de Ferreira (5.30); Paredes (4.75) Notes: Non-specialized municipalities are those whose Location Quotient (LQ) < 1.00; Specialized municipalities are those whose LQ is > 1.00 but lower than the value corresponding to the 95th percentile of the LQ; Highly specialized municipalities correspond to those whose LQ is above the 95th percentile of the LQ. Source: Authors’ computations based on micro-data from the Matched Employer-Employee Databases, GEE/ ME, Ministry of Economy, Portugal (2009). 93 Figure 3.1 (cont.): Agglomeration patterns of Portuguese core creative groups TV and Radio Music and the Performing Arts Film, Video and Photography Software and Digital Media Teaching, training and research Highly specialized municipalities: 15 Highly specialized municipalities: 15 Highly specialized municipalities: 15 Highly specialized municipalities: 15 Highly specialized municipalities: 15 Specialized municipalities: 20 Specialized municipalities: 65 Specialized municipalities: 72 Specialized municipalities: 1 Specialized municipalities: 109 Non-specialized municipalities: 263 Non-specialized municipalities: 228 Non-specialized municipalities: 221 Non-specialized municipalities: 292 Non-specialized municipalities: 184 Concentrated in most important urban centres. Dispersed across small urban centres and rural areas. Concentrated in large urban centres Dispersed across small urban centres and inland/ rural areas. Concentrated in large urban centres. Very dispersed across large and intermediate urban centres. Oeiras (6.53); Ponta Delgada (Azores Island); (4.59); Lisboa (3.43); Vila Nova de Gaia (3.38); Funchal (Madeira Island) (2.48) Pedrógão Grande (11.4); Calheta (Madeira Island) (7.11); Lagoa (6.60); Albufeira (4.54); Lagos (3.88) Seia (3.74); V. Franca de Xira (2.74); S. João da Madeira (2.71); Espinho (2.70); Oeiras (2.28); Tavira (2.18) Oeiras (5.72); Amadora (3.04); Lisboa (2.57); Porto (1.83); Matosinhos (1.20) Oliveira de Frades (5.31); Beja (2.17); Oeiras (2.06); Porto (1.77); Coimbra (1.74) 94 4.2. Co-location of core creative employment Considering all the ten core creative groups and based on principal component and factor analysis, 25 we estimated the latent factors able to explain the correlational behaviour among the LQs of each core creative sector and, thus, capture the co-location patterns of core creative employment. The analysis of the Rotated Component Matrix (cf. Table 3.4) uncovered four main latent factors, which together explained approximately 57% of total variance of the original variables. Factor 1, labelled ‘Intellectual Property creative employment’, associates the core creative groups ‘Advertising/ Marketing’, ‘Publishing’, ‘TV/ Radio’, and ‘Software/ Digital Media’ with component 1. These activities appear co-located as they are human-capital/ knowledge-intensive activities generally dedicated to the production of intangible creative contents subject to intellectual property rights. Table 3. 4: Co-location analysis of the ten core creative groups: estimated Rotated Component Matrix Factors’ labels Core Creative groups Location Quotients (LQs) FC1 FC2 FC3 FC4 IP core creative Software/ Digital Media 0.799 0.230 0.055 0.151 TV/ Radio 0.668 -0.127 -0.126 -0.080 Advertising/ Marketing 0.631 0.286 0.181 0.141 Publishing 0.495 -0.065 0.013 -0.079 Traditional core creative Design/ Visual Arts 0.091 0.791 -0.138 -0.058 Crafts -0.222 0.658 0.175 -0.130 Architecture 0.213 0.579 -0.082 0.187 Leisure versus Intellectual/ Mental core creative Music/ Performing Arts 0.058 -0.163 0.809 0.171 Teaching/ Training/ Research 0.008 -0.158 -0.605 0.459 Independent/ freelance core creative Film/ Video/ Photography -0.019 0.026 0.049 0.864 Total variance explained 19% 16% 11% 11% ‘Architecture’, ‘Design/ Visual Arts’ and ‘Crafts’ emerge as highly correlated with component 2. We labelled factor component 2 as ‘Traditional core creative’ as it 25 Factor analysis assesses the structure of a set of interrelated observed variables in order to find a low number of intrinsic/ latent factors that may partially explain the behaviour of original variables. If two variables are (not spuriously) correlated, their interdependency results from a common, not directly observable feature, i.e., a latent factor (Maroco, 2011: 471). 101 highlight the importance of considering the specificities of each core creative sector in location studies. ‘Leisure creative’ activities (‘Music/ Performing Arts’), understood as a catch-all measure of entertainment/ bohemian activities (Florida, 2002a), are mainly found in tourism, coastal municipalities, with a low presence of ‘Intellectual/ Mental creative’ activities (‘Teaching/ training/ research’), which, in turn, are dispersed around university cities in inland areas of the country. These results differ from Florida’s (2002a,b, 2004) popular findings. Portugal’s intermediate urban development and the high concentration of leisure and tourism activities in coastal areas may help explain these outcomes. Nevertheless, our findings suggest that municipalities with a higher presence of Leisure activities are those with higher levels of tolerance. This corroborates Florida et al.’s (2008) argument that there is a positive relation between tolerance and the presence of entertainment/ artistic activities in a region. The diversity of geographical patterns becomes evident when detailing the spatial analysis of each core creative group. From their agglomerative behaviour to co-location patterns, creative employment reveals heterogeneous characteristics across creative groups. This heterogeneity has to be appropriately acknowledged when designing and implementing policy strategies which relate creativity with regional development. Such policy measures should not be only creativity-oriented but also specific to each type of core creative group (knowledge-intensive, traditional, leisure, intellectual, independent/ freelance). Furthermore, the results obtained allow us to conclude that the differentiated (co)location patterns of creative activities are mainly a regional phenomenon, distinguishing regions within the same country, and not only an aspect differentiating countries in international comparisons, as in Lazzeretti et al. (2012). This study did not extend its analysis to factors behind the spatial patterns of agglomeration and co-location observed. Further research on the determinants of location would broaden our understanding of these patterns and sustain the formulation of more appropriate regional policies on creativity and regional growth. Undoubtedly, this constitutes grounds for future lines of research and further steps to be explored. 102 Acknowledgements A word of sincere appreciation to the referees of the Annals of Regional Science for their insightful comments and suggestions. The authors are also grateful to Professor Armindo Carvalho (Faculdade de Economia, Universidade do Porto, Portugal) for the valuable suggestions on the use of statistical procedures and the study of spatial patterns using ArcGIS 10.1®. We acknowledge the courtesy of the GEE/ ME - Gabinete de Estratégia e Estudos/ Ministry of Economy, Portugal, for permitting the access to data which turned this study possible. The GEE/ ME is not responsible for the results and interpretations contained in this paper. These are of the authors’ full responsibility. This study had the financial support of the Portuguese government - Foundation for Science and Technology (FCT), Ministry of Education and Science, and was co-financed by the European Social Fund [FCT Doctoral Grant SFRH-BD-69571-2010]. References Andersen, K., Hansen, H., Isaksen, A., Raunio, M. (2010), “Nordic city regions in the creative class debate - Putting the creative class thesis to a test”, Industry and Innovation, Vol. 17, Nº2, pp. 215-240. Barbour, E., Markusen, A. (2007), “Regional occupational and industrial structure: Does one imply the other?”, International Regional Science Review, Vol. 30, Nº1, pp. 72-90. Bathelt, H. (2002), “The re-emergence of a media industry cluster in Leipzig”, European Planning Studies, Vol. 10, Nº5, pp. 583-611. Bertacchini, E., Borrione, P. (2013), “The geography of the Italian creative economy: The special role of the design and craft-based industries”, Regional Studies, Vol. 47, Nº2, pp. 135-147. Boix, R., Lazzeretti, L., Capone, F., De Propris, L., Sánchez, D. (2013), “The geography of creative industries in Europe: Comparing France, Great Britain, Italy and Spain”, in L. Lazzeretti (Ed.), Creative Industries and Innovation in Europe: Concepts, Measures and Comparative Case Studies, pp. 23-44, New York: Routledge. 103 Boix, R., Galletto, V. (2006), “El mapa de los distritos industriales de España”, Economía Industrial, 359, pp. 95-112. Borén, T., Young, C. (2013), “The Migration Dynamics of the “Creative Class”: Evidence from a Study of Artists in Stockholm, Sweden”, Annals of the Association of American Geographers, Vol. 103, Nº1, pp. 195-210. Boschma, R., Fritsch, M. (2009), “Creative class and regional growth: Empirical evidence from seven European countries”, Economic Geography, Vol. 85, Nº4, pp. 391-423. Campbell-Kelly, M., Danilevsky, M., Garcia-Swartz, D., Pederson, S. (2010), “Clustering in the Creative Industries: Insights from the Origins of Computer Software”, Industry and Innovation, Vol. 17, Nº3, pp. 309-329. Capone, F. (2008) “Mapping and analysing creative systems in Italy (1991–2001)”, in P. Cooke and L. Lazzeretti (Eds.), Creative Cities, Cultural Clusters and Local Economic Development, Chapter 14, pp. 338-364, UK: Edward Elgar. Chapain, C., Cooke, P., De Propris, L., MacNeill, S., Mateos-Garcia, J. (2010), Creative Clusters and Innovation. Putting Creativity on the Map, London, UK: NESTA. Clifton, N., Cooke, P. (2010), “Creative knowledge workers and location in Europe and North America: A comparative review”, Creative Industries Journal, Vol. 2, Nº1, pp. 73-89. Cruz, S., Teixeira, A. (2013), “Assessing the magnitude of creative employment: a comprehensive mapping and estimation of existing methodologies”, European Planning Studies (DOI: 10.1080/09654313.2013.822475). Currid, E., Stolarick, K. (2010a), “The occupation-industry mismatch: New trajectories for regional cluster analysis and economic development”, Urban Studies, Vol. 47, Nº2, pp. 337-362. Currid, E., Stolarick, K. (2010b), “Cultural capital and metropolitan distinction: Views of Los Angeles and New York”, City, Culture and Society, Vol. 1, pp. 217-223. 104 Currid, E., Williams, S. (2010), “Two cities, five industries: Similarities and differences within and between cultural industries in New York and Los Angeles”, Journal of Planning Education and Research, Vol. 29, pp. 322-335. Currid-Halkett, E., Stolarick, K. (2011), “The arts: Not just artists (and vice versa)” in D. Andersson, A. Andersson and C. Mellander (Eds.), The Handbook of Creative Cities, pp. 259-283, London/ New York: Edward Elgar Publishing. Department for Culture, Media and Sport (DCMS) (1998), Creative Industries Mapping Document 1998, London: UK DCMS. Department for Culture, Media and Sport (DCMS) (2001), Creative Industries Mapping Document 2001, London: UK DCMS. Department for Culture, Media and Sport (DCMS) (2010), Creative Industries Economic Estimates – December 2010 (Experimental statistics) - Full Statistical Release, London: DCMS, https://www.gov.uk/government/uploads/ system/uploads/attachment_data/file/77514/CIEE_Full_Release_Dec2010.pdf [accessed September 2014]. Department for Culture, Media and Sport (DCMS) (2011), Creative Industries Economic Estimates – December 2011, London: DCMS, https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/779 59/Creative-Industries-Economic-Estimates-Report-2011-update.pdf [accessed September 2014]. Department for Culture, Media and Sport (DCMS) (2014), Creative Industries Economic Estimates January 2014 Statistical Release, London: DCMS, https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/271 008/Creative_Industries_Economic_Estimates_-_January_2014.pdf [accessed September 2014]. De Propris, L., Chapain, C., Cooke, P., MacNeill, S., Mateos-García, J. (2009), The Geography of Creativity, London: NESTA Interim Report, August. Faggian, A., Comunian, R., Jewell, S., Kelly, U. (2013), “Bohemian graduates in the UK: Disciplines and location determinants of creative careers”, Regional Studies, Vol. 47, Nº2, pp. 183-200. 105 Florida, R. (2002a), The rise of the Creative Class – and How it’s Transforming Work, Leisure, Community and Everyday Life, New York: Basic Books. Florida, R. (2002b), “The Economic Geography of Talent”, Annals of the Association of American Geographers, Vol. 92, Nº4, pp. 743-755. Florida, R. (2004), The rise of the Creative Class – and How it’s Transforming Work, Leisure, Community and Everyday Life (Revised paperback ed.), New York: Basic Books. Florida, R. (2005), The Flight of the Creative Class: The New Global Competition for Talent, New York: Harper Collins Publishers. Florida, R., Mellander, C. (2007), “The creative class or human capital? Explaining regional development in Sweden”, CESIS Electronic Working Paper Series. Florida, R., Mellander, C., Stolarick, K. (2008), “Inside the black box of regional development - human capital, the creative class and tolerance”, Journal of Economic Geography, Vol. 8, pp. 615-649. Fritsch, M., Stuetzer, M. (2009), “The geography of creative people in Germany”, International Journal of Foresight and Innovation Policy, Vol 5, Nº1/2/3, pp. 723. Fritsch, M., Stuetzer, M. (2012), “The geography of creative people in Germany revisited”, Jena Economic Research Papers, No. 2012/065. Hansen, H., Asheim. B., Vang, J. (2009), “The European creative class and regional development: how relevant is Florida's theory for Europe?”, in L. Kong and J. O’ Connor (Eds.), Creative Economies, Creative Cities: Asian-European Perspectives, pp. 99-120, Dordrecht: Springer. Hansen, H., Niedomysl, T. (2009), “Migration of the creative class: evidence from Sweden”, Journal of Economic Geography, Vol. 9, pp. 191-206. Higgs, P., Cunningham, S., Bakshi, H. (2008), Beyond the Creative Industries: Mapping the Creative Economy in the United Kingdom, London: National Endowment for Science, Technology and the Arts (NESTA). Jacobs, J. (1969), The Economy of Cities, New York: Vintage. 106 Kolenda, R., Yang Liu, C. (2012), “Are Central Cities More Creative? The Intrametropolitan Geography of Creative Industries”, Journal of Urban Affairs, Vol. 34, Nº5, pp. 487-512. Krätke, S. (2002), “Network analysis of production clusters: The Potsdam/Babelsberg film industry as an example”, European Planning Studies, Vol. 10, Nº1, pp. 2754. Krätke, S. (2010), “‘Creative cities’ and the rise of the dealer class: A critique of Richard Florida’s approach to urban theory”, International Journal of Urban and Regional Research, Vol. 34, Nº4, pp. 835-853. Lazzeretti, L. (2013), Creative Industries and Innovation in Europe: Concepts, Measures and Comparative Case Studies, New York: Routledge. Lazzeretti, L., Boix, R., Capone, F. (2008), “Do creative industries cluster? Mapping creative local production systems in Italy and Spain”, Industry and Innovation, Vol. 15, Nº5, pp. 549-567. Lazzeretti, L., Capone, F., Boix, R. (2012), “Reasons for clustering of creative industries in Italy and Spain”, European Planning Studies, Vol. 20, Nº8, pp. 12431262. Lorenzen, M., Frederiksen, L. (2008), “Why do cultural industries cluster? Localization, urbanization, products and projects”, in P. Cooke and L. Lazzeretti (Eds.), Creative Cities, Cultural Clusters and Local Economic Development, Chapter 7, pp. 155-179, Cheltenham. UK: Edward Elgar. Lorenzen, M., Andersen, K. (2009), “Centrality and Creativity: Does Richard Florida’s Creative Class Offer New Insights into Urban Hierarchy?”, Economic Geography, Vol. 85, Nº4, pp. 363-390. Markusen, A., Wassall, G., DeNatale, D., Cohen, R. (2008), “Defining the Creative Economy: industry and occupational approaches”, Economic Development Quarterly, Vol. 22, Nº1, pp. 24-45. Maroco, J. (2011), Análise Estatística com o SPSS Statistics (fifth Ed.), Pero Pinheiro: ReportNumber. 107 Martin-Brelot, H., Grossetti, M., Eckert, D., Gritsai, O., Kovács, Z. (2010), “The Spatial Mobility of the ‘Creative Class’: A European Perspective”, International Journal of Urban and Regional Research, Vol. 34, Nº4, pp. 854-870. Mellander, C. (2009), “Creative and knowledge industries: an occupational distribution approach”, Economic Development Quarterly, Vol. 23, Nº4, pp. 294-305. Miguel-Molina, B., Hervas-Oliver, J., Boix, R., Miguel-Molina, M. (2012), “The importance of creative industry agglomerations in explaining the wealth of European regions”, European Planning Studies, Vol. 20, Nº8, pp. 1263-1280. Pratt, A. (2006), “Advertising and creativity, a governance approach: a case study of creative agencies in London”, Environment and Planning A, Vol. 38, Nº10, pp. 1883-1899. Roodhouse, S. (2006), Cultural Quarters: Principles and Practice, Bristol: Intellect. Scott, A. (2000), The Cultural Economy of Cities, London: Sage. Scott, A. (2002), “A new map of Hollywood: the production and distribution of American motion pictures”, Regional Studies, Vol. 36, Nº9, pp. 957-975. Stoneman, P. (2009), Soft innovation: Towards a More Complete Picture of Innovative Change, UK: NESTA. Turok, I. (2003), “Cities, clusters and creative industries: The case of film and television in Scotland”, European Planning Studies, Vol. 11, Nº5, pp. 549-565. UNCTAD (2008), Creative economy - Report 2008, Geneva: UNDP/ UNCTAD. Wiesand A., Söndermann M., (2005), The “Creative Sector” - An engine for diversity, growth and jobs in Europe, Amsterdam: European Cultural Foundation. Wu, W. (2005), “Dynamic Cities and Creative Clusters”, World Bank Policy Research Working Paper 3509, February. 108 ESSAY 4 ___________________________________________________________ The determinants of spatial location of creative industries start-ups: Evidence from Portugal using a discrete choice model approach ____________________________________________________________ 109 The determinants of spatial location of creative industries start-ups: Evidence from Portugal using a discrete choice model approach Abstract This paper assesses the location determinants of the newly created firms in the creative sector within the framework of Discrete Choice Models. Estimations using a conditional logit model, which incorporate spatial effects of neighbouring regions in the location choices of firms, yield the following results: i) the concentration of creative and knowledge-based activities, due to agglomeration economies, play an important role in location decisions of new creative establishments; ii) in contrast, the concentration of service-business activities has a negative impact on location choices, which may be due to the fact that creative firms privilege interdependencies with other activity sectors, such as innovation/ knowledge-based activities; iii) creative firms tend to favour a diversified industrial tissue and related variety, in order to enjoy from inter-sectorial synergies; iv) higher education at a regional level has a highly significant, positive effect on location decisions, while lower educational levels of human capital negatively affect those decisions, explained by the specific requirements that creative firms usually have of a highly skilled labour force; v) tolerant/ open environments attract creative activities; vi) creative firms tend to favour municipalities where the stock of knowledge and conditions for innovative activity are higher. Location decisions of creative firms also vary according to the creative sector they belong to and to their own characteristics, firm’s educational level or technologyintensity. Finally, municipality attributes are more important in terms of firms’ location decisions than the characteristics of nearby regions. Keywords: Spatial economics; industrial location; econometric models; creative industries. JEL codes: C01, R12, R30. 110 1. Introduction It is widely documented that firms tend to co-locate and that industrial agglomeration leads to localization economies (e.g., Marshall, 1890/1920; Hoover, 1937; Krugman, 1991; Fujita and Thisse, 2002; Devereux et al., 2004; Ellison et al., 2007; Arauzo-Carod and Viladecans-Marsal, 2009). For over a century, since the seminal study of Marshall (1890/1920) with the definition of spatial agglomeration economies (externalities deriving from the clustering of firms in space), researchers have studied the location behaviour of economic activities and the major reasons explaining geographical patterns of the industrial activity. The empirical literature on the determinants of industrial location (e.g., agglomeration economies, human capital, taxes, wages) has increased in recent decades (e.g., ArauzoCarod and Manjón-Antolín, 2004; Arauzo-Carod and Viladecans-Marsal, 2009; AlamáSabater et al., 2011; Guimarães et al., 2011; Arauzo-Carod, 2013). Two different approaches have been used in terms of modelling the location choices. One is focused on the choice behaviour of the firm/ agent (e.g., Arauzo-Carod and Manjón-Antolín, 2004; Alamá-Sabater et al., 2011). The other puts emphasis on the perspective of the territory where the firms are to be located (e.g., Arauzo-Carod and Viladecans-Marsal, 2009; Arauzo-Carod, 2013). Discrete Choice Models (DCM) are applied when the focus is on the firm and how the respective features of the firm (firm size, industrial sector, employment) or of the territory (infrastructures, inhabitants) have an impact on location choices. If the perspective is on the region and the determinants affecting location choices are studied in terms of firm entries on the region, then Count Data Models (CDM) are employed (Arauzo-Carod et al., 2010). These modelling techniques have been mainly used for estimating the location patterns of manufacturing industries (e.g., Arauzo-Carod and Viladecans-Marsal, 2009; ManjónAntolín and Arauzo-Carod, 2011; Alamá-Sabater et al., 2011; Liviano and ArauzoCarod, 2012; Arauzo-Carod, 2013). The study of location patterns of creative industries has mostly been comprised of exploratory analyses using the region as the unit of analysis (e.g., Lazzeretti et al., 2012; Miguel-Molina et al., 2012; Bertacchini and Borrione, 2013; Boix et al., 2013; Lazzeretti, 2013). Although such studies refer to the importance of studying the location determinants of creative activities, the modelling of their location behaviour using 117 quality of life (Arauzo-Carod and Manjón-Antolín, 2004; Arauzo-Carod and Viladecans-Marsal, 2009). In contrast, on the location determinants of industrial establishments in all manufacturing industry sectors, across all the municipalities of Murcia, Spain, a significant positive effect of human capital - measured by the percentage of labour force that has completed secondary and tertiary level education - is described by Alamá-Sabater et al. (2011), who conclude that the role of highly skilled workers on firms’ location decisions is important. Also Manjón-Antolín and ArauzoCarod (2011), on their analysis of new and relocated establishments in all manufacturing industry sectors (from highto low-technology sectors) in Catalan municipalities, find a significant positive effect of human capital (percentage of population working in science and technology/ percentage of graduates with a university degree in population over 25-years old) on start-ups’ location choices. In turn, Liviano and Arauzo-Carod (2012), using a database comprising medium-to-low technology firms of the natural-resource and manufacturing industry sectors across Catalonian municipalities, find a negative effect of human capital (measured by the average years of schooling of the population over twenty-five years of age) on firms’ decisions, which arguably might be explained by lower requirements for highly-skilled human capital, as in Arauzo-Carod and Manjón-Antolín (2004) and Arauzo-Carod and Viladecans-Marsal (2009). Addressing the issue of sector/ industry characteristics more explicitly, suggested to some extent in Arauzo-Carod and Viladecans-Marsal (2009), in his location study of manufacturing firms in Catalonian municipalities, Arauzo-Carod (2013) demonstrates that the requirements of human capital are industry-specific, and only in the case of high-tech firms, the human capital in the region - measured by the number of individuals with higher education relative to the number of jobs - has a significant positive effect on firms’ location choices. Also, the residence region of the highly-skilled workers/ human capital may not coincide with the place where the firms are located. This mismatch is explained by the preference of the workforce to live neighbouring regions, which leads to spatial lags of human capital (Alamá-Sabater et al., 2011; Arauzo-Carod, 2013). 118 Table 4. 2: Location determinants and respective effects in empirical literature: human capital Human capital Statistical Effect Authors/ Study Territorial perspective Average years of schooling of the population over twentyfive years of age: negative effect on the entry of new firms. Liviano and ArauzoCarod (2012) Percentage of labor force with secondary and tertiary education by municipality: positive, statistically significant, most important effect. Alamá-Sabater et al. (2011) Percentage of population working in science and technology// % of population with a university degree// average years of education of population over 25 years old: statistically significant, positive effects on the location of start-ups. Manjón-Antolín and Arauzo-Carod (2011) Human Capital (number of people with medium and high levels of education per km2): negative coefficient. Arauzo-Carod and Manjón-Antolín (2004) Industry/ sectorial perspective Human-capital variables (Nº individuals in each degree of educational attainment relative to nº jobs (illiterate // incomplete primary// primary education// middle school// technical high school// high school// intermediate university degree// advanced university degree): non-significant effects. Human-capital/ Highly skilled labour: Only for high-tech firms, there is a positive effect (human capital is an industryspecific factor). Arauzo-Carod (2013) Spatially lagged human-capital variables: some significant and positive effects. Human-capital Intermediate level (percentage of the population with complete secondary school): significant, positive effect on firms in all industries. Human-capital University level (percentage of the population with a university degree): significant negative impact for firms in intermediate and low-technology industries. Arauzo-Carod and Viladecans-Marsal (2009) Thus, empirical studies show negative, positive, mixed or non-significant effects of human capital on firms’ location decisions, largely depending on the database or on the measure of human capital that is used. It is also suggested that, besides considering the role of human capital as an attribute of regions, it is important to take into account the industry-specific and firm-level characteristics - in terms of knowledge-base, employees’ skills and educational level of the labour force - when analysing the impact of human capital on firms’ location choices. Given these considerations, we present a second hypothesis as follows: H2a. The region’s human capital is positively related to creative firms’ location choices. H2b. Human capital existent in each creative firm is related to its location choices. Tolerance Tolerance can be also considered as a key location determinant, since higher receptivity to newcomers, new influences and lifestyles are likely to attract creative firms to a 119 particular region (Florida, 2002, 2005; Florida et al., 2008). Although this factor is not usually considered in location models, recent research on the geography of creative industries acknowledges the importance of institutional and tolerance-related variables on the analysis of these firms’ location behaviour (e.g., Hansen, 2007; Florida et al., 2008; Lazzeretti et al., 2012; Mellander et al., 2013). Specifically, it is found that large urban centres are more likely to have a tolerant atmosphere, characterized by their openness to racial and sexual minorities as well as to other nationality groups/ foreigner people/ immigrants. This openness promotes a diversified local social network, where trust and social capital increase the effectiveness of relationships (Florida, 2002, 2005). On a study on location determinants for the creative class and regional development across all U.S. metropolitan areas, Florida et al. (2008) proved that tolerance (proxied by gay and bohemian indexes) allows for a higher accumulation of human capital and creative workers, complementary skills embodied in the immigrants, and artistic networks as channels of information among firms/ industries in the region. Thus, the more tolerant a region is the more favourable it will be to an open business climate characterized by urbanization economies, positively affecting the location decisions of creative firms and creative workers (Jacobs, 1969; Florida et al., 2008). Given these arguments, the third hypothesis is established as: H3. The region’s tolerance is positively related to creative firms’ location choices. Technology Technological endowments (facilities, provisions, firms, products, networks) represent an important factor of firms’ location patterns, particularly for knowledge-intensive and creative firms (Florida, 2002, 2005), given the role of localized, shared knowledge in the development of innovative and creative activities. As innovations and the outcomes of technological/ R&D facilities tend to spread locally, mainly due to aspects such as trust and reciprocity characterizing the networks where local knowledge is transferred (Feldman, 2000), technology provisions are a critical asset in promoting an environment where externalities arise in the form of tacit knowledge and encourage the creation of further knowledge/ innovative activities (Audretsch et al., 2007). There is a wide corpus of empirical literature corroborating the relation between technology, knowledge and the spatial clustering of firms and industries (e.g., Jaffe et al., 1993; Audretsch and Feldman, 1996; Tödtling et al., 2004; Autant-Bernard, 2006; 120 Audretsch et al., 2007). The mechanisms behind the relationship between technological endowments and the geographical clustering of firms are related to the ways through which local knowledge is diffused (Tödtling et al., 2004). Knowledge spillovers arise from labour mobility, local buzz, social networks, regular firms’ inter-relations, face-toface contacts, spinoffs or innovation joint projects, among others (Feldman, 2000; Audretsch et al., 2007). These spillovers explain the findings of Jaffe et al. (1993) on their study on the geographic location of patent citations and their spatial flows across the metropolitan areas of U.S. states, where the authors conclude that knowledge created at a regional level tends to be highly localized and stimulates the accumulation of additional knowledge in the same territorial unit. Likewise, on the geography of innovative activities across all U.S. states, Audretsch and Feldman (1996) discover that industries where knowledge spillovers (through industry innovations/ university research/ skilled labour) are more important show a higher tendency for the spatial clustering of innovative activities than other industries for which knowledge externalities are less significant. Allowing a deeper understanding of the mechanisms through which local knowledge is transferred, Tödtling et al. (2004) undertake a firm survey in Austria, comprising the manufacturing medium-tech sectors, high-tech industries, knowledge and innovation-based services and research firms, among others. The authors conclude that in the case of manufacturing and knowledge and innovationbased services, knowledge is mainly transferred through supplier-buyer relationships/ markets, informal interactions and expert/ labour mobility. In high-tech firms, there is a particular relevance for research projects, formal networks, R&D joint collaboration and consultancy as mechanisms of knowledge exchange. Research firms make more use of explicit/codified knowledge such as scientific patents, formal contracts and research collaboration. Also proving the spatial clustering of knowledge activities is the study of Autant-Bernard (2006) on the location determinants of research and development firms/ labs across all regions of France, where the stock of knowledge available in the region (proxied by private R&D expenditures of the other labs located in the region), as well as the presence of knowledge spillovers (spatial lag of those expenditures) have significant positive effects on research labs’ location decisions. These findings are also described in the study of Audretsch et al. (2007) on the location determinants of 75 German planning regions, where it is concluded that R&D facilities/ headquarters tend to concentrate in urban centres characterized by knowledge diversity, creativity and a business climate receptive to the creative innovation. 121 As shown in the empirical studies, the presence of a network of interdependent hightech/ knowledge-based firms promotes the development of local innovation processes and encourages the transmission of knowledge, new ideas and patents (Tödtling et al., 2004). This ultimately leads to growth of the region, which attracts even more knowledge-based and creative capital, given that the industries that most rely upon this asset tend to locate where their potential might be reinforced (Florida, 2002, 2005; Audretsch et al., 2007). Besides the role played as a territorial determinant (reflected, for instance, by a region’s research and development investments/ number of patents created/ density of high-tech firms), technology can be also considered as an industry-specific factor (high, medium and low-technology industries), which affects creative firms’ location choices. In this line of reasoning, the fourth hypothesis is set as follows: H4a. The region’s technological endowments are positively related to creative firms’ location choices. H4b. Industry technological intensity is related to creative firms’ location choices. Inter-territorial spillovers The benefits for firms locating in a particular region may be affected by the characteristics of surrounding locations. Inter-territorial spillovers are the effects that territory-specific (economic, social, cultural, geographic) attributes of neighbouring regions may have on a particular location. They have been recently studied and appear to be relevant in industrial location choices (e.g., Autant-Bernard, 2006; Arauzo-Carod, 2007; Alamá-Sabater et al., 2011; Guimarães et al., 2011). Indeed, there are flows characterized by supplier-buyer linkages, company interactions, industry interdependencies, labour/ human capital mobility, intellectual/ knowledge spillovers, which not only explain the (co)agglomeration patterns within each region, but also occur beyond the established frontiers of each territorial unit, with an influential effect on firms’ location choices (Autant-Bernard, 2006; Ellison et al., 2007; Alamá-Sabater et al., 2011). For instance, firms may get benefits from locating near regions (e.g., large urban centres) with large consumer markets, intensive production linkages, high population density, human capital, supplier and distribution chains, but may choose to avoid those territorial areas because of congestion effects. In these cases, the attributes 122 of nearby regions have a significant positive effect in firms’ location choices (ArauzoCarod, 2007). Despite the importance of neighbouring effects, to the best of our knowledge this issue has not yet been specifically addressed in the empirical literature on the location of creative industries. Inter-territorial spillovers are reflected in spatial autocorrelation, which occurs when the observations of a variable at a particular region are partially correlated with the variables of neighbouring locations (Arauzo-Carod, 2007). From this perspective, location choices are not only affected by the attributes of the chosen territory but may also depend on the characteristics of nearby areas. This is analysed by Autant-Bernard (2006), on the study of regional determinants of R&D labs/ firms across the French NUTS 2 regions, where it is proved that spatial knowledge spillovers, proxied by the spatially-lagged term of private R&D expenditures, exert a significant positive effect in R&D labs/ firms’ location decisions. The author concludes that the selection of a particular region is not only influenced by the relative stock of knowledge present in the region but also by that of nearby regions. The significance of inter-territorial spillovers is also observed in Alamá-Sabater et al. (2011) on the location factors of 8,429 industrial establishments in the 45 municipalities of Murcia, Spain. Their findings show that spatial spillovers have a significant impact on firms’ location decisions, with a declining effect as municipalities become more distant. In fact, the authors find that the attributes of neighbouring regions have a similar impact as those of the chosen municipality in firms’ location decisions. This is due to the presence of spatial dependence effects, which become more important when the analysis is undertaken at a more disaggregated level (e.g., municipalities, local metropolitan areas) and there is a sharing of economic, socio-cultural, infrastructural/ connectivity and other territorial aspects among neighbouring regions (Arauzo-Carod, 2007; Alamá-Sabater et al., 2011). Applying to employment data by industry/ establishment of manufacturing industry sectors across U.S. states/ counties, Guimarães et al. (2011) incorporate spatial neighbouring effects in measures of industrial concentration, 28 and conclude in support 28 The authors develop a spatially weighted Ellison-Glaeser index accounting for the spatial neighbouring effects, which offers more detailed information in measuring spatial economic concentration than popular measures of localization such as Gini, Herfindhal and common Ellison-Glaeser indexes that only consider 123 of the improvements obtained in the spatially-weighted index when compared to the original corresponding measure. In this line of argumentation, we hypothesise that: H5. Inter-territorial spillovers of neighbouring regions explain creative firms’ location choices. 3. Methodology 3.1. Data considerations The data comprises all (369) creative start-ups or new establishments created in 2009, 29 in all the creative industries, distributed across all 308 Portuguese municipalities. The source of the data is the Linked Employer-Employee Databases of GEE/ ME, Portugal. It covers all employment in industries and establishments operating in the national territory with at least one employee, excluding Public Administration and selfemployment. 30 According to the latest data available (2009), national employment in the private, structured sector totalled 3,128,126 workers, operating in a total of 407,235 establishments in all the activity sectors. Although in 2009 a total of 12,246 creative establishments ran businesses in Portugal, we had to restrict our analysis to the newly created establishments in order to avoid any endogeneity effects between firms’ location choices and the determinants of such choices. Nine major creative industries were considered for the analysis - Advertising and Marketing; Architecture; Design; Film, Video and Photography; TV and Radio; Music/ Entertainment and the Performing Arts; Publishing; Software and Digital Media; and Research (cf. Table 4.3). the information inside each pre-defined territorial unit. Besides all the information within the limits of each geographical unit, these authors’ index includes the spillovers that lie outside the boundaries of each territory. 29 This is the latest data available at the time of this study (June 2014). Courtesy of GEE/ ME, Gabinete de Estratégia e Estudos, Ministry of Economy, Portugal (Quadros de Pessoal, Linked EmployerEmployee Databases). 30 Further implications on the aspects of this database are discussed in Cruz and Teixeira (2013). 124 Table 4. 3: Creative industry sectors - mapping the creative startups/ new establishments (n=369) Core Creative sectors Industries Portuguese CAE - Rev. 3 Industry codes (compatible with ISIC - Rev. 4 codes) 1. Advertising and Marketing Advertising; Market research/ public opinion polling 7311; 7312; 7320 2. Architecture Architectural activities 7111 3. Design Design activities 7410 4. Film, Video and Photography Motion picture, video and television production, postproduction, distribution and projection activities; Photographic activities 5911; 5912; 5913; 5914; 7420 5. TV and Radio Radio activities; Television activities 6010; 6020 6. Music/ Entertainment and the Performing Arts Sound recording/music publishing activities; Performing arts; Support activities to performing arts; Artistic and literary creation; Operation of arts facilities; Amusement/ recreation activities 5920; 9001; 9002; 9003; 9004; 9321; 9329 7. Publishing Publishing of books, periodicals/ others; Translation/interpretation activities; Libraries/archives/ museum activities; Information service activities (news agencies) 5811; 5812; 5813; 5814; 5819; 7430; 9101; 9102; 9103; 9104; 6391; 6399 8. Software and Digital Media Software publishing; Computer programming/ consultancy; Data processing/hosting/Web portals 5821; 5829; 6201; 6202; 6203; 6209; 6311; 6312 9. Research Research on natural sciences, engineering, social sciences and humanities 7211; 7219; 7220 Note: For a detailed account of the relevant creative industries see Cruz and Teixeira (2014). Given that our purpose includes the testing for neighbourhood effects on creative firms’ location behaviour, through the use of spatially-lagged explanatory variables, in order to account for the spatial dependence among regions, the most suitable territorial unit of analysis is the municipality - as is shown in most recent empirical literature (e.g., Alamá-Sabater et al., 2011; Liviano and Arauzo-Carod, 2012; Arauzo-Carod, 2013). 3.2. Location determinants: variables selected and respective indicators In order to account for the location economies and to capture the benefits from the colocation of creative firms with interdependent activities/ knowledge-based firms, we used a standard measure, which is usually applied in the empirical literature for its analytical tractability (e.g., Alamá-Sabater et al., 2011; Miguel-Molina et al., 2012; Lazzeretti et al., 2012) - the location quotient (LQ) 31 (see Table 4.4). Based on the employment by industry sector in each region, we calculated the LQ in all the 31 The LQ captures the degree of specialization in a given industry, for each region, in comparison with the national average in that industry. 125 municipalities for: i) creative firms (LQ Creative firms), service-based firms (LQ Service firms); iii) knowledge-based activities (LQ Knowledge firms). Regarding urbanization economies, we used a traditional proxy describing the effects of urban agglomeration, Population Density (e.g., Arauzo-Carod and Viladecans-Marsal, 2009; Arauzo-Carod, 2013), which is robust to differences in land surface sizes and allows control for urban scale economies deriving from populated regions (Melo et al., 2009). To account for the industrial mix and the external economies transversal to all firms/ industries, we computed indexes based on the Herfindahl-Hirschman Index, usually adopted by the extant empirical research on industrial location (e.g., ArauzoCarod and Manjón-Antolín, 2004; Alamá-Sabater et al., 2011; Manjón-Antolín and Arauzo-Carod, 2011; Liviano and Arauzo-Carod, 2012): Index of industrial diversity (Industrial Diversity) and Index of creative industries’ diversity (Creative Diversity), for all 308 municipalities (cf. Table 4.4). Then, the variables LQ Creative firms, LQ Service firms, LQ Knowledge firms, Population Density, Industrial Diversity and Creative Diversity were included in our model to test Hypothesis 1 (”Agglomeration economies are positively related to creative firms’ location choices”). To examine the implications of Hypothesis 2a. (“The region’s human capital is positively related to creative firms’ location choices”), human capital variables at the municipality level - graduates of higher education human capital, measured by the percentage of population with a completed degree (Higher Education) and intermediate human capital, proxied by the gross enrolment rate in upper secondary education (Secondary Education) - were incorporated in the model. Since human capital is also a firm-level asset, we also considered the average educational attainment of the workers in each of the firms in our database, to test for the Hypothesis 2b. (“Human capital existent in each creative firm is related to its location choices.”). Following Florida (2002, 2005) and Lazzeretti et al. (2012), tolerance-related indicators include local cultural amenities (Culture) proxied by the number of museums and recreational facilities by municipality, immigrant legalization rate (Foreigners), and a social inequality ratio (Social Inequality) (cf. Table 4.4), with the aim of checking the Hypothesis 3 (“The region’s tolerance positively affects creative firms’ location choices”). 126 To test Hypothesis 4a. (“The region’s technological endowments are positively related to creative firms’ location choices”), technology endowments at a regional level are proxied by the proportion of business research and development expenditures in regional gross domestic product (R&D Firms), in line with Autant-Bernard (2006). In each region, technology is usually proxied in terms of R&D expenditures (in total turnover), R&D workers (in total workers), or patents owned (e.g., Jaffe et al., 1993; Audretsch et al., 2007). We opted for not including patents (‘codified’ knowledge), as R&D private investments more properly capture all the localized knowledge, ‘tacit’ and ‘codified’ (Autant-Bernard, 2006), that is likely to be incorporated in the innovation process of creative firms (Florida et al., 2008). At the industry level, and in order to test for the Hypothesis 4b. (“Industry technological intensity is related to creative firms’ location choices”), we categorize the industries/ firms in terms of their technology intensity: very high, high, medium-high and mediumtechnology. The neighbouring effects in firms’ location decisions are analyzed by introducing spatially-lagged explanatory variables in the model, calculated on the basis of spatialweights matrices (e.g., Alamá-Sabater et al., 2011). We carry out this analysis by constructing a spatially-lagged model, composed of the explanatory variables and their respective spatial lags, for the purpose of testing Hypothesis 5 (“Inter-territorial spillovers of nearby regions explain creative firms’ location choices”). All the variables selected and respective indicators are presented in Table 4.4. Given that the firm micro-data available comprises all the new creative establishments of the year 2009, each indicator computed for the analysis of regional location determinants refers to 2008 and 2009 or earlier periods, to best describe the existing conditions at the time that those establishments were created. 133 unrestricted CLM, with all the explanatory variables, is suitably specified when compared to the alternative restricted model. All coefficients are statistically significant (at one, five and ten percent levels), most of them highly significant (at one percent level). From Table 4.5, and similar to the results obtained by the bulk of research on the location of manufacturing industries (cf. Section 2), it is noticeable that (co)location economies play an important role in creative firms’ location decisions. The concentration of creative firms (LQ Creative Firms) and the clustering of knowledgebased activities (LQ Knowledge Firms) are statistically significant and exert a positive effect on the decisions of creative establishments. There is enough evidence to maintain that creative firms tend to locate where other creative and knowledge-based activities are clustered, suggesting co-location among these sectors/ activities, due to potential interdependencies and local synergies. In contrast, the concentration of service-business activities (LQ Service Firms) has a negative impact on choices. This is a similar result to that obtained in Alamá-Sabater et al. (2011) and may derive from the fact that in large urban centres, service-based activities are not so highly concentrated, or that creative firms privilege interdependencies with other activity sectors such as innovation/ knowledge-based activities. It is mostly in inland/ remote municipalities that services (e.g., health, accountancy or legal activities) usually have more relative importance at a local level. Regarding urbanization economies, population density (Population Density), denoting externalities from urban agglomeration, has a significant, positive effect in firms’ location decisions, suggesting the tendency of creative establishments to locate near large consumer markets. In terms of regional industrial mix, estimated coefficients for the diversity indexes of all the activity sectors (Industrial Diversity) and of creative industry sectors (Creative Diversity) have positive, significant impacts in location choices. This evidence suggests that creative firms tend to favour a diversified industrial matrix both in terms of all the industrial sectors and of the mix of creative industries, substantiating the argument in Lazzeretti et al. (2012) that creative firms privilege local related variety in order to benefit from inter-sectorial, transversal synergies. 134 Table 4. 5: Standard CLM estimates (n=369 cases/ creative establishments; j=308 alternatives/ municipalities) Hypotheses Variable/ Location Determinant Estimated Coefficient Standard Error H1. Agglomeration (location and urbanization) economies Population Density 0.0001* 0.0000541 LQ Creative Firms 1.053*** 0.3500657 LQ Service Firms -0.769** 0.3957541 LQ Knowledge Firms 0.775*** 0.280637 Industrial Diversity 0.141*** 0.0377678 Creative Diversity 11.726*** 2.530715 H2. Human Capital Higher Education 0.206*** 0.0245704 Secondary Education -0.005*** 0.0009556 H3. Tolerance/ Openness Culture -0.033*** 0.0108462 Foreigners 0.268*** 0.0597044 Social Inequality -0.125*** 0.0205913 H4. Technology R&D Firms 1.198* 0.6898952 Log-likelihood -1566.4406 Wald chi2(12) (joint significance of the variables in the model) 1228.17 [Prob > chi2 = 0.0000] Pseudo R2 0.2592 Nr. Observations 113,652 Likelihood-ratio (LR)test – Unrestricted with all variables vs restricted (measure of fit for CLM specification) LRfull/ restricted = 460.59 [Prob > chi2 = 0.0000] ***, **, * one, five and ten percent significance levels, respectively. Source: Authors’ computations based on STATA 13 ® and micro-data from the Linked Employer-Employee Databases, GEE/ ME, Portugal (year 2009). From this it is possible to conclude that the effects of traditional location factors - location and urbanization economies - support the empirical literature due to benefits arising from industry-specific (creative sectors) clustering, urban agglomeration and due to externalities transversal to all co-located firms/ industries, which validates our Hypothesis 1 (H1) that agglomeration economies are positively related to creative firms’ location choices. Regarding human-capital estimates, it is noticeable that higher education at a regional level (Higher Education) plays a statistically highly significant and positive effect in creative firms’ location decisions. A unit increase in this factor leads to a positive increment of 23% (e0.206) on the odds of locating at a particular municipality versus all the other alternative locations. In turn, lower educational levels, such as upper 135 secondary schooling rate (Secondary Education), have a negative, statistically significant effect. These facts are in overall accordance with the empirical/ exploratory research on the location of creative industries (Florida, 2002, 2005; Florida et al., 2008; Lazzeretti et al., 2012), contrasting with results obtained by studies (e.g., Arauzo-Carod and Manjón-Antolín, 2004; Arauzo-Carod and Viladecans-Marsal, 2009; Liviano and Arauzo-Carod, 2012) on the location of medium-to-low technology manufacturing firms (cf. Table 4.2), which can be explained by the specific requirements that creative firms usually have of a highly skilled labour force. These findings validate the implications of Hypothesis 2a. that the region’s higher education human capital is positively related to creative firms’ location decisions. Concerning tolerance-related variables (institutional factors), we observe a positive, significant impact of immigrant legalization rate (Foreigners), denoting openness to immigrants/ newcomers, in location decisions. These decisions are negatively affected by the existence of social inequalities (Social Inequality) in the municipality, which finds support in the empirical literature that tolerant/ open environments are a locus of creative activities (Florida et al., 2008). The coefficient for cultural amenities (Culture) is significant and negative, which can be due to the fact that museums, libraries and cultural facilities are spread across inland and coastal municipalities, and are much more related with heritage and historical sites than with the contemporary art facilities, usually found in large metropolises as mentioned by Florida (2002, 2005). Summing up, although cultural infrastructures repel creative firms, their location choices favour more tolerant and equal environments, where openness to newcomers and less social inequality are present. This evidence partially confirms Hypothesis 3. Finally, the estimate for regional technological endowments (R&D Firms) shows a positive and significant coefficient, which corroborates the empirical literature (e.g., Autant-Bernard, 2006; Audretsch et al., 2007) that creative firms tend to favour municipalities where the stock of knowledge (developed by private firms) and the conditions for the innovative activity are higher. This confirms our Hypothesis 4a, that a region’s technological endowments are positively related to creative firms’ location choices. Since human capital is not only an attribute of the region but also of the firm, in order to test H2b., we estimate the baseline model for three groups of firms (cf. Table 4.6): firms with high educational levels (Model I), those with intermediate levels (Model II) and 136 ones with basic levels (Model III). Depending on the type of firm (with high, intermediate or basic educational level), the location determinants differ. This means that Hypothesis 2b, postulating that the human capital/ educational level existent in creative firms is related to their location choices, is validated. Specifically, firms with higher educational levels tend to favour location determinants such as (co)location economies (LQ Creative Firms; LQ Knowledge Firms) and withinindustry variety (Creative Diversity) in order to take advantage of complementary linkages; higher education/ graduate human capital (Higher Education) in opposition to lower educational levels; tolerant environments (positive significant effect of Foreigners; negative significant sign for Social Inequality), and local innovation (R&D Firms) in their location choices (cf. Table 4.6, Model I). These factors generally describe creative firms’ location determinants in the empirical literature, and they are usually found in large urban centres (Florida et al., 2008; Lazzeretti et al., 2012). In turn, creative establishments with intermediate and basic educational levels tend to privilege more industrial diversity and not clustering with complementary creative/ knowledge industries; human capital (Higher Education), particularly evident in the case of intermediate-level firms; and institutional factors (positive, significant effect of Foreigners; significant negative impact of Social Inequality) for both types of firms (cf. Table 4.6, Models II and III). Creative firms with higher educational levels are most likely to portray intellectual property activities which require a highly skilled labour force, and they are usually colocated with other innovative/ knowledge-intensive firms (e.g., Advertising and Marketing; Software and Digital media; Research), whereas establishments with intermediate/ basic educational levels, more concerned with leisure, entertainment and artistic activities (e.g., Film, Video and Photography; Music/ Entertainment and Performing arts), mainly tend to privilege industrial and socially diversified environments. 137 Table 4. 6: Standard CLM estimates according to the educational level (high, intermediate, basic) in creative establishments (n=369; j=308 municipalities) Location Determinants Explanatory Variables Standard CLM Estimated Coefficients Model I - High educational level establishments Model II - Intermediate educational level establishments Model III - Basic educational level establishments Agglomeration economies Population Density 0.0002** (0.0000705) 0.00001 (0.0001117) -0.00009 (0.0001497) LQ Creative Firms 0.946** (0.4927382) 0.920 (0.6580498) 1.831* (0.7637896) LQ Service Firms -0.692 (0.5516932) -1.286* (0.7615708) -0.034 (0.8838747) LQ Knowledge Firms 0.903** (0.3808698) 0.584 (0.5456962) 0.918 (0.6657408) Industrial Diversity 0.096* (0.0517234) 0.180*** (0.0698426) 0.239*** (0.0929636) Creative Diversity 12.074*** (3.563947) 10.376** (4.731422) 13.902** (5.86167) Human Capital Higher Education 0.231*** (0.0315043) 0.247*** (0.0472813) 0.028 (0.0687083) Secondary Education -0.005*** (0.0012599) -0.005*** (0.0018979) -0.002 (0.0024325) Tolerance/ Openness Culture -0.041*** (0.0143194) -0.014 (0.0228775) -0.041 (0.0276566) Foreigners 0.178* (0.093723) 0.417*** (0.0991971) 0.226* (0.1333904) Social Inequality -0.084*** (0.0295553) -0.168*** (0.0381912) -0.176*** (0.0459044) Technology R&D Firms 1.717* (0.942001) -0.431 (1.467819) 2.207* (1.33419) Nr. Observations / Cases 65,604 obs./ 213 cases 30,492 obs./ 99 cases 17,556 obs./ 57 cases ***, **, * one, five and ten percent significance levels, respectively. Standard Errors in brackets. Source: Authors’ computations based on STATA 13 ® and micro-data from the Linked Employer-Employee Databases, GEE/ ME, Portugal (year 2009). Technology is also a characteristic of the industry sector to which a firm belongs. Thus, in order to test for the Hypothesis 4b, we estimate four models according to the technological intensity of the industry to which the creative establishment belongs: ‘very high-tech’ (Model I), ‘high-tech’ (Model II), ‘medium-to-high tech’ (Model III), and ‘medium-tech’ (Model IV) (cf. Table 4.7). 138 Table 4. 7: Standard CLM estimates according to technological intensity of creative firms (n=369; j=308 municipalities) Location Determinants Explanatory Variables Standard CLM Estimated Coefficients Model I – Very-High tech creative firms Model II - High-tech creative firms Model III - Medium-Hightech creative firms Model IV - Medium-tech creative firms Agglomeration economies Population Density 0.0001 (0.0000939) 1.22e-06 (0.0000814) 0.0003** (0.0001481) 0.0002 (0.0002621) LQ Creative Firms 0.783 (0.6313214) 1.341*** (0.5018942) 0.149 (0.9689093) 2.014 (1.500396) LQ Service Firms -0.396 (0.767213) -0.545 (0.5837014) -1.617* (0.913261) 0.200 (1.882107) LQ Knowledge Firms 1.073** (0.4863913) 1.159*** (0.4017348) -1.563* (0.8547248) 0.897 (1.460686) Industrial Diversity 0.2786*** (0.0933306) 0.157*** (0.0581652) 0.146* (0.0847599) -0.052 (0.0784532) Creative Diversity 10.057** (4.575143) 13.278*** (3.625556) 5.233 (7.408923) 24.362* (12.0456) Human Capital Higher Education 0.223*** (0.0422254) 0.183*** (0.0366901) 0.227*** (0.0639556) 0.290** (0.1453894) Secondary Education -0.007*** (0.0017053) -0.004*** (0.0013831) -0.00009 (0.0023451) -0.008 (0.0049457) Tolerance/ Openness Culture -0.036* (0.0193225) -0.030* (0.0162479) -0.013 (0.0294751) -0.075 (0.0474394) Foreigners 0.316*** (0.1104592) 0.275*** (0.0828753) 0.359*** (0.1454878) -1.639 (1.343929) Social Inequality -0.129*** (0.0418044) -0.145*** (0.031077) -0.015 (0.0478961) -0.227*** (0.0859189) Technology R&D Firms 1.251 (1.28737) 0.616 (1.080231) 1.965 (1.564127) 4.829** (2.397027) Nr. Observations / Cases 37,576 obs./ 122 cases 54,208 obs./ 176 cases 15,400 obs./ 50 cases 6,468 obs./ 21 cases ***, **, * one, five and ten percent significance levels, respectively. Standard Errors in brackets. The division in terms of technology-intensity was made following the taxonomy of Silva and Teixeira (2011). Source: Authors’ computations based on STATA 13 ® and micro-data from the Linked Employer-Employee Databases, GEE/ ME, Portugal (year 2009). 139 In the case of very high and high-technology creative establishments, agglomeration economies due to the co-location with creative and knowledge-based firms (LQ Creative Firms; LQ Knowledge Firms); urbanization economies from related variety (Industrial and Creative Diversity); higher levels of human capital (Higher Education); and institutional factors of tolerance (Foreigners; Social Inequality) play important roles as location determinants. In the case of medium-to-high and medium-technology creative establishments, decisions are mainly affected by human capital (Higher Education) and institutional tolerance-related factors (Foreigners; Social Inequality). Moreover, these firms avoid or are indifferent to the co-location with creative/ knowledge-based activities, as shown by the sign and significance of LQ Creative Firms and LQ Knowledge Firms (cf. Table 4.7, Models III and IV). This provides evidence for different patterns of location behaviour according to the technology-level of creative firms, which validates our Hypothesis 4b. Finally, in order to account for the inter-territorial spillovers of neighbouring municipalities in creative firms’ location choices (H5), we estimate an ‘enlarged’ model, adding the spatial lags of each explanatory variable in the CLM (cf. Table 4.8). It is evident from the estimates that when including the attributes of neighbouring regions, the most important determinants of creative firms’ location choices remain much the same as in the standard CLM estimations (Table 4.5). The attributes of chosen locations have a significant effect on firms’ decisions while those of nearby regions only show significance for the case of Secondary Education_spl and the institutional factor Social Inequality_spl. Here, it is possible that since upper secondary education is a variable which is widely distributed throughout the country, and social inequality is an institutional factor, their effects may extend beyond the boundaries of each municipality. In short, Hypothesis 5 (H5) is partially sustained by the data. Although it is critical to account for inter-territorial spillovers, in the particular case of our database, location behaviour is strongly shaped by municipality characteristics and not by the aspects of contiguous regions. This can be understood in that creative firms are mainly located in large or important urban centres, with an ample supply of resources (e.g., human capital, knowledge networks and technological endowments), related variety and large consumer markets, with little need to resort to resources beyond the borders of their region. 140 Table 4. 8: CLM with spatially lagged variables - parameter estimates (n=369 cases/ creative establishments; j=308 alternatives/ municipalities) Hypotheses Variable/ Location Determinant Estimated Coefficient Standard Error H1. Agglomeration (location and urbanization) economies Population Density 0.0002** 0.0001019 LQ Creative Firms 0.879** 0.4047041 LQ Service Firms -0.147 0.484064 LQ Knowledge Firms 0.785** 0.3524896 Industrial Diversity 0.068* 0.0400985 Creative Diversity 10.925*** 2.871029 H2. Human Capital Higher Education 0.165*** 0.0394276 Secondary Education -0.00007 0.0014653 H3. Tolerance/ Openness Culture -0.033 0.0219262 Foreigners 0.293** 0.1279316 Social Inequality -0.063** 0.0313443 H4. Technology R&D Firms 2.194** 0.9108805 H5. Inter-territorial spillovers of neighbouring regions Population Density_spl 0.0003 0.0002221 LQ Creative firms_spl 1.178 1.017497 LQ Service firms_spl -1.062 0.7401107 LQ Knowledge firms_spl -0.034 0.7902569 Industrial Diversity_spl 0.028 0.0544824 Creative Diversity_spl 10.645 7.551003 Higher Education_spl 0.026 0.0945354 Secondary Education_spl 0.005** 0.0025297 Culture_spl -0.002 0.0551487 Foreigners_spl -0.009 0.1638044 Social Inequality_spl -0.067* 0.03949 R&D Firms_spl -2.288 1.622236 Log-likelihood -1548.1567 Wald chi2(24) 1229.31 [Prob > chi2 = 0.0000] Pseudo R2 0.2678 Nr. Observations 113,652 Likelihood-ratio (LR) test LR full/ restricted = 497.16 [Prob > chi2 = 0.0000] ***, **, * one, five and ten percent significance levels, respectively. Source: Authors’ computations based on STATA 13 ® and micro-data from the Linked Employer-Employee Databases, GEE/ ME, Portugal (year 2009). 141 4.2. Empirical results by creative industry sector The location patterns of creative industries reveal heterogeneous characteristics across creative sector groups (Cruz and Teixeira, 2014). Thus, it is expected that creative firms’ location behaviour is differentiated according to the industry sector to which they belong. Indeed, standard CLM estimates by sector (cf. Table 4.9) indicate that creative establishments/ start-ups are affected by different combinations of location factors, depending on their industry sector. Creative firms in the sectors of ‘Advertising and Marketing’ and ‘Software and Digital media’ tend to favour regions with higher concentrations of creative and knowledgebased activities, benefiting from synergies of co-location with complementary industries and from industrial and creative diversification/ related variety; with higher human capital and tolerance/ openness, reflected by the foreigners’ acceptance rate and lower levels of social inequality. These location factors, characterizing large urban centres (such as Lisbon and Oeiras), support the arguments usually raised in the empirical literature on creative industries (e.g., Florida, 2002, 2005; Florida et al., 2008; Lazzeretti et al., 2012; Cruz and Teixeira, 2014). In turn, establishments that belong to ‘Publishing’, ‘Architecture’, ‘Design’ and ‘Film, Video and Photography’ industries, mostly located across intermediate or important urban centres in the North-Centre of the country, share some similarities in their major determinants, mostly related with industrial/ related diversity, institutional and human capital factors. In the ‘Publishing’ industry, where firms are quite dispersed across intermediate urban centres in the country’s North-Centre (around Porto, Coimbra and Lisbon), firms emphasise creative diversity, human capital and social equality as location determinants. In the ‘Design’ sector, where firms are mainly located in Northern intermediate urban centres (around Porto), creative establishments tend to favour municipalities with industrial diversity, lower concentrations of services-based firms, higher levels of human capital and lower social inequalities. The clustering of services mainly occurs in inland/ coastal/ tourism municipalities, thus the negative estimate in Design (cf. Table 4.9) might be explained by design firms’ preference to locate near relevant 142 manufacturing industries (e.g., fashion/ textiles design, furniture/ equipment design, industrial/ product design, graphic design), that are mostly located in the North of Portugal. Concerning ‘Architecture’, creative establishments favour co-location with other creative activities, creative diversity/ related variety and higher levels of human capital. These firms tend to be located in intermediate urban centres in the North-Centre municipalities (mainly around Porto). Firms belonging to ‘Film, Video and Photography’ prefer regions with lower social inequalities and higher levels of human capital. These firms are scattered all over the territory, with some prevalence around and in the two largest urban centres (Lisbon and Porto). In the ‘Research’ sector, creative establishments prefer to locate where there are high levels of human capital (higher education) and avoid municipalities with lower levels of human capital (secondary education), mainly privileging cities with universities, highereducation institutions and research centres. In contrast, firms belonging to ‘Music, Entertainment and the Performing arts’ avoid locations with higher concentrations of knowledge-based activities and reveal a preference to locate in regions with larger consumer markets/ population density and higher openness/ immigration acceptance rate (mainly tourism/ coastal municipalities).