City specialisation and diversification in South East Europe (SEE) countries

42019, XXII, 2 Economics DOI: 10.15240/tul/001/2019-2-001 Introduction Cities are considered centres of economic activity and, presumably, they remain attractive locations for manufacturing fi rms so as long as benefi ts agglomeration economies prevail over the costs of agglomeration diseconomies. Agglomeration economies attract fi rms and labour to co-locate, while agglomeration diseconomies push fi rms and labour to relocate to decentralised locations (Richardson, 1995). Industry patterns formed across urban landscape of a country or a region will largely depend on the interplay of these opposite forces, as well as on industryand fi rm-specifi c issues. The size of agglomeration and the economic structure may be interrelated and in some economies, mostly larger, patterns of city specialisation emerge. All cities are characterised by being either specialised or diversifi ed, depending on whether their economic activity is concentrated in similar or dissimilar types of production – and larger cities tend to be more diversifi ed (Duranton & Puga, 2000). An overriding problem in Urban Economics research in South-East Europe is the lack of data. In the case of ex-Yugoslav countries, we see no regular yearly statistics on cities since the 1990es. The data hindrance is probably the main reason why research in CEE is also mostly focused on regions and not on cities. Research topics in the regional context mostly range within the income convergence/ disparities issue. Results from this research can implicitly shed some light on the issue of city specialisation. Economic development of cities in the South-East Europe since the end of the World War II can be seen through two distinct economic development periods: a) The fullscale industrial development of cities. Cities expand their borders as economic migrants from all the countries’ regions join the wave of rising economic activity; b) The tertiarisation phase upon the breakdown of communism and the onset of social and economic transition to democracy and market economy. Economic planning is quickly dismissed as a relic of socialist planning and many cities are left to cope with the rising unemployment coming from the weakened industries. It must be noted that the deindustrialisation process in the metropolitan regions has started in some cities even before the transition process, in the 1980s, as documented in Croatia’s capital development (Svirčić-Gotovac, 2006). However, while western EU economies experienced urban shrinking since the 1970s, CEE countries continued to support industry and neglect service sector up to 1989 which resulted in overgrowth of industrial agglomerations (Rumpel et al., 2013). The development of metropolitan regions contrasts the development of former industrial cities. The newest indicators of development of metropolitan regions in European transition and post-transition economies available at EUROSTAT and OECD indicate that metropolitan regions have largely converged to the EU average in income per capita and other economic indicators. According to the extensive research of Central and Eastern Europe cities by Lintz et al. (2004), market forces spurred city development, but mostly in the case of metropolitan regions. Urban attraction forces of metropolises have drawn foreign investors and have further strengthen the position of the capital-cities within the national urban system. Dogaru et al. (2014) have found that capital city regions received more greenfi eld FDI and attracted a wider variety of investments in both sectors and functions. The fast growth dynamics of metropolitan regions is unmatched by former industrial or industrial & agricultural regions. To some CITY SPECIALISATION AND DIVERSIFICATION IN SOUTH EAST EUROPE (SEE) COUNTRIES Ivana Rašić Bakarić, Katarina Bačić, Sunčana Slijepčević EM_2_2019.indd 4EM_2_2019.indd 4 19.6.2019 15:10:4819.6.2019 15:10:48 5 2, XXII, 2019 Economics extent, this can explain the level of regional disparities at NUTS level III in the new member states that are signifi cantly higher than in the incumbent members: the average level of the measure of dispersion in the year 2011 in new member states is 35.1 percent, compared to 25.8 percent in incumbent members. Lintz et al. (2004) report the context of the setback of many old industrial cities that hosted important industrial plants through the lower business start-up rate of local population. Lintz et al. (2007) noticed that many old industrial cities in CEE countries, which have been drivers of growth in CEE countries for decades, have been facing major problems (socioeconomic and environmental) since the economic transition and that they have experienced a setback. It could be debated that the reliance of the local population on one or few big industries that have employed the whole city and regional working population has inhibited the entrepreneurial spirit and thus also cities’ economic recovery and further growth. New disparities between different CEE cities during transition have been dependent on location, inherited economic structure and environmental quality (Lintz et al., 2007), but also stirred by demographic changes and outmigration (Scott & Kühn, 2012). While mostly capital cities and regional centres developed (partly due to European union funds), peripheries experienced urban economic decline connected with closure of industrial enterprises (Lintz et al., 2007; Sýkora & Bouzarovski, 2011). As pointed out, results from regional research can be useful in learning more about cities in CEE. Regional development policies in European transition economies have been revived with the political and economic desire to join the EU. Evidence on regional specialisation has shown that increase in specialisation in the incumbent EU members was insubstantial (OECD, 1999; 2004), thus risks associated with greater specialisation brought by EU integration as debated by the US economist Krugman (1991), were probably illusory. Recently presented evidence for the new EU transition members has shown that in the case of Central European countries (CEC) changes in the relative regional specialisation are more dynamic than in the incumbent members (Stierle–von Schütz & Stierle, 2013). The main aim of this paper is to discern whether patterns in the relative specialisation and diversifi cation of manufacturing industry in 98 SEE (Bosnia and Herzegovina, Bulgaria, Croatia, Serbia, Slovenia and Romania) cities exist, by constructing specialisation and diversifi cation measures over the period 2006-2013. The article is structured as follows: Section 1 provides a literature overview on city specialisation issues, Section 2 describes the dataset, the sample of cities and discusses representativeness issues. In this section measures of specialisation (and diversifi cation) are applied to the dataset, with the purpose to capture the traits of cities’ economic structure. Section 3 analyses differences within manufacturing industry across cities in terms of their technological complexity and uses cluster analysis to identify groups of cities on the basis of multivariate data, namely indices of manufacturing-industry specialisation, and city size using population data. These measures enable recognising industrial patterns in urban space throughout countries and the SEE region. Section 4 offers conclusions. 1. Literature Review Specialisation of cities is an important topic in urban economics. Duranton and Puga (1999) summarize the most important fi ndings on diversifi cation and specialisation of cities, mostly coming from city research in the US, in few stylised facts (paraph., p. 2-9): “Both specialised and diversifi ed cities exist; Larger cities tend to be more diversifi ed; Individual city specialisations are stable over time; Most relocations are from diversifi ed to specialised cities.” In Henderson (1997) research on diversity and city size, he has found that larger cities (population above 500,000) in the US were more specialised in services and medium-sized cities were more specialised in manufacturing. The latter were also more specialised in mature-industries (textiles, food, pulp and paper) than in new industries on lower levels of aggregation. Duranton and Puga (2005) revealed a shift in city specialisation in the US from sectoral to functional specialisation that came as a result of disintegration of functions: management was located in larger cities and production functions in smaller cities. Davies and Henderson (2008) found that this separation is useful for headquarters because of the availability of differentiated local suppliers and because of the presence of the other headquarters. EM_2_2019.indd 5EM_2_2019.indd 5 19.6.2019 15:10:4819.6.2019 15:10:48 62019, XXII, 2 Economics Cuadrado-Roura and Rubalcaba-Bermejo (1998, p. 134) posit that specialisation is “foremost a historical fact” that “tends to follow its own life cycle” and the specialisation cycle may end if specialisation become uncompetitive. Other reasons may be changes in international dynamics or if “suffi cient size is reached to impose a different logical pattern”. Cities are undoubtedly recognised as centres of knowledge generation, diffusion and accumulation (Fujita & Mori, 2005). The literature is largely inconclusive as to whether agglomeration externalities arise between fi rms belonging to either the same or to different industries (Henderson, 1986; van Hagen & Hammond, 1994; Glaeser, Kallal, & Scheinkman, 1992), though. As put forward by Marshall (1890), Arrow (1962), and Romer (1986) knowledge spillovers may arise between fi rms within the same industry and be supported by local concentrations of a particular industry (the Marshall-ArrowRomer (MAR) type localization or ’specialization’ externalities). According to the MAR type externalities, knowledge spillovers in specialized geographically-concentrated industries stimulate growth. On the other hand, the notion that industrial diversity directly contributes to agglomeration economies comes from Jacobs (the Jacobs-Porter type externalities, or economies of diversifi cation Jacobs (1969), who argues that the most important knowledge transfers stem from variety and diversity of geographically proximate industries (Da Silva Catela, Goncalves, & Porcile, 2010). Applied to cities, the diversifi cation of an industry in a city helps knowledge spill-overs between fi rms, and therefore the growth of that city. Kolehmainen (2003) argues that the environment that agglomeration constitutes is far more complex and includes more interrelated elements to be considered, including institutional setting and behaviour of fi rms. Stojcic and Orlic (2015) point out that knowledge spillovers and cooperation between fi rms and scientifi c institutions can ultimately lead to increasing returns, drawing on NEG literature. Through trade liberalisation and EU integration processes, post-transitioning Europe, has opened to structural changes. These processes can lead to increase in specialisation across key manufacturing locations, including cities. Krugman (1991) has predicted that the removal of trade barriers and European integration will bring about more industrial specialisation (or concentration) across EU and, as a consequence, more exposure to asymmetric economic shocks. Evidence from the incumbent EU members over the last two decades was not supportive of this prediction, at least at the regional level (OECD, 2004). A more evident process of structural change can be observed in the expansion of the service sector, the tertiarisation, at least across most developed regions (Marelli, 2004). However, Longhi, Musolesi and Baumont (2014) argue that the metropolitan areas and major regional centres of larger EU countries may accumulate most benefi ts from European integration. Urban regeneration projects in some cities (e.g. Bucharest in Romania) were connected with realisation of European Union funds (Hlaváček et al., 2016). Observing these spatial units as “the European core”, with the highest probability for increasing returns to emerge through concentration of economic activities, with an econometric model Longhi, Musolesi and Baumont (2014) show how 35 metropolitan areas over the period of 19802005 have undergone structural changes due to integration and development processes. Longhi, Musolesi and Baumont’s research has shown that specialisation has increased and that sectoral structures have become more similar in services. Moreover, the integration jointly with development positively infl uence specialisation in the sense that positive effect of development on specialisation is stronger in metropolitan areas that are better integrated with EU. Stierle-von Schütz and Stierle (2013) use a comprehensive set of measures of concentration and specialisation for the period ranging from 1995 to 2010 to provide an overview of change in the economic structures in the EU, regarding integration processes as the trigger of change. Measures were applied to different indicators of economic activity at various sectoral breakdowns for all EU member states, including transition economies. The spatial unit of observation were regions. On an aggregated level, dynamics of the level of specialisation and concentration appeared slow, even when observed in the light of EU enlargement and in the post-2008 period of economic crisis. By observing structural traits by sectors, the authors found that in EU-15 and EU transition economies the concentration of some low-tech industries was higher and EM_2_2019.indd 6EM_2_2019.indd 6 19.6.2019 15:10:4819.6.2019 15:10:48 7 2, XXII, 2019 Economics the concentration of high-tech industries was lower. As expected though, new member countries’ concentration of high-tech industries is comparatively lower. In some regions and in some sectors there is indication of convergence of new member countries’ production structures to that of EU-15, most markedly in high-tech sector manufacturing. Longhi, Nijkamp and Traistaru (2005) researched effects of European integration on patterns of manufacturing location and specialisation for Bulgaria, Estonia, Hungary, Romania and Slovenia, accession countries at that time, over the period 1990-1999. Based on the presumption that economic integration will result in more effi cient allocation of resources through structural changes and adjustment, authors carry out an econometric analysis searching for the determinants of manufacturing location. The economic geography of manufacturing in these countries was more in line with the predictions of classical and new trade theories than with predictions of NEG as it was determined by factor endowments and proximity to large markets. Industry patterns formed across urban landscape of a country or a region will largely depend on the interplay of these opposite forces, as well as on industryand fi rm-specifi c issues. The size of agglomeration and the economic structure may be interrelated and in some economies, mostly larger, patterns of city specialisation emerge. Evidence on the productivity advantages that fi rms can appropriate by locating in larger cities and in more diversifi ed locations can be found in the empirical literature. Firms in larger cities are overall more productive than fi rms in smaller cities (Combes et al., 2012; Rosenthal & Strange, 2004), due to a number of reasons, including foremost the agglomeration economies, but also localised natural advantage, stronger worker and fi rm selection (Combes et al., 2012). Furthermore, productivity advantages of fi rms located in cities as more diversifi ed locations are noted over fi rms in more specialised industrial-districttype of areas (Di Giacinto et al., 2014) provide convincing evidence for Italy). However, evidence also shows that links between effi ciency and the type of agglomeration economies are to be observed in context of the countries’ development level. For example, Da Silva Catela, Gonçalves, and Porcile’s research (2010) on the relation between types of agglomeration economies (specialization vs. diversifi cation) and labour productivity for 524 Brazilian municipalities in 1997 and 2007 had shown that specialised municipalities were over-performing. Authors have used fi nitemixture regressions and municipalities have shown polarised, with the over-performing group of municipalities being signifi cantly specialised, while the opposite was found for the underperforming group. 2. City Specialisation and Diversifi cation in SEE 2.1 Dataset The main aim of this part of the paper is to discern whether patterns in the relative specialisation and diversifi cation of manufacturing industry in 98 SEE cities exist, by constructing specialisation and diversifi cation measures over the period 2006-2013. Indices that are used in measuring specialization (relative specialisation index) and diversity (relative diversifi cation index) for 98 cities in six SEE countries are constructed following the work of Duranton and Puga (2000). The empirical analysis was conducted on large dataset of 63,506 manufacturing fi rms observed over the period 2006-2013. The data employed in this analysis were obtained from the large pan-European fi rm level database Amadeus provided by Bureau van Dijk. The unit of analysis is the fi rm defi ned as a legal entity, as opposed to the establishment. The industry data were aggregated from fi rm-data obtained in Bureau van Dijk’s Amadeus database for Central and Eastern Europe for the period 2006-2013. The geographical units are cities above 50,000 inhabitants in six SEE countries (Bosnia and Herzegovina, Bulgaria, Croatia, Serbia, Slovenia and Romania). The number of cities in each country and their share of the total national population are as follows: Bosnia and Herzegovina – 12 cities or 35.6% of the 2011 population; Bulgaria – 18 cities or 44.7% of the 2011 population; Croatia – nine cities or 35.4% of the 2011 population; Romania – 38 cities or 32.8% of the 2011 population; Serbia – 17 cities or 35.7% of the 2011 population; Slovenia – four cities or 24.8% of the 2011 population. To produce the city-class-size ranges, the total population of the city area was used. The greatest number of the observations, 37.1% refers to fi rms located in large cities (cities EM_2_2019.indd 7EM_2_2019.indd 7 19.6.2019 15:10:4819.6.2019 15:10:48 82019, XXII, 2 Economics with more than 500,000 inhabitants), followed by fi rms located in cities between 100,000 and 249,000 inhabitants (24.2% of the observations) and fi rms located in cities between 50,000 and 99,000 inhabitants (Tab. 1). Tab. 1 displays the sectoral distribution of the observations in the sample. About 63% of the observations refer to four industry groups: food, beverages and tobacco industry, furniture, other manufactured goods and textiles, and to apparel and leather industry. Since data on the number of fi rms in manufacturing were not publicly available for the selected cities the total number of manufacturing fi rms in the country is resorted to as another option for assessing data coverage. The analysis of manufacturing industry diversifi cation and specialisation in the cities is based on the relation between agglomeration economies of the Marshall-Arrow-Romer type (economies of location or specialization) and the Jacobs-Porter type (economies of urbanization or diversifi cation). Henderson’s research (1997) on US cities has shown that patterns of city specialisation-size can be observed. His research shows that cities of similar sizes are often specialised in similar industries i.e. cities with population above 500,000, considered larger cities, were more specialised in services and medium-sized cities were more specialised in manufacturing. The latter were also more specialised in matureindustries (textiles, food, pulp and paper) than in new industries on lower levels of aggregation. In his research on manufacturing industries in the U.S. and Brazil Henderson did not fi nd urbanisation economies, but did fi nd evidence of localisation economies explaining that by the fact that resources in manufacturing are generally not more productive in larger cities Industry City size class Total less than 99,000 100,000249,000 250,000499,999 > 500,000 Food, beverages, tobacco 17,715 16,315 10,997 23,840 68,867 Textiles, apparel, leather 15,783 16,292 9,947 20,677 62,699 Wood, cork, paper, printing, recorded media 11,792 11,053 8,759 22,474 54,078 Coke and refi ned petroleum products 128 124 12 281 545 Chemicals, pharmaceuticals, rubber, plastic 7,943 8,336 5,647 14,308 36,234 Other non-metallic mineral products 4,196 4,111 2,239 6,184 16,730 Basic metals, metal products 14,855 15,772 9,871 19,796 60,294 Machinery and equipment 3,161 3,909 2,738 6,399 16,207 Computer, el. and optical products, el. equipment 3,469 5,490 3,317 12,056 24,332 Transport equipment 1,534 2,407 1,087 2,483 7,511 Furniture, other manufactured goods 12,588 16,506 12,491 25,515 67,100 Businesses by size Small fi rms (less than 50 employed) 82,668 88,653 60,269 139,904 371,494 Medium sized fi rms (50-250 employed) 7,606 8,559 4,557 9,630 30,352 Large fi rms (more than 250 employed) 2,890 3,103 2,279 4,479 12,751 Total 93,164 100,315 67,105 154,013 414,597 Source: Amadeus, National Statistical Offi ces Tab. 1: The sample – number of observations EM_2_2019.indd 8EM_2_2019.indd 8 19.6.2019 15:10:4919.6.2019 15:10:49 9 2, XXII, 2019 Economics and that resources in any industry are more productive in places where there is more of similar activities (Henderson, 2003). CuadradoRoura and Rubalcaba-Bermejo (1998, p. 134) posit that “there is a close relationship between the role of a city within a hierarchical system and its degree and type of specialisation”. Looking into the specialisation fi rst, the share of employment in city u contained in industry j is calculated. Since industries come in different sizes, the size of industries is normalized by their share at national level (in this case by share at the sample level for this country). In such way, the extent to which an industry j is represented in city u is identifi ed. By doing so, the observed industry distribution in city u is compared with that of national sample as a whole. Any value of specialization index above 1 indicates that an industry in a city area has an employment concentration above the national average. The specialisation index (normalised share or localization coeffi cient) is given as share of industry i in city j is divided by the corresponding sectoral share for whole population: (1) where i is the share of industry i in city j, while xi is the corresponding share at the national level. The relative-specialisation index is given as: (2) The extent of city diversifi cation is measured using the relative diversifi cation index (RDI). The basis for its calculation is the inverse of Herfi ndahl-Hirschman index, which is obtained from the ratio between one and the sum of the squares of the sectoral share in city’s employment. The diversifi cation index is: (3) DI takes the value 1 in the case where economic activity in the city under consideration is fully concentrated in a sector. Higher levels of this index suggest greater level of diversity, while smaller levels are associated with higher specialization. However, according to Duranton and Puga it is important to correct this index for differences in sectoral employment shares at the national level (Duranton & Puga, 2000), what leads to relative diversifi cation index. Relative diversifi cation index (RDI) is the inverse of the sum of the absolute values of the difference between each sector’s share in city’s employment and its share in national employment for each city over all sectors. This index will be higher as the structure of activities in the city under confi guration tends to refl ect the diversity of the national economy. Relative diversifi cation index (RDI) is given by: (4) Measures of specialisation and diversifi cation were applied to the dataset used in this research. Tab. 2 displays most and least specialized cities with over 50,000 inhabitants in SEE countries, in the period from 2006-2013. Most specialized city (Burgas) is 13 times more specialised by its index value than the least specialised one (Veliko Tarnovo). Two out of 10 most specialized cities are specialised in coke and refi ned petroleum products industry sector and three cities in tobacco industry sector. This type of specialisation is associated to cities’ natural resources or, in other words, to localised natural advantage. When considering the least specialized cities, the highest values of RSI are present among sectors of manufacturing industry. In the case of the least specialized cities (Veliko Tarnovo and Gabrovo in Bulgaria) the structure of manufacturing industry of a particular city is closer to the national. Although not strong, the negative correlation between city specialization and its size measured by number of inhabitants (-0.314) confi rms previous fi ndings that specialized cities tend to be generally smaller (Duranton & Puga, 2000). In line with that, it should be noted that among 15 most specialized cities in SEE there is no city with more than 250,000 inhabitants. It is interesting to note that among the top 20 most specialized cities there is no Slovenian city, which could lead to the conclusion that countries with higher GDP per capita in SEE have lower share of highly EM_2_2019.indd 9EM_2_2019.indd 9 19.6.2019 15:10:4919.6.2019 15:10:49 10 2019, XXII, 2 Economics specialized cities, but the economy size cannot be ruled out as an alternative explanation. In most of cases of high specialization in Europe, industrial specialisation refers to the high share of textiles, food processing, and in these cases manufacturing sectors are not diverse enough to support other sectors. According to Attaran (1986, p. 45) economic diversity is defi ned as “the presence in an area of a great number of different types of industries” or “the extent to which economic activity of a region is distributed among a number of categories. Much of the literature and research on economic diversity points that diverse economies are more resistant to fl uctuations associated with the business cycles (Hackbart & Anderson, 1975; Dissart, 2003), and lower output volatility is related with higher economic growth (Ramey & Ramey, 1995). Rank City Sector Relative specia lization index (RSI) City population 1 Burgas (BG) Coke and refi ned petroleum products 27.22 198,725 2 Bijeljina (BA) Other transport equipment 24.36 114,663 3 Ploieşti (RO) Tobacco products 17.99 209,945 4 Yambol (BG) Motor vehicles, trailers and semitrailers 16.18 71,561 5 Giurgiu (RO) Coke and refi ned petroleum products 15.90 61,353 6 Blagoevgrad (BG) Tobacco products 15.70 70,293 7Drobeta-Turnu Severin (RO) Chemicals and chemical products 15.18 92,617 8 Smederevo (RS) Basic metals 14.87 64,175 9 Nis (RS) Tobacco products 14.06 183,164 10 Galaţi (RO) Basic metals 13.69 249,432 11 Velika Gorica (HR) Other manufacturing 13.37 63,517 12 Kraljevo (SR) Other transport equipment 13.23 64,175 13 Târgoviște (RO) Basic metals 13.08 79,610 14 Banja Luka (BA) Basic pharmaceutical products and preparations 13.00 199,191 15 Mostar (BA) Repair/installation of machinery and equipment 12.63 113,169 ……. …….. ……………………………. ……….. 94 Pleven (BG) Basic metals 2.57 101,978 95 Ruse (BG) Furniture 2.33 147,055 96 Plovdiv (BG) Printing and reproduction of recorded media 2.32 341,567 97 Gabrovo (BG) Rubber and plastic products 2.28 56,003 98 Veliko Tarnovo (BG) Rubber and plastic products 1.95 68,676 Source: prepared by the authors on Bureau Van Dijk’s Amadeus database for Central and Eastern Europe, 2006-2013. Note: Average value of RSI over the period is given. Tab. 2: Most and least specialized cities in the period 2006-2013 EM_2_2019.indd 10EM_2_2019.indd 10 19.6.2019 15:10:4919.6.2019 15:10:49 11 2, XXII, 2019 Economics Country GDP per capita (current prices, euro) Cities 20 most specialized cities total sub-sample by countries 2013 2006-2013 2013 BA 3,569 3 12 BG 5,800 4 18 HR 10,200 1 9 RO 7,200 7 38 RS 4,800 5 17 SI 17,400 0 4 Total – 20 98 Source: prepared by the authors on Bureau Van Dijk’s Amadeus database for Central and Eastern Europe and Eurostat database, 2006-2013. Rank City (country) RDI 1 Plovdiv (BG) 2.91 2Sofi a (BG) 2.84 3 Vratsa (BG) 2.81 4 Belgrade (SR) 2.77 5 Cluj-Napoca (RO) 2.77 6 Bucharest (RO) 2.54 7 Veliko Tarnovo (BG) 2.34 8 Novi Sad (RS) 2.30 9 Buzău (RO) 2.07 10 Bihać (BH) 1.99 11 Dobrich (BG) 1.97 12 Gabrovo (BG) 1.90 13 Baia Mare (RO) 1.90 14 Stara Zagora (BG) 1.90 15 Varna (BG) 1.88 …………. 94 Drobeta-Turnu Severin (RO) 0.83 95 Bârlad (RO) 0.82 96 Vranje (RS) 0.79 97 Kranj (SI) 0.78 98 Smederevo (RS) 0.71 Source: prepared by the authors on Bureau Van Dijk’s Amadeus database for Central and Eastern Europe, 2006-2013. Note: Capital city rank: Sarajevo (29) Zagreb (33) and Ljubljana (48). Tab. 3: Distribution of 20 most specialised cities by countries and GDP per capita in SEE, 2006-2013 Tab. 4: Most and least diversifi ed cities in SEE, 2006-2013 EM_2_2019.indd 11EM_2_2019.indd 11 19.6.2019 15:10:4919.6.2019 15:10:49 12 2019, XXII, 2 Economics The relationship between city size measured by total employment in manufacturing/total population and its relative diversifi cation index is shown on Figs. 1 and 2. There is a pretty strong positive correlation between the city size (measured by employment in manufacturing) and the relative diversifi cation index, 0.6681. The correlation between relative diversifi cation index and city size measured by total population is even higher, 0.7543 and indicates that diversifi cation increases with city size i.e. with the size of the local labour market. These results are in line with the literature (Duranton & Puga, 2000). Also Da Silva Catela, Goncalves and Porcile (2010), measuring relative specialisation and diversifi cation in 524 Brazilian cities confi rmed that diversifi cation increases with city size. Larger cities have larger variety of production inputs, thus enabling concentration and economies of scales. Greater number of different industries can foster the exchange of ideas, so the fi rms can benefi t from the generation and diffusion of the knowledge. Larger cities are usually more diversifi ed and knowledgeintensive and have multiple specializations, while medium-sized and small cities sized cities may have just one or two, or none (Duranton & Puga, 2000; Audretsch, 2002; Crescenzi, RodriguezPose, & Storper, 2007; Drennan, 2002). Tab. 4 displays 15 most diversifi ed cities and 5 least diversifi ed cities. Within the 10 most diversifi ed cities there are three capital cities of the largest economies from the sample: Sofi a, Beograd and Bucharest. On the other hand, the capital cities of Bosnia and Herzegovina, Croatia and Slovenia are ranked in 29th, 33th and 48th place respectively, what can be related to the issue of the country size (in terms of population). Actually, the most diversifi ed cities are not to be found among top ten most specialized cities. This is additionally confi rmed by an indicative negative relationship between city specialization and diversifi cation, albeit with correlation coeffi cient of -0.317. However, diversity and specialisation are not exact opposites, as there are cities which are both diversifi ed and specialised. Fig. 1: Relative diversifi cation index and city size in terms of employment in SEE Source: Prepared by the authors based on Bureau Van Dijk’s Amadeus database for Central and Eastern Europe, 2006-2013. EM_2_2019.indd 12EM_2_2019.indd 12 19.6.2019 15:10:4919.6.2019 15:10:49 19 2, XXII, 2019 Economics is recorded in the medium low-tech group. The results of the k-means cluster analysis revealed that among larger cities, capital cities and most of regional centres underlying pattern in specialisation exists (specialization in hightech and medium high-tech industry) and this is also the case for smaller cities that have similar specialization pattern in low-tech and mediumlow tech industries. This research was supported by a grant from the CERGE-EI Foundation under a program of the Global Development Network. All opinions expressed are those of the author(s) and have not been endorsed by CERGE-EI or the GDN. References Abdel-Rahman, H., & Anas, A. (2004). Theories of systems of cities. In J. V. Henderson & J. F. Thisse (Eds.), Handbook of Regional and Urban Economics. Amsterdam: Elsevier. Arrow, K. J. (1962). The economic implications of learning by doing. Review of Economic Studies, 29(3), 155-172. https://doi. org/10.2307/2295952. Attaran, M. (1986). Industrial Diversity and Economic Performance in US Areas. The Annals of Regional Science, 20(2), 44-54. https://doi.org/10.1007/BF01287240. Audretsch, D. B., & Feldman, M. (1996). Innovative clusters and the industry life cycle. Review of Industrial Organization, 11(2), 253-273. https://doi.org/10.1007/BF00157670. Audretsch, D. B. (2002). The Innovative Advantage of US Cities. European Planning Studies, 10(2), 165-176. https://doi. org/10.1080/09654310120114472. Combes, P.-P., Duranton, G., Gobillon, L., Puga, D., & Roux, S. (2012). The productivity advantages of large cities: Distinguishing agglomeration from fi rm selection. Econometrica, 80(6), 2543-2594. https://doi.org/10.3982/ECTA8442. Cuadrado-Roura, J. R., & RubalcabaBermejo, L. (1998). Specialization and Competition amongst European Cities: A New Approach through Fair and Exhibition Activities. Regional Studies, 32(2), 133-147. https://doi. org/10.1080/00343409850123026. Crescenzi, R., Rodriguez-Pose, A., & Storper, M. (2007). The territorial dynamics of innovation: a Europe-United States comparative analysis. Journal of Economic Geography, 7(6), 673-709. https://doi.org/10.1093/jeg/lbm030. Da Silva Catela, E. Y., Gonçalves, F., & Porcile, G. (2010). Brazilian municipalities: agglomeration economies and development levels in 1997 and 2007. CEPAL Review, 101, 141-156. Dissart, J. C. (2003). Regional Economic Diversity and Regional Economic Stability: Research Results and Agenda. International Regional Science Review, 26(4), 423-446. https://doi.org/10.1177/0160017603259083. Di Giacinto, V., Gomellini, M., Micucci, G., & Pagnini, M. (2014). Mapping local productivity advantages in Italy: industrial districts, cities or both? Journal of Economic Geography, 14(2), 365-394. https://doi.org/10.1093/jeg/lbt021. Dogaru, T., Burger, M., Karreman, B., & van Oort, F. (2014). Functional and Sectoral Division of Labour within Central and Eastern European Countries: Evidence from Greenfi eld FDI. Journal of Economic and Social Geography, 106(1), 120-129. https://doi.org/10.1111/tesg.12093. Drennan, M. P. (2002). The Information Economy and American Cities. Baltimore and London: Johns Hopkins University Press. Duranton, G., & Puga, D. (2005). From sectoral to functional specialization. Journal of Urban Economics, 57(2), 343-370. https://doi.org/10.1016/j.jue.2004.12.002. Feldman, M. P., & Audretsch, D. B. (1999). Innovation in cities: Science-based diversity, specialization and localized competition. European Economic Review, 43, 409-429. https://doi.org/10.1016/S0014-2921(98)00047-6. Fujita, M., & Mori, T. (2005). Frontiers of the New Economic Geography. Papers in Regional Science, 84(3), 377-405. https://doi. org/10.1111/j.1435-5957.2005.00021.x. Giffi nger, R., Fertner, C., Kramar, H., & Meijers, E. (2007). City-ranking of European Medium-Sized Cities. Final report. Retrieved April, 2016, from http://smart-cities.eu/ download/city_ranking_fi nal.pdf. Glaeser, E. L., Kallal, H. D., Scheinkman, J. A., & Shleifer, A. (1992). Growth in Cities. The Journal of Political Economy, 100(6), 11261152. https://doi.org/10.1086/261856. Hackbart, M. M., & Anderson, D. A. (1975). On measuring economic diversifi cation. Land Economics, 51, 374-378. https://doi. org/10.2307/3146208. Hammond, G., & von Hagen, J. (1994). Industrial Localization. An Empirical Test for Marshallian Localization Economies [CEPR Discussion Papers 917]. EM_2_2019.indd 19EM_2_2019.indd 19 19.6.2019 15:10:5119.6.2019 15:10:51 20 2019, XXII, 2 Economics Henderson, J. V. (1986). Effi ciency of Resource Usage and City Size. Journal of Urban Economics, 19(1), 47-70. https://doi. org/10.1016/0094-1190(86)90030-6. Henderson, J. V. (2003). Marshall’s scale economies. Journal of Urban Economics, 53(1), 1-28. https://doi.org/10.1016/S00941190(02)00505-3. Hlaváček, P., Raška, P., & Balej, M. (2016). Regeneration projects in Central and Eastern European post-communist cities: Current trends and community needs. Habitat International, 56, 31-41. https://doi. org/10.1016/j.habitatint.2016.04.001. Jacobs, J. (1969). The Economy of Cities. New York: Random House. Kolehmainen, J. (2003). Territorial agglomeration as a local innovation environment – the case of a digital media agglomeration in Tampere, Finland. [MIT Special working papers series on local innovation systems (MIT-IPCLIS-03-002)]. https://ipc.mit.edu/sites/default/ fi les/documents/03-009.pdf. Lintz, G., Müller, B., & Schmude, K. (2007). The future of industrial cities and regions in central and eastern Europe. Geoforum, 38(3), 512-519. https://doi.org/10.1016/j.geoforum.2006.11.011. Longhi, S., Nijkamp, P., & Traistaru, I. (2005). Economic Integration and Manufacturing Location in EU Accession Countries. Journal of International Business and Economy, 6(1), 1-22. Longhi, C., Musolesi, A., & Baumont, C. (2014). Modeling structural change in the European metropolitan areas during the process of economic integration. Economic Modelling, 37, 395-407. https://doi.org/10.1016/j. econmod.2013.10.028. Marelli, E. (2004). Evolution of employment structures and regional specialisation in the EU. Economic Systems, 28(1), 35-39. https://doi. org/10.1016/j.ecosys.2004.01.004. Marshall, A. (1890). Principles of Economics (8th ed.). London: Palgrave Macmillian. Ramey, G., & Ramey, V. A. (1995). Crosscountry evidence on the link between volatility and growth. American Economic Review, 85(3), 1138-1151. https://doi.org/10.3386/w4959. Romer, P. M. (1986). Increasing Returns and Long-Run Growth. The Journal of Political Economy, 94(5), 1002-1037. https://doi. org/10.1086/261420. Rosenthal, S. S., & Strange, W. C. (2004). The Determinants of Agglomeration. Journal of Urban Economics, 50(2), 191-229. https://doi. org/10.1006/juec.2001.2230. Rumpel, P., Slach, O., & Koutský, J. (2013). Shrinking cities and Governance of Economic Regeneration: The Case of Ostrava. E&M Ekonomie a Management, 16(2), 113-128. Scott, J. W., & Kühn, M. (2012). Urban Change and Urban Development Strategies in Central East Europe: A Selective Assessment of Events Since 1989. European Planning Studies, 20(7), 1093-1109. https://doi.org/ 10.1080/09654313.2012.674345. Stierle-von Schütz, U., & Stierle, M. H. (2013). Regional Specialisation and Sectoral Concentration in an Enlarged EU: A comprehensive updated overview. Retrieved May 5, 2014, from http://www.infer-research. net/files_publications/WP%202013_2%20 -%20Stierle%20von%20Schutz%20Stierle.pdf. Svirčić Gotovac, A. (2006). Kvaliteta stanovanja u mreži naselja Hrvatske. Sociologija i prostor, 44(171), 105-126. Sýkora, L., & Bouzarovski, S. (2012). Multiple transformations: conceptualising the post-communist urban transition. Urban Studies, 49(1), 43-60. https://doi. org/10.1177/0042098010397402. Van Hagen, J., & Hammond, G. W. (1994). Industrial localization: an empirical test for Marshallian localization economies. In Location of Economic Activities: New Theories and Evidence. London: CEPR and Consorcio de la Zona de Vigo. Ivana Rasic Bakaric, PhD Institute of Economics, Zagreb Department for Regional Development Croatia [email protected] Katarina Bacic, PhD Mreža znanja Croatia [email protected] Suncana Slijepcevic, PhD Institute of Economics, Zagreb Department for Regional Development Croatia [email protected] EM_2_2019.indd 20EM_2_2019.indd 20 19.6.2019 15:10:5119.6.2019 15:10:51 21 2, XXII, 2019 Economics Abstract CITY SPECIALISATION AND DIVERSIFICATION IN SOUTH EAST EUROPE (SEE) COUNTRIES Ivana Rašić Bakarić, Katarina Bačić, Sunčana Slijepčević The main objective of the paper is to study the role of localisation and the urbanisation (or diversifi cation) economies in urban post-transition SEE, by constructing and analysing manufacturing specialisation and diversifi cation measures over the period 2006-2013. The second objective of the paper is to analyse differences within manufacturing industry across cities in terms of their technological complexity. Industries are mapped across cities with over 50,000 populations (98 cities in six SEE, covering 35.3% of the total SEE population), a population threshold that is in line with previous literature. The data were obtained from Bureau Van Dijk’s Amadeus fi rm-level database containing, most importantly, balance sheet data and profi t-and-loss account data for CEE. The analysis of manufacturing industry diversifi cation and specialisation in the cities is based on the relation between agglomeration economies of the Marshall-Arrow-Romer type (economies of location or specialization) and the Jacobs-Porter type (economies of urbanization or diversifi cation). Analysis results revealed that a particular specialisation pattern that would point to a homogenous system of cities throughout the region could not be confi rmed. City specialisation in manufacturing was negatively correlated to city size in SEE, but this relation has not shown particularly strong. Similarly to other countries, top-specialised cities are specialised in manufacturing closely related to natural resources such as petroleum products and tobacco, pointing to advantages arising from “fi rst nature” geography. However, diversity and specialisation are not exact opposites, as there are cities which are both diversifi ed and specialised. The results of the second part of the analysis show that medium-low technology and low technology groups of industries in manufacturing prevail in total turnover, with 36.2% and 35.0% share, respectively. City specialization in the prevailing technology group in SEE, in medium-low technology, is highest in Bulgarian, Bosnian and Herzegovinian and in Croatian cities. Key Words: City, specialisation, diversifi cation, manufacturing, agglomeration economies. JEL Classifi cation: R00, R12, O18. DOI: 10.15240/tul/001/2019-2-001 EM_2_2019.indd 21EM_2_2019.indd 21 19.6.2019 15:10:5119.6.2019 15:10:51