- 1 The data show the latest available years for the related countries. See also Figure-1.
1The share of manufacturing value added in GDP tends to decrease in almost all upper-middle and high income countries, excluding some South Asian economies. For example the share of manufacturing in GDP decreased to around 11 % (2009) in France and 11% (2010) in United Kingdom, 13% (2010) in United States, 19% (2010) in Japan, 12% (2011) in Chile, 13% (2011) in South Africa and 18 % (2011) in Turkey during the last two decades.1 This trend can be attributed to the increasing shares of services and financial sectors through deepening of financial globalization in addition to the consequences of economic growth on the sectoral composition á la Kuznets (1973). The term deindustrialization which is used for explaining the dynamics behind the decrease in the share of manufacturing nests several interrelated features from variations in the speed of technological change across sectors to location choice of investments. Shifting the focus from deindustrialization of a country to deindustrialization of a region puts emphasis on issues such as agglomeration, regional policies, and city planning policies or policies in urbanization.
- 2 Urban Agglomeration concept is based on Audretsch, Falck and Heblich (2007). The classification of (...)
2Last three decades some special policies have been implemented in order to decelerate industrial growth in Istanbul. Istanbul has very long history of industrialization. In spite of decrease in the share Istanbul in total Turkish manufacturing due to the implementations of these policies, this city still is the major industrial region (or urban agglomeration) of Turkey.2 The primary aim of the paper is to discuss the deindustrialization of Istanbul. However, in order to emphasis the distinctive characteristics of the deindustrialization of this old industrial center, the following section gives a brief theoretical background on deindustrialization; the aim of the brief note is to clarify the reasons behind deindustrialization at the country and regional level. The approach of the paper is to identify the links between deindustrialization and trade and industrialization policies and urbanization policy. Departing from this dual character of deindustrialization, Section 3 discusses the pattern of deindustrialization in a comparative framework at the country level; Section 4 focuses on the pattern of deindustrialization at the regional level in Turkey. Thus, by considering the regional deindustrialization issue, the paper focuses on the changes in the spatial concentration and shifts of Turkish manufacturing. The last section concludes the paper.
3This section explains the concept of deindustrialization. It also presents a brief explanation on the location theory which is important dimension of deindustrialization in a country or region: The location theory due to its emphasis on spatial characteristics may provide information to identify the long-term development in a space. In other words, there is a link between the spatial characteristics of a region and its deindustrialization tendency.
- 3 O’Rourke and Williamson (2002) state that “globalization defining term of the 1990s.” And, the 1990 (...)
- 4 Niepmann and Felbermayr (2009) include the following countries in the analysis : Australia, Austria (...)
4Rowthorn and Ramaswamy (1997: 5) assert that “… long-term decline in share of manufacturing employment in the advanced economies - a phenomenon referred as “deindustrialization.” In the long-run firstly, deindustrialization may appear as an outcome of the changes in development level. One of the expected results of development process in an economy is the change in the sectoral shares. In the first phase of development the share of agriculture decreases while the share of manufacturing increases. In the following phase of development, the share of service sector increases against the share of manufacturing. Kuznets (1973) states that “Major aspects of structural change include the shift away from agriculture to nonagricultural pursuits and, recently, away from industry to services; a change in the scale of productive units, and a related shift from personal enterprise to impersonal organization of economic firms, with a corresponding change in the occupational status of labor.” Therefore, we may anticipate a decreasing trend in the share of manufacturing in the sectoral composition of GDP in the developed countries while an increasing or at least a stable trend in the middle income countries. Secondly, globalization may force deindustrialization by lowering the trade costs:3 Niepmann and Felbermayr (2009) empirically show that lower trade costs have effect on the distribution of industrial production across countries.4
5Deindustrialization can be happen in both country level and regional level. At the country level, economic development and trade openness, which mentioned above, are critical factors behind the changes industrialization level of a country. However, deindustrialization may also differ at the regional level due to the effect of domestic policies. Deindustrialization at the regional level is related to the shift of manufacturing from on space to another in a country. The domestic policies to stimulate a shift in manufacturing may be either long-term or short term policies, or both. One of long-term domestic policy is industrialization strategy which has important outcomes on the formation of new agglomerations in the country. For example import substitution strategies give a way to the concentration of industries in a one center while openness of country has mixed results. An industrialization strategy basically rests on to modify the domestic relative prices which directly or indirectly affect the location choice of entrepreneurs. Other major domestic policies are urban policies and regional or sectoral support measures.
- 5 The seminal paper of Krugman (1991) entitled “Increasing Returns and Economic Geography” is the fir (...)
6The critical concept behind the movement of firms between regions or countries is transportation cost. The New Economic Geography models (NEG) emphases the effect of transportation cost. In a simple form of NEG general equilibrium model, lower transportation cost creates increasing return and consequently positive externalities.5 Firms tend to move to the regions which have positive externalities. Nevertheless, sometimes the opposite is possible: firms may quit a region due to negative externalities. The pull or push effects of a region are defines as “centripetal forces” and “centrifugal forces”. “Centripetal forces” covers “market size effects (linkages), thick labor markets and pure external economies while the “centrifugal forces” include immobile factors, land rents and pure external diseconomies (Krugman, 1999).
7However, questions in the location theory are not limited to the movement of firms in the space. The location of an industry causes other questions which emerge from supply and demand sides. Ottaviano and Thisse (2004: 2575-6) state five points on this issue:
“ … legacy of location theory can be summarized in five points: i) the economic space is the outcome of a trade-off between various forms of increasing returns and different types of mobility costs; ii) price competition, high transport costs and land use foster the dispersion of production and consumption; therefore iii) firms are likely to cluster within large metropolitan areas when they sell differentiated products and transport costs are low; iv) cities provide a wide array of final goods and specialized labor markets that make them attractive to consumers/workers; and v) agglomerations are the outcome of cumulative processes involving both the supply and demand sides”.
- 6 Audretsch et al. (2007) quote these amenities from Glaeser, E., Kolko, J., Saiz, A. (2001) Consumer (...)
8The first four points define the environment in which positive externalities contain. The fourth one points out the demand side while the first three points are related to the supply side. The NEG models are defined in this type of environments. However, the opportunities offered by a city are not limited an economic environment mentioned above. Structures of cities are not homogeneous. Therefore, the discussions should also include the heterogeneity of cities and concepts of city agglomerations. Agglomerations may be in different forms: Audretsch, Falck and Heblich (2007) classify them as “industrial district,” “industrial agglomeration” and “urban agglomeration.” They state that urban agglomerations “… are not dominated by one manufacturing industry but are, instead, historically grown centers rich with cultural life and other amenities that support a certain lifestyle” (Audretsch et al., 2007). Other amenities may be public infrastructure related education, effective crime prevention, transportation facilities, cultural buildings and so on.6 In addition to these amenities, we can also add good quality health infrastructures.
9The patterns of deindustrialization path in Turkey at the country level can be evaluated considering the deindustrialization concept and the location theory approach outlined above. The discussions in this section carried out in the framework of international comparison with selected middle income and high income countries. Figure-1 displays the share of manufacturing value added in GDP for high and middle income countries from different geographies during the second half of the last century. The data show that the manufacturing sector has a declining trend in Argentina, Brazil, Chile, Mexico and Peru from Latin America; Finland, Portugal, Romania, Spain and Turkey from Europe; and South Africa. On the other hand two East Asian countries, Republic of Korea and China displayed entirely different path: Especially Republic of Korea has a success story in manufacturing activities. Although Morocco displays a volatile trend in the share of manufacturing in GDP, Egypt and Tunisia exhibit strong manufacturing development. Turkey performs better but one can say that Turkey and South Africa have similar trend in the share of manufacturing in GDP. Both countries failed to maintain their success in 1990s (Figure-1): The share of manufacturing in GDP in South Africa has decreased from 24 percent in 1990 to 13 percent in 2011; in Turkey, this indicator has declined from 23 percent to 18 percent in the same period. The dramatic decreases in the share of the manufacturing employment are observed in France, United Kingdom and United States; however the declines in Germany and Japan are modest.
10These results are in line with the Kuznets (1973) approach; however, the quick decline in the share of manufacturing sector may be related to the openness as Niepmann and Felbermayr (2009) pointed out. Niepmann and Felbermayr (2009) state the decline of manufacturing sector in the OECD countries. However, the indicators displayed in Figure-1 shows that this does not happen only in the high-income OECD countries; deindustrialization is also a phenomenon in some upper-middle income countries.
Figure 1. Table. Manufacturing, value added
Classification by country from 1960 to 2011
World Bank, 2012.
11This section focuses on the changes in the spatial concentration of Turkish manufacturing sector. Therefore, the section discusses both the deindustrialization of old industrial centers and the emergence of new manufacturing centers.
12The Turkish industrial policy has changed dramatically after the 1980s. The import substitution strategy was replaced by export orientation approach. Trade and financial openness created a new business environment for Turkish firms during the last three decades. Export promotion strategies provided new opportunities also for local investors in various part of the Anatolia. Due to some local advantages some new industrial centers emerged during this period. The industrial centers are called as Anatolian Tiger by linking their success to the fast growing East Asian Tigers. Some of new industrial centers are originally ancient artisanal cities. For example Denizli, which is currently a new industrial center, has been known as an old textile city. One can argue that the new trade policies may also gave path to revitalization of industrial heritage. Some of emerging industrial regions are neighbor of metropolitan areas: Kocaeli and Manisa are respectively hinterlands of Istanbul and Izmir, and these cities benefited from the locational advantages. However, Kayseri in the middle of Anatolia and Gaziantep in the Southeast Anatolia have displayed rapid industrial development without these sorts of advantages.
- 7 “ The NUTS classification (Nomenclature of territorial units for statistics) is a hierarchical syst (...)
13Dogruel and Dogruel (2010 and 2011) define a classification of regions based on the development pattern of manufacturing in the region. In this section this classification has been employed for the evaluation of the shifts in the spatial concentration of Turkish manufacturing during the last three decades. This period also corresponds to a major change in economic policies in Turkey. Turkish regional system has three levels according to Nomenclature of territorial units for statistics of European Commission. In Turkey, there are 12 NUTS 1, 26 NUTS 2 and 81 NUTS 3 regions. Dogruel and Dogruel (2010 and 2011) classified the 26 NUTS 2 regions into five groups; they consider NUTS 2 level basic regions’ definition due to it can be used “for the application of regional policies7.” The indicator for the classification is the share of each NUTS 2 region in the total manufacturing employment: The initial period is represented by the average of the 1983-1985 and the final period calculated as the average employment level of the 1998-2000 period. In these works, certain industrialization characteristics of the regions are employed to classify the regions in addition to manufacturing employment level. These characteristics are “i) geographic location of the region; ii) whether a region is in the hinterland of another region or it is an industrial cluster, and iii) whether the region has an industrial tradition or not;” the five groups consist of industrial zones, hinterlands, emerging regions, minor industrial regions and poorly industrialized regions (Dogruel and Dogruel, 2011: 9). Dogruel and Dogruel (2011: 9) provide the following information to explain how they have classified the regions according to employment shares:
- 8 The New Geography Model see Krugman (1991) and Fujita (2010).
“A region is called as industrial zone if its average employment share is greater than 4 percent. Average employment shares in hinterlands and emerging regions are 3 to 4 percent, and they tend to increase during the period the paper covers. Hinterland region is the neighbor of an industrial zone. In the New Geography Models8, the formation and externality creation capacity of an agglomeration are related to these types’ proximities in a location. Therefore, a region is called as hinterland region if it is in the hinterland of an industrial zone; otherwise, as emerging region if it is a cluster without having any proximity with an agglomeration. A region is called as minor industrial region if its average share of employment is 1.5 percent and not classified as hinterlands and emerging regions. The regions are classified as minor industrial regions if their average shares of employment are below 1.5 percent.”
Figure 2. Table. Industrial regions by industrialization characteristics (private & state firms)
Industrial regions by industrialization characteristics (private & state firms)
Dogruel and Dogruel, 2011.
- 9 Due to data limitation the figure does not include the year 2005.
14Figure-2 displays the industrial regions with the classification into five groups. The map in Figure-3 illustrates the five industrial regions classified according to their manufacturing share and industrialization characteristics. Figure-4 covers the period from 2003 to 2009.9 The period of 1983-2000 covers both private and state firms. There remain few state firms after 2000 due to vast privatization implementation. Therefore, the data which cover the period of the years 2003-2009 does not include state firms; the data represent only private firms. The difference between the employment shares of two periods is due to the changes in the data structure of manufacturing which may elucidate the inconsistency between two periods. The new data collection method considers firm headquarters as the unit in contrast to previous method which considers production units (plants). Therefore, due to headquarters are in the big cities or metropolitan areas, the shares of industrialized zones in the second period (2003-2009) show a leap, and are higher than the first period (1983-2000): As an example, the average employment share of manufacturing in Istanbul reduced to 27 percent in the final period (1998-2000) from 30 percent (Figure-2); however, the beginning year 2003 of the second period (2003-2009) this share is 35 percent. We can observe also a leap in the share of TR41 (Bursa, Eskisehir, Bilecik). These differences may indirectly reflect the localization of headquarters: The headquarters of the most manufacturing activities are still located in Marmara region.
Figure 3. Map. Industrial regions of Turkey
Industrial regions of Turkey
Dogruel and Dogruel, 2011.
- 10 “Pole” and “industrial belt” are similar concepts with agglomeration. For the old industrial pole o (...)
15“Industrial zones” in Turkey consist of three distinct industrial poles or belts.10 Figure-2 displays the share of each industrial region. The shares changed slightly over the period. Hence, after 1980, the average share of manufacturing employment in the industrial zones has decline only from 67 percent to 65 percent in the end of period. The “industrial zones” depict the industrial heart of Turkey. This structure did not chance after 2000s: The share of manufacturing employment is 67 percent in 2003 and 65 percent in 2009 (Figure-4).
Figure 4. Table. Industrial regions
Industrial regions by industrialization characteristics for private firms, 2003-2009
Calculated from TURKSTAD Annual Industry and Service Statistics (NUTS2 level data).
16Although the share of industrial zones has remained approximately the same, the manufacturing sector shifted within the sub regions of industrial zones. Figure-2 and Figure-4 show that the share of manufacturing employment of Istanbul decreased. Dogruel and Dogruel (2010) state that this decline as the outcome of the deindustrialization policies implemented in Istanbul: These policies decreased the advantages of the region (stimulate centrifugal forces). Dogruel and Dogruel (2010) also assert that the deindustrialization policies of Istanbul indirectly magnified the effect of centripetal forces in Bursa (TR41), Kocaeli (TR42) and Tekirdag (TR21). The deindustrialization policies have also effect on the technological composition of manufacturing (a shift to low technology industries), firm size and the productivity in Istanbul (Dogruel and Dogruel, 2010).
17These three centers are the hinterland of Istanbul (Dogruel and Dogruel, 2010). However, Bursa is an exceptional case: The city has been the capital of the Ottoman Empire in the ancient time for nearly four decades.11 Bursa was also a very important silk production and textile center until 17th century. In spite of its historical advantages, proximity to Istanbul can be seen as the main determinant of the increase in the manufacturing employment during the last three decades. The manufacturing employment share in GDP has declined in Adana - Mersin region (TR62) and Izmir (Figure-2). However, the rate of decrease in Izmir is smaller than the rate of decrease in Adana and Mersin (TR62). Adana – Mersin and Izmir regions have relatively long industrialization history, starting from end of 19th century. Dogruel and Dogruel (2010) does not provide any clear evidence which may explain the reasons of manufacturing shift from these two old industrial regions. Dogruel and Dogruel (2010) capture only the case of deindustrialization of Istanbul. However, the local people and entrepreneurs in Izmir frequently mention about the negative effect of Manisa (TR33) which is a new industrial center in the hinterland of Izmir: They claim that the regional incentives toward Manisa (TR33) create positive externalities in this city. And, the firms prefer to localize in Manisa rather than Izmir due to the relative advantages of Manisa through selective regional policies.
18The total share of hinterlands and emerging regions has increased in the 1983-2000. This shows that the industrial centers in the hinterlands and emerging regions had an agglomeration capacity in that period. On the other hand, the agglomeration effect strongly persists across the industrial zones due to the insignificant changes in the share of manufacturing employment. The conditions were unfavorable for the minor industrial regions and poorly industrialized regions in the same period: The share of manufacturing employment has declined in both regions (Figures-2 and 7). However, the data demonstrate a slightly changing position in favor of the minor industrial regions and poorly industrialized regions in the period after 2003. The share of manufacturing employment in these two regions increased during the period of 2003-2009 (Figures-4 and 7). The indicator displayed a slight decrease in the industrial zones. Hinterland and emerging regions were unchanged after 2003 (Figure-7). Figures-5 and 6 show the shifts of manufacturing sector between regions. Sharp changes in the indicators from the period before 2000 to the period after 2003 should be attributed to the change in the data structure after 2003. The unchanged situations can also be taken as the outcome of the changes of the data structure.
Figure 5. Graphics
Shares of the five groups, 1983-2009 (Regional total)
Figure 6. Graphics
Shares of the four groups, excluding industrial zones, 1983-2009 (Regional total)
Figure 7. Table. Shares of the five industrial region groups (1983-2009)
Share of manufacturing employment
For the period of 1983-2000 : Dogruel and Dogruel, 2011 ; for the period of 2003-2009 : TURKSTAD Annual Industry and Service Statistics (NUTS2 level data).
19Deindustrialization became a worrying phenomenon during the last two decades for developed as well as developing countries. However, deindustrialization is not limited with the country borders; in some cases, deindustrialization happened at the regional level in a country. The paper examines the deindustrialization of Turkey. The evaluations cover both the country and regional level. The focus is the deindustrialization of Istanbul and other old industrial centers.
20The dynamics of behind deindustrialization are related with the policies at the international as well as domestic level. Economic development and globalization are the major dynamics behind the deindustrialization at the country level. At the regional level, industrialization strategy which is indirectly has link with the trade policies. Urban policies and regional incentives are other important domestic policies. The manufacturing sector in a country may shift due to these long and short term policies through positive externalities in favor of a region created by these policies.
21In spite of data inconsistency between pre-2000 and post-2003 periods, we observed that the manufacturing share of old industrial zones has been steadily decreasing. Considering the nature of the change in data structure, it is possible to conclude that the hinterlands of old industrial centers and new emerging centers benefited from deindustrialization of the old industrial centers.