The Effects of Trade Shocks on Regional

(1)

The Effects of Trade Shocks on Regional

Competitiveness

and the Italian Labor Force

Master Thesis

University of Groningen

Faculty of Economics and Business

MSc International Economics & Business

Abstract

Italy has always been characterized by a strong level of heterogeneity, which entails that the shocks that defined the second unbundling era had differential impacts across the country. My analysis aims at assessing the extent to which the competitiveness of Italian regions was affected by the consequences of two factors; the ICT revolution and trade openness of emerging economies. Such phenomena created Global Value Chains and changed the very nature of competitiveness. In my analysis, the latter is defined as the ability of regions to retain value-added and jobs derived from activities involved in the production of manufactured final products and it is measured following an input-output approach.

Keywords: Regional Competitiveness, Global Value Chains, Input-Output Analysis

Author: Michela Gasperini Student number: 3755479

(2)

Tables and Figures

Figure 1. Labor Productivity of the G7, annual growth rate (%) ...5

Figure 2. Percentages of Sectoral Value-Added in GDP in 1995, by regions ...10

Figure 3. Employment by Sector (in %) ...11

Figure 4. Employment Shares by Skills (in %) ...11

Figure 5. Stylized Global Value Chain ...12

Figure 6. Determinants of Competitiveness ...14

Figure 7. Structure of an Input-Output Table ...16

Figure 8. Input-Output Table with Matrix Algebra ...17

Figure 9. GVC Income Matrix ...18

Figure 10. GVC Income Shares of North-East Textile and Leather Products (in %)...22

Figure 11. GVC Workers of North-East Textile and Leather Products (in thousands) ...22

Table 1. Competitiveness of the Northeastern Italy Apparel Industry ...21

Table 2. Net Impact of GVC Income between 1995 and 2006 ...23

Table 3. Total Value Lost by GVC of Northeastern Apparel Products, 1995-2006...24

Table 4. GVC Income of Northeastern Textile Industry Lost in GVCs of European Countries’ Products, 1995-2006 ...24

Table 5. Regional GVC Income (GVCI) ...26

Table 6. GVC Workers, by sector ...27

Table 7. Growth in GVC Workers (in %) by Skill Level, between 1995-2001 and 2001-2006 ...27

Table 8. GVC Income Lost to China and Eastern Europe between 1995 and 2006 ...28

Table 9. Net Impact of GVC Income (GVC I.) in million US$, between 1995 and 2006 ...29

Table 10. Impacts of China and Eastern Europe on GVCs of Italian Regions, between 1995 and 2006 ...30

(4)

Chapter 1 Introduction

Since the end of the 20th century, the global economy was hit by two shocks; the ICT revolution and the opening up of emerging economies such as China. Both phenomena had implications on the traditional global trade patterns. Indeed, the exposure to low-cost goods and services and the cheap labor force of emerging economies had repercussions on the competitiveness of developed nations. For instance, in order to be able to face the new competitors, firms began to relocate part of their production in geographical areas where it was cheaper to execute (e.g. had lower labor costs). Italy was not immune to these changes and its economy suffered the consequences of these shocks under two perspectives, which are emphasized by the headlines of two newspapers; “Cosí la Cina ci rovina” (“Thus China will ruin us”) from Panorama (2005) and “La carica delle duemila imprese” (“The attack of the two thousands firms”) from Il sole 24 ore (2004). The former underlines how the competition with low-cost Chinese exports has jeopardized Italian regions’ competitiveness, whereas the latter refers to the job losses that followed the choice of the relocation of some of the production stages abroad. At this point, one question arises; what were the effects of the ICT revolution and trade liberalization of Eastern European countries and China on the competitiveness of regions in Italy? In particular, I look at the period that goes from 1995 to 2006, which is when these trends became more prominent.

The consequences of these shocks were not the same for the whole country but differed across Italy. This is because, since its Unification in 1861, the country has always been characterized by a large degree of heterogeneity, in terms of development disparities between regions. These differences are accentuated especially between the North and the South of the country, with the latter being characterized by persistent backwardness and the former being more industrialized. Suffice it to say that, in 2016, the South had an average income of 16.113 euros, much lower than the average incomes of the North and Central Italy of 24.356 euros and 21.189 euros, respectively (Istat, Bes 2016). Moreover, regions are marked by striking differences in factor endowments, with the North being more abundant in high-skill labor and capital, and the Mezzogiorno, which comprises Southern regions plus the islands, being limited to a less-skilled labor force.

(5)

In my study, I will precisely quantify these trends. More specifically, I will assess the changes, and the extent to which trade shocks were responsible for these changes, in the level of competitiveness of Italian regions. The period I analyze goes from 1995 to 2006, which coincides with the outset of those trade shocks. In particular, the years at the turn of the 21st century, are often referred to as the second unbundling era (Baldwin, 2006) in order to capture the essence of the new trends, e.g. the fragmentation of production in the global trade structure and the creation of the so-called Global Value Chains (GVCs). In GVCs, the production of goods and services is broken up into small parts, each of which takes place in different countries (or regions). Therefore, the final product is an assembly of parts and components that are sourced from several geographical locations. The origination of the unbundling process implied the divergence from traditional structures under two perspectives. Firstly, it changed the very nature of competitiveness. If before countries competed against each other in terms of products, nowadays it is possible to be competitive only in specific activities that are required for the production of that product. The second perspective, instead, refers to the implications that the unbundling process had on trade statistics traditionally used to measure competitiveness, i.e. gross exports. The increase in trade in intermediate products created the need to develop a new methodology to measure the exports value of trade. In my study, I follow the approach used by Timmer et al. (2013), who consider the value-added of trade, i.e. the value of gross exports discounted by the value of imports. To trace the changes in patterns in income and employment levels generated by participating in global value chains more accurately, I make use of global input-output tables with regional details for Italy.

In order to provide an answer to the overarching question driving this analysis, I analyze the competitiveness of Italian regions under three perspectives: i) the level of competitiveness within GVCs, ii) the level of competitiveness of GVCs, and iii) the level of competitiveness of the regions in three European countries- UK, Germany and France. For each one of these three determinants, I isolate the impacts generated by the participation of China and Eastern European countries in Western trade. In general, I find that every region decreased in competitiveness throughout the period studied and the increase in prominence of China and Eastern Europe further exacerbated the regional positions in the global economy. However, it is also true that some regions, mainly in the North, were able to profit from the opportunities generated by the combination of the ICT revolution and the opening up of new and low-cost countries, by relocating production and upgrading. Indeed, I find traces of the latter in the shift of demand towards a more educated labor force, which signals the movement towards activities that yield more value-added whilst relocating abroad low value-added ones.

(6)

Chapter 2 Literature Review

2.1 Defining (regional) Competitiveness

Despite its importance and its increasing relevance on the policy agenda, the concept of competitiveness is still elusive (Reinert, 1994). A first definition of regional competitiveness comes from the EC’s Sixth Periodic Report on the Regions, which defines it as “the ability to produce goods and services which meet the test of international markets, while at the same time maintaining high and sustainable levels of income or, more generally, the ability of (regions) to generate, while being exposed to external competition, relatively high income and employment levels.” (Commission of the European Communities, 1999). According to Storper (1997) competitiveness is “the capability of a region to attract and keep firms with stable or increasing market shares in an activity, while maintaining stable or increasing standards of living for those who participate in it.” (Storper, 1997). Similarly, Cooke’s (2004) interpretation refers to the “capability of a sub-national economy to attract and maintain firms with stable or rising markets shares in an activity, while maintaining stable or increasing standards of living for those who participate in it.” (Cooke, 2004).

All of the above definitions have a similar underlying concept, in that they all refer to the ability of the region to retain income and employment growth in the face of increasing competition (Timmer et al., 2013). This is precisely the approach I follow, given its coherence with the purpose of my analysis. More specifically, the study I conduct aims at quantifying the amount of income and jobs that have been maintained by Italian regions despite the competition shocks from Eastern Europe and China. It is important to mention that my focus is only on income and jobs generated from the production processes of final products that are internationally contestable and prone to fragmentation, that is from activities that are involved in the production of manufactured final products. This does not mean that I consider only manufacturing activities, but rather all the sectors that contribute to the assembly of manufactured final products, such as services.

2.2 Trade before and after the Second Unbundling

Competitiveness has always been high on governments’ agenda. However, while in the past regional policymakers aimed at hosting the whole production process of a product, nowadays they focus only on specific activities. An explanation can be found in the analysis of the evolution of trade patterns and in the literature on the trends that have taken place since the 1990s, that is the opening up of emerging countries and the ICT revolution. These two shocks have altered the traditional trade structure of the global economy, creating additional competitive pressure on developed countries in the Western world.

(7)

it is increasingly more viable for firms to benefit from the differences in comparative advantages between countries (and regions) (Los et al., 2017) by slicing up the value chain (Krugman, 1995). More specifically, firms find it profitable to offshore stages of production to countries where they are relatively cheaper to be undertaken, creating the so-called Global Value Chains (GVCs). The latter are defined as “the value added of all activities that are directly and indirectly needed to produce [a manufactured final product] [...].” (Timmer et al., 2014) and are the consequence of the unpacking process firms have carried out (Baldwin, 2006). Baldwin (2006), refers to this period as the second unbundling and argues that the changing nature of globalization affected the levels at which international competition occurs. Indeed, during the first unbundling, countries competed mainly at the sectoral level, e.g. Northeastern Italian versus German shoes, whereas after the second unbundling, international competition started to take place at the level of the production stages, e.g. laces production in Northeastern Italy versus laces production in China.

The changes in technology coincided with the liberalization, and subsequent participation in trade agreements, of emerging economies. For instance, countries in Eastern Europe dismantled trade barriers in 1989, with the fall of the Iron Curtain, and submitted the application for EU membership soon after. During the 2000s, governments in those countries carried out a number of reforms to prepare for the accession in the European Union, including the implementation of a functioning free-market economy. In 2001, China became a member of the World Trade Organization (WTO), which entailed, among other things, the abolishment of trade barriers and the opening up of domestic markets to foreign firms. The consequence was a staggering increase in trade, especially in intermediate inputs, between Western European countries and the newcomers. The availability of a significant pool of low-cost labor, capable manufacturers, large quantities of raw materials and substantial new markets were indeed a game changer for Western firms.

The changes in trade patterns have implications for the competitiveness of Italian regions. To see this, it is important to give some context about the situation in Italy. Since the 1990s, the economic growth of the country has significantly slowed down and the main reason, put forward by the literature, was a decrease in productivity growth rate. Figure 1 precisely stresses this point by showing the lower labor productivity rates in Italy compared to the G7 between 1995 and 2006. Daveri and Jona-Lasinio (2005) argue that this fall was mainly due to a decline in total factor productivity (TFP), rather than in capital deepening. Prior to the 1990s, the economy was characterized by dynamism and innovative capacity that slowed down with the changes in the world economy. The features of Italian firms that were previously being identified as key components for the manufacturing success in the 1980s, were now being considered by scholars as anomalies (Rabellotti et al., 2009). For instance, 82.7% of Italian manufacturing firms had fewer than 10 employees (Agostino et al., 2016) and Italian industries, especially in the Northeastern and Central regions, have always been spatially concentrated in clusters so to benefit from Marshallian externalities1_{. Moreover, Italian firms have always had a comparative}

advantage in the traditional industries of the Made in Italy- Italian products that are particularly

(8)

appreciated abroad, thus are mainly earmarked to export-, which in that period were experiencing a dramatic decrease in TFP (Daveri and Jona-Lasinio, 2005).

Figure 1.Labor Productivity of the G7, annual growth rate (%)

Source: OECD Productivity Statistics, GDP per capita and productivity growth.

While, in the past, those aspects played in favor of Italian firms, the new nature of globalization created serious challenges. Therefore, many producers started to search outside the borders of the districts, and outside Italy, for alternative and lower-cost suppliers. This resulted in the offshoring of production processes and in the disintegration of the districts (Rullani, 1997).

Empirical analyses that quantify the level of fragmentation of production in Italy are mainly case studies and focus on specific value-chains in specific locations. Even though it is doubtful whether their results can be generalized, case studies are still helpful in giving a preliminary flavor of the trends taking place in different regions in Italy. Prota and Viesti (2007), for instance, found that the majority of the relocation moved towards Eastern European countries, as Romania, and the main justification was the availability of cheap labor. Indeed, it was estimated that in the first half of the 1990s, the cost of labor per unit of output in the sector of textile and clothing, was three times higher in Italy than in Eastern European countries (Baldone et al., 2002). In the footwear sector, Amighini and Rabellotti (2006) found that the districts specialized in low-priced markets, as Southern districts, moved a higher percentage of the production of intermediate inputs abroad relative to districts in Veneto and the Marche. Therefore, Southern regions imported parts from foreign subcontractors more than the rest of the peninsula. This means that the outsourcing strategies varied according to the market position of the district. Similarly, Gianelle (2005) reports empirical evidence on the fragmentation of the production process in textile and leather industries in the region of Veneto. He depicts a picture in which producers were increasingly moving abroad the stages of production that concerned exclusively manufacturing activities. Simultaneously, they retained higher value-added activities, such as design, marketing and distribution, within the national borders, producing consequences on the employment structure in the firm. Indeed, as the activities retained at home required higher degrees of knowledge, the author found a shift of demand towards a more educated labor force. Corò et al. (2005) computed an indicator of

-0,5 0 0,5 1 1,5 2 2,5 3 1995-1997 1998-2000 2001-2003 2004-2006

(9)

“Foreign Involved Employees”2_{for three value chains, textile and apparel, leather and shoes and wood}

furniture, and found that the textiles and apparel chain was the one more open to internationalize production and that all three value chains were mainly linked to Eastern European countries.

2.3 Competitiveness in Global Value Chains

As previously mentioned, the production of a final good requires the implementation of many activities that take place in more than one country and that add value to the product. However, it is important to understand that the amount of value-added depends on the type of the activity. In general, knowledge-intensive occupations, as R&D, marketing, advertising and distribution, add greater value than manufacturing operations. Countries (or regions) tend to retain these activities that yield more value-added and outsource the low-skill labor tasks to countries with cheaper labor force. This is important because it implies that upgrading to higher value-added activities is a viable way to increase the level of competitiveness. According to Gereffi and Memedovic (2003), upgrading refers to the acquisition of “technological, institutional and market capabilities” that allows firms to improve their products and their productive efficiency or to move into activities that generate higher value-added.

The literature on competitiveness that takes a global value chains approach advocates on the importance of upgrading in order to face the threats of low-cost producers that continuously enter the global economy. Major contributions to the field were given by Gereffi (1994, 1999), Gereffi et al. (2005), Gereffi and Korzeniewicz (1994), Gereffi and Kaplinsky (2001) and Humphrey and Schmitz (2002). Humphrey and Schmitz (2002), who were among the first scholars to put the spotlight on regions, discuss four types of upgrading possibilities. Process upgrading entails the improvement of the production system either by increasing its efficiency or by introducing new technologies. Product upgrading requires the shift of production to more sophisticated products, thus increasing the unit value. Functional upgrading means increasing the skills needed for a certain activity by acquiring new functions. Lastly, in inter-sectoral (or chain) upgrading firms take on new value-chains, often related to the one in which they were previously operating. For instance, Taiwan used the knowledge acquired in producing televisions to make monitors and upgrade in the skill-intensive computer sector3_.

The focus of this stream of literature is on the participation of developing countries in GVCs. However, it is not far-fetched to associate the Southern part of Italy to a developing country. It suffices to think that, to this date, Italy is the highest beneficiary of European Funds4 and, according to De Angelis et al. (2018), 70% of the total EU cohesion policy fund of 2007-2013 allocated for Italy went to the South. Moreover, one can apply the same concepts of upgrading to developed nations. According to Agostino et al. (2015), if suppliers are capable of taking advantage from the relation with lead firms, i.e. the large multinational that operates as a leader in the value-chain by setting the criteria to which suppliers must conform, then the benefits that firms enjoy in countries such as Italy are as those that firms in

2_{The indicator measures the extent to which international production was involved in each industrial district they analyze and consists} of the amount of foreign workers that are directly and indirectly involved in Italian GVCs (Corò et al., 2005).

3_{The concept of upgrading is further clarified in section 3.2.1 and, in the analytical part of the paper (chapter 4), I discuss evidence of} upgrading patterns.

(10)

developing countries enjoy. Empirical evidence about the process of quality upgrading of Italian exports is given for instance by De Nardis and Pensa (2004). They studied the extent to which traditional Italian industries were jeopardized by international competition and found that, during the 1980s and 1990s, Italian exporters were not completely vulnerable to foreign competitors, not even those located in low cost countries. Consistent with these findings, Monti (2005) found that, as Italian exports had, on average, better quality than the products exported by developing nations, only a small share was in direct competition with the latter. Thus, the author gives evidence of quality upgrading of Italian exported products in the 1990s. One example is the textile district in Biella (Piedmont) where some SMEs shifted their production towards very high quality fabrics, e.g. cashmere, alpaca and vicuna (Rabellotti et al., 2009).

Another stream of literature also points out that the threat of emerging countries, which were rapidly expanding their competition to medium technology products, was looming. Monti (2005) showed that direct competition between Italian and emerging countries’ exports was growing since the beginning of the 2000s. Similarly, Sammarra and Belussi (2006) analyzed the Vibrata-Tordino-Vomano clothing industry (Abruzzo) and found that the economic slowdown that started in the 1990s had been worsened by the increased competition from low-labor cost countries, mainly from Asia. These empirical studies suggest that the misfortune of Italian producers was that their exported goods were very similar to the ones of emerging countries, i.e. low-tech and labor-intensive, implying a reduction in foreign, and most likely also national, demand throughout the period studied.

2.3.1 An Input-Output Approach

(11)

economy over the period of 1995-2007, and this was especially true for the newcomers Hungary, Czech Republic, Slovakia, Poland and Bulgaria.

The consequences of the Eastern Europe enlargement for Western European countries were discussed by Lurweg and Westermeier (2010), who put the spotlight on Germany. The authors explored the relationship between trade shocks and employment patterns during the period of 1995-2006. They found that trade positively influenced jobs creation; around 7% of total German jobs would have not been created in an autarkic state. Feenstra and Sasahara (2018) extended the analysis of Autor et al. (2016) on the China shock on the exports and employment in the US by using an input-output approach. They compared the jobs created due to the growth of US exports to China with job destruction due to an increase in Chinese imports and found a net effect of about 1.7 million jobs creation.

The use of an input-output approach has some major advantages. First of all, it goes beyond the limited scope of case studies. Indeed, even though the latter can provide insightful information, it remains uncertain whether the results hold beyond the case considered and can thus be generalized. For instance, the findings of Amighini and Rabellotti (2006) are sector-specific and it is questionable whether they remain valid also for sectors other than the footwear one they consider. By using input-output analysis, instead, it is possible to depict a more general picture that takes into account all the industries and regions of the country and gives a perspective on the trends in competitiveness that is valid for a broader set of circumstances.

It is important to realize that the two methods (case studies and input-output approach) are complementary. Both give insightful information by taking different perspectives and do not arrive to contrasting results, rather, their outcomes are different pieces of a same puzzle.

Another key advantage is that an input-output approach goes beyond the traditional measures of competitiveness that, due to the new trends in the global trade patterns, e.g. increase in intermediate inputs trade, are not reliable anymore. More specifically, gross exports figures do not accurately represent the value contributed by the specific country (or region), as they do not distinguish between trade in intermediate inputs and trade in final products. Thanks to the well-known Leontief inverse, it is indeed possible to differentiate between the foreign value-added and the value added by the activities taking place in the specific country or region considered.

(12)

2.3.2 Hypotheses Formulation

Combining the knowledge gathered from the studies highlighted in this chapter, it is possible to put forward expectations on the potential outcomes of my empirical investigation. Firstly, I expect a general loss of competitiveness for all Italian regions, due to the hurdles they faced, highlighted by the literature, and their inability to be flexible in an increasingly competitive market. This decrease in competitiveness may be rooted in an expanded participation of Chinese and Eastern European countries in GVCs of the regions’ industries, which would have jeopardized the latter’s relative competitive positions.

Secondly, I expect a more intense impact of Chinese and Eastern European countries’ competition on Northern regions. This is because the latter were more integrated into the global economy and thus more exposed to potential shocks. Istat figures on the turn of the century reveal that gross exports of Northern regions, although being a flawed statistic as argued before, accounted for more than 70% of the total national amount (ICE report, 2007). However, it is important to reiterate that the ICT revolution and the openness of the newcomers offered advantages as well, such as relocation opportunities and demand increases. I expect Northern regions to have benefitted the most by, for instance, undertaking processes of upgrading or by relocating the manufacturing stages of production and concentrating on higher value-added activities. The same reasoning would not hold for Southern regions, as they lagged behind in economic and development terms.

Thirdly, looking at the regional competitive positions in the European economy, I expect that the opening up of emerging economies in the early 1990s has harmed Italian regions under two perspectives. Firstly, in terms of lower foreign demand, as the newcomers had comparative advantage in goods and services with similar characteristics to the ones exported by the regions, that is cheap and low-tech. Secondly, in terms of relocation choices of European countries, as a e.g. German firm may have preferred to relocate production in e.g. China rather than in Italy.

(13)

Chapter 3 Data and Methodology

3.1 Value-Added and Employment Patterns: Descriptives

Describing the changes in sectoral composition and employment patterns gives a first hint on the evolution of the competitiveness in the regions. Figure 2 shows that in 1995 industrial activities were particularly important in the North, especially in the Western part, and not very prominent in Central and Southern regions. For every macro-region, market services occupied a dominant role in the economy, yielding the most value added. This is true especially for North-Central Italy and to a lesser extent for the rest of the peninsula. Non-market services, instead, were particularly of high value in Southern regions. On the opposite hand, the value-added in the manufacturing sector was minimal if compared to the service sectors and this was true especially for the regions located in the Central-Southern parts of the country. Indeed, the majority of the value-added in the industry sector was concentrated in Northern regions.

Figure 2. Percentages of Sectoral Value-Added in GDP in 1995, by regions

Source of data: Cherubini and Los (2015).

Turning to changes in the labor force, figure 3 shows the employment patterns for the period considered. It is noticeable that, in 1995, the majority of workers in every region was employed in service sectors. Moreover, industry played an important role especially in the economies of Northeast and Northwest Italy, where it employed approximately 35% of the whole labor force. In 2006, the share of services in every region increased and the labor force in manufacturing decreased. This hints to a shift of workers away from manufacturing production and towards the service sectors, especially market services. This phenomenon had implications on the relative demand for skills required to work in each sector.

Figure 4 shows precisely the shares of each skill level in total employment in each region in 1995, 2001 and 2006. In general, the North has always had a higher percentage of high- and medium-skilled labor force relative to the South. For instance, in 1995, the Northwest had approximately 3000 thousand high and medium skilled workers involved in the production of manufacturing final products, whereas the South counted approximately 2500 thousand (see Appendix A for disaggregated employment figures by skill levels). Moreover, for all three years considered, the South shows a higher

0 10 20 30 40 50

Primary Industry Costruction Market Services NonMarket Services

(14)

share of low-skill workers, e.g. in 2006 Southern regions counted as many low-skilled workers as the rest of Italy had five years prior.

Figure 3. Employment by Sector (in %)

Source of data: Cherubini and Los (2015).

Figure 4. Employment Shares by Skills (in %)

(15)

3.2 Approach

3.2.1 Stylized GVC

Before delving into the methodology used, one should clarify the concept of GVCs, discussed in this subsection, and the concept of input-output tables, developed in sub-section 3.3.1. Figure 5 shows a highly stylized example of a GVC for the apparel industry that includes three countries (can also be interpreted as regions). The final product is produced in country C, which is thus called the “country-of-completion” as it hosts the activities that complete the product before it is shipped to retailers, wholesalers or consumers.

A final clothing product is made up of: i) raw materials (e.g. natural or synthetic fibers) and additional decorations (e.g. zippers), ii) other components (e.g. yarns and fabrics), and iv) many activities (e.g. cutting, sewing and ironing) and services (e.g. logistics, marketing and financial services). Therefore, the production of the final textile product requires intermediate inputs and activities that take place in country C itself but also in other parts of the world, as in countries A and B. In this example, an increase in the demand of a piece of clothing assembled in country C, would increase the production in the country itself, but also the activities taking place in countries A and B, such as spinning and knitting or the production of raw materials. Therefore, an increase in the demand, foreign or domestic, of country C’s final product, increases production also in intermediate inputs and services sourced from other countries. Moreover, the manufacturing of those intermediate inputs and service activities taking place in the chain, require labor. This means that the demand for final product of country C creates value-added and jobs all along the value chain.

The amount of value-added and jobs produced depends on the different activities. For instance, activities such as distribution marketing and sale services, yield higher value than pure production activities, such as spinning, knitting, cutting and sewing. Moreover, activities that entail higher value-added require also a more skilled labor force. In contrast, manufacturing activities are able to absorb a large number of workers, mainly unskilled.

Figure 5. Stylized Global Value Chain

Notes: The yellow highlighter indicates that Country C hosts the last stage of production (i.e. country-of-completion).

(16)

GVCs. Product upgrading in this scenario means that e.g. country A improves the quality of the raw materials, enhancing the overall quality of the final piece of clothing. Process upgrading can happen, for instance, in the efficiency betterment of spinning activities taking place in country B, by improving the technology of the machinery used. Functional upgrading is identifiable in country C, in case the latter upgrades to handle the marketing of the product as well. In this case, country C would move down the value chain by incorporating an activity that would yield high value-added. Lastly, to explain inter-chain upgrading, one should also consider a second global value chain, e.g. a GVC for the machinery industry. Succeeding in inter-chain upgrading means that e.g. country B manages to become part of the GVC of the machinery industry by starting to produce parts and components needed to e.g. sew. It can do so by utilizing the knowledge acquired through the specialized service of cutting and sewing developed in the apparel GVC depicted in figure 5.

3.2.2 Discussion on the Methodology

Before outlining the methodology used, it is important to make two clarifications. First of all, the value of GVC income is not a pure measure of the competitiveness of the manufacturing sector, as it does not include manufacturing activities that produce intermediate inputs for non-manufacturing products. GVC income is rather a measure of the degree of competitiveness of all those activities that are related to the production of manufactured final products. Therefore, it includes also those services, e.g. financial services, that participate in GVCs and are fundamental for the making of the manufactured final product. These indirect contributions are included in the structure of input-output tables by accounting for linkages across different sectors, as will be shown in section 3.3.1. Moreover, competitiveness is measured on the domestic, rather than national, level. This means that it ignores the ownership of the production factors, e.g. the activity of spinning and knitting in figure 5, may be owned by a foreign firm that is operating in country B but it will be considered part of the value added by country B.

The methodology used measures competitiveness as the total GVC income of each region to the total worldwide GVC income. For instance, changes in the amount of the value-added generated by Northeastern Italy regions in the production of manufacturing final products, relative to the amount of value-added generated in the world economy in the production of manufacturing final products, would reflect the changes in competitiveness of the region relative to other countries.

(17)

value-added to the total GVC income, entails a reduction in value-added of country C. To isolate the impacts, I look at the change in e.g. Chinese contributions of value-added to the total size of GVCs with e.g. Northeastern Italy being the region-of-completion, between 1995 and 2006. Then, I proportionally redistribute the increase in Chinese contributions to Northeastern regions, based on the latter’s contribution of 1995. In this way, I have a measure of the amount of value-added that would have been generated by Northeast, had China not increased its share in the GVC.

Figure 6. Determinants of Competitiveness

It is important to mention that the increase in openness of China and Eastern European countries has also benefitted the regions, as the latter were able to participate in new, fast-growing GVCs. Therefore, just looking at the amount of value added lost would give a biased picture. As I discuss in section 3.3.2, I deal with the problem by looking at the amount of value-added contributed by the regions to Chinese and Eastern European GVCs. Subtracting the amount of value-added gained and the amount lost, I have a net measure of the impact of the newcomers on regional competitiveness.

The second determinant concerns the relative sizes of GVCs of each region to the total amount of GVC income produced in the world economy. This determinant can be thought as reflecting the competitiveness level of a value chain, e.g. the competition between Chinese and Northeastern shoes production, but also the competition between GVCs of Northeastern textile and electrical equipment. An increase in the shares that GVCs completed in the regions captured of total GVC income reflects increased competitiveness. Again, as I am particularly interested in the effects of Chinese and Eastern European trade shocks, I isolate the latter following the same methodology outlined for the first determinant. Indeed, I proportionally redistribute the increase in size of e.g. GVCs having China as country-of-completion, to Italian regions. In this way, I show the extent to which GVCs completing production in the regions reduced in sizes due to an increase in the relative sizes of GVCs having China hosting the last stage of the production process.

(18)

value-added to Italian regions, in order to show the amount of value-value-added that would have been retained by the regions had China not increased its shares. It is important to emphasize that the last determinant of competitiveness looks only at the within GVCs component.

3.3 Data

My research question poses the focus on regions, rather than the nation as a whole, and, as the World Input-Output Database (WIOD) does not include information at the sub-national level, data for my study are taken from Cherubini and Los (2015). The latter integrated sub-national input-output tables for Italy into the WIOD for three years (1995, 2000 and 2006) by incorporating Interregional Supply and Use tables (IRSUTs)5_{and trade data into the WIOD. Italy was divided into four macro-regions;}

Northwest, Northeast, Centre and South and Islands (Mezzogiorno). To incorporate Italian IRSUTs in the WIOD the authors replaced the Supply and Use Tables of Italy as a whole by the ones for the “Four Italies” (Cherubini and Los, 2015). Each macro-region is thus considered as a country, resulting in a final database of 44 countries, instead of the original 41. To link the new SUTs to each other and to the rest of the tables in the WIOD system, the authors used trade flows data among Italian regions and between those macro-regions and other countries. In particular, the WIOD contains information for economic linkages across 35 industries, mostly at the two-digit ISIC rev. 3 level or groups (see Appendix B for a description of the industries in WIOD).

A key feature of the dataset constructed by Cherubini and Los (2015) is that it also includes data on employment and workers are further classified by skill levels. In order to disaggregate labor force information at the regional level, as data are usually available only at the national level from EU KLEMS, the authors used two datasets; the Italian Regional Accounts and the Labors Force Survey (LFS). These were used to split employment information from the EU KLEMS database into macro-regional data for the “Four Italies”. Furthermore, workers were differentiated between low-skilled (LS), medium-skilled (MS) and high-skilled (HS), according to the International Standard Classification of Education (ISCED). University graduates are considered to be HS workers (in ISCED categories 5 and 6), workers who reached a higher education below degree, intermediate vocational plus advanced education and low intermediate are considered to be MS (in ISCED 3 and 4) and workers with no formal education are considered to be LS (in ISCED 1 and 2).

3.3.1 Structure of Input-Output Tables

(19)

and intermediaries that can be either sourced from abroad or from the domestic economy. The output is used either to satisfy final demand, domestic and foreign, or it is used as an intermediary input, either at home or abroad.

Figure 7. Structure of an Input-Output Table

Notes: It is important to note that the submatrices in figure 7 have different dimensions. The rows and columns related to “Rest of Italy” include all industries in Northeast, Central and South Italy. Therefore, e.g. the submatrix of “int. use of domestic outputs” related to Rest of Italy is three times the same submatrix related to Northwest or Country 1. Similarly, “ROW” includes all other countries of the dataset. Moreover, “Ind.”, “Int. use”, “VA” and “1” are the abbreviations for “Industry”, “intermediate use”, “Value-added” and “Country 1”, respectively. The table is expressed at basic prices (borne by producers) meaning that wholesale,

retail and transport services are not part of GVCs.

Figure 8 gives a more technical perspective by introducing matrix algebra. Z is defined as the CNxCN matrix with intermediate deliveries, with CN being the number of industries in the global economy. This means that the elements zij are the intermediate inputs that go from countries-industries i on the

rows to countries-industries j on the columns. W is the RxCN matrix with R primary inputs (such as wages, salaries, employers’ contributions, capital depreciation, indirect taxes, price-decreasing subsidies, operating surplus or other income, and imports). For instance, the element wij gives the use

of primary input i, such as imports, by sector j for its production. F is the CNxK matrix with K being the number of final demands, i.e. household and government consumption and investment. Lastly, x is the total gross output vector.

The sales value of a product is the sum of the value shipped to satisfy final demands plus the value shipped as intermediary. Therefore, the equilibrium where supply is met by demand is found when the value of a good produced is equal to the value of a good or service demanded, as,

𝑥𝑖(𝑛) = ∑ 𝑓𝑗 𝑖𝑗(𝑛)+ ∑ ∑ 𝑧𝑗 𝑚 𝑖𝑗(𝑛, 𝑚), (1)

where xi(n) is the total output produced in sector n and country i, fij(n) is the amount produced to satisfy

the final use of country j, and zij(n,m) is the amount produced to satisfy the intermediate use in country

(20)

𝐱 = (𝐈 − 𝐀)−𝟏_{𝐟, (2)}

with A being the matrix CNxCN of intermediate coefficients, i.e. aij(n,m)=zij(n,m)/xj(m), I being the

identity matrix, and f being the vector of final demands constructed by multiplying the F matrix with the summation vector e, as Fe. The (I - A)-1 factor is the well-known Leontief inverse- a matrix which contains elements that give the amount sector n needs to produce for sector m to produce one unit of output.

Figure 8. Input-Output Table with Matrix Algebra

Notes: The prime in x’ is used to indicate row vectors. Vectors and matrices are indicated in bold and lower case for the former, and bold and upper case for the latter. Scalars are instead indicated by italic. The fourth quadrant - (Q4) - is assumed to be empty as it

does not influence the analysis. However, it generally includes imports by households for private consumption and several taxes. Adding the R terms in (t1) gives GDP plus imports, whereas adding the K terms in (t2) gives the amount of final demands plus exports.

3.3.2 Construction of the GVC Income Matrix

To construct the measure of GVC income, one should firstly introduce p. The latter is the vector with direct value-added coefficients, constructed by summing all primary inputs in the matrix W, as e’W with e’ being a summation vector, and then dividing each industry’s value added by its total output level, for each country. GVC income should also include the indirect contributions of sectors that are accounted for in the structure of input-output tables. Therefore, one should multiply the value-added coefficients with the Leontief inverse, as

𝐆 = 𝐩̂(𝐈 − 𝐀)−𝟏_{𝐟̂, (3)}

(21)

dimensions CNxCN, where the element in the n-th row and m-th column gives the amount of value-added generated in industry n contributed to the global value chain for product m.

The columns of the GVC income matrix reflect the total of GVCs of each industry in each country, where the latter is the country-of-completion. As I previously mentioned, I am interested in the S manufacturing sectors; therefore, the columns that do not relate to manufacturing industries-of-completion were removed from the G matrix6.

Figure 9 precisely illustrates the structure of the matrix. The row sums of matrix G (i.e. Ge) give the vector gvci and represent the total contributions of value-added of each country-industry (on the rows), i.e. the total amount of GVC income that each country-industry (on each row) generated in the world economy. The sum of all GVC incomes is h̄ and computing the shares of GVC incomes of each country to worldwide GVC income, by dividing the elements of the vector gvci by h̄, gives our measure of competitiveness (as in depicted in figure 6).

Figure 9. GVC Income Matrix

Notes: The yellow highlighter indicates that those countries (regions) host the last stage of production (i.e. countries (regions)-of-completion).

Following this reasoning, the measure of competitiveness can be split up in three determinants of competitiveness, presented in figure 6. The first one is given by the shares of value-added that countries contributed in GVCs, relative to the total size of the GVCs. For instance, it reflects the value-added contributed by China in the GVC of Northeastern textile and leather industries, with Northeast being the region-of-completion. It is possible to compute the shares of value-added by dividing the elements of the G matrix by the total sizes of GVCs- the y’ vector. Thus, we would have a H matrix (CNxCS) which elements represent the shares of value-added that each country-industry on the rows contributed to the GVCs having countries on the columns as countries-of-completion, i.e. the competitiveness within value chains, as illustrated in figure 6.

The second determinant of competitiveness refers to the countries’ sizes of GVCs relative to the total amount of GVC income generated in the world in a given year. This can be computed by simply dividing each element in the y’ vector by the total GVC income h̄, which yields the row vector b’ (1xCS). The latter represents the competitiveness of a value chain, as depicted in figure 6.

(22)

It is possible to show that the vector 𝐠𝐯𝐜𝐢 is equal to the multiplication of the two determinants of competitiveness, as:

𝐠𝐯𝐜𝐢 = 𝐇𝐛ℎ̄, (4) and dividing its elements by the total GVC income h̄ gives our measure of competitiveness.

Replacing the p vector with employment coefficients one can do the exact same reasoning to construct GVC jobs. To compute the vector of labor input coefficients q, one needs employment information for all the countries and regions in the dataset. Data for the regions are taken from Cherubini and Los (2015), whereas data for all the other countries are taken from the Socio Economic Accounts database (WIOD website). I used the Release of 2013, as it is consistent with the database constructed by Cherubini and Los (2015). Once the vector of labor inputs emp is constructed, one should divide its elements by the total output of each sector, as qij(n,m)=empij(n,m)/xj(m). Therefore, we would have:

𝐂 = 𝐪̂(𝐈 − 𝐀)−𝟏_{𝐟̂. (5)}

After having constructed the G matrix, the second step is the computation of the impacts of China and Eastern European countries increase in dominance. When considering the first determinant of competitiveness (within a value chain), one can isolate the impacts by proportionally redistributing the increase in e.g. China’s contributions of value-added to GVCs of the regions’ industries, based on the latter shares of 1995. By following this methodology, I am able to individually look at the consequences for the regions (in terms of lost value-added) had China and Eastern European countries not increased their relative shares in GVCs completed in the regions. Therefore, the first step is the computation of the increase in Chinese and Eastern European countries contributions of value-added in the GVCs completed in the regions, as:

∆𝑔_𝑖𝑗𝑛𝑠 _{= 𝑔}

𝑖𝑗𝑛𝑠(2006) − 𝑔𝑖𝑗𝑛𝑠(1995), (6)

with i being equal to China and Eastern European countries and j being equal to each Italian region. Therefore, ∆𝑔_𝑖𝑗𝑛𝑠 gives the changes in the elements of the G matrix, with rows corresponding to emerging nations’ industries and columns corresponding to each Italian region’s manufacturing industries. For instance, this means that if i represents China and j equals Northeast, then ∑ ∑_𝑛 _𝑠∆𝑔_{𝐶𝐻𝑁𝐸}𝑛𝑠 would give the total changes in the contributions of value-added from Chinese industries in GVCs having Northeast as region-of-completion.

(23)

completed in China or Eastern European countries, the regions’ value-added increased. To account for the gains, I look at the amount of value-added contributed by the regions in GVCs having China and Eastern Europe as countries-of-completion. This can be easily computed from the G matrix by summing up the elements that have the regions’ industries on the rows and the emerging countries’ industries on the column, as ∑ ∑_𝑛 _𝑠𝑔_𝑖𝑗𝑛𝑠, with i corresponding to each region and j corresponding to China and Eastern Europe. The last step concerns the evaluation of the net impact, i.e. the difference between the gained and the lost value-added for each region.

For what it concerns the isolation of the Chinese and Eastern European impacts in the second determinant of competitiveness, the same reasoning was followed. However, instead of looking at the income matrix G, one should consider the vector y’. Therefore, equation (6) would be transformed as:

∆𝑦_𝑗𝑠 _{= 𝑦}

𝑗𝑠(2006) − 𝑦𝑗𝑠(1995), (7)

with j being equal to China and Eastern European countries. Assuming j equal to China, then ∑_𝑠∆𝑦_𝐶𝐻𝑠 represents the changes in the sizes of GVCs of Chinese industries. Again, I proportionally redistributed the total increase ∑_𝑠∆𝑦_𝐶𝐻𝑠 by multiplying the latter with the elements in the b’ vector corresponding to Italian regions, as (∑ 𝑏_𝑗𝑠_(1995))

𝑠 (∑ ∆𝑦𝑠 𝐶𝐻𝑠 ), with j equal to each region.

Finally, to isolate the impacts in the last determinant of competitiveness, it is possible to use equation (6) by replacing the j with the three European countries considered- UK, Germany and France- and the i with China and Eastern Europe. After having computed the increase in Chinese and Eastern European countries’ contributions to GVCs completed in the three European countries, one should multiply the increase by the shares of the region in the same GVCs in 1995. Therefore, I would have a measure of the loss of the regions’ competitiveness within GVCs of the three European countries’ industries due to the increased contributions of Eastern European countries and China throughout the period considered.

3.3.3 Example of Northeast Textile and Leather Industries

This section illustrates the approach with an example of an apparel commodity chain, where the last stage of production takes place in Northeast Italy. As I explain in section 3.2.1, the demand for Italian clothing increases Northeastern production of final output but also the production of intermediate goods and services involved in the value chain. All these activities are usually not located in Northern Italy, but rather in either other regions of the peninsula or abroad. This means that from the value-added of gross exports in North Italy should be discarded the value-value-added by all the other stages of the production process that do not take place in that region (i.e. ‘indirect’ contributions to value-added). Following the methodology described above, I proceed in four steps, which follow the diagram of section 3.2.2.

(24)

up the elements of the gvci column vector7, where the rows correspond to Northeastern apparel industries, and then by dividing the sum by the total GVC income h̄.

As table 1 shows, the competitiveness of the apparel GVC of Northeastern Italy decreased over time. The share of its GVC income in worldwide GVC income fell from 0.12% to 0.07%. The change may be due to different reasons and the next three steps of the analysis are devoted to disentangling the three determinants of competitiveness, introduced in section 3.2.2.

Table 1. Competitiveness of the Northeastern Italy Apparel Industry

Share of worldwide GVC income (in %)

1995 2001 2006

Northeast apparel

industry 0.12 0.09 0.07

Author’s computation with data from Cherubini and Los (2015).

Notes: The numbers represent the measure of competitiveness as described in figure 6.

The first one is related to the competitiveness within the value chain. Figures 10 and 11 report information about the relative contributions of countries and regions in the GVCs of Northeastern Italy textile and leather industries. This was computed from the G matrix of figure 9 by, firstly constructing the H matrix and then taking the appropriate sums8. More specifically, the shares were computed by dividing each country or regions’ contributions to the GVC of Northeastern apparel industry by the total size of the value chain.

Figure 10 shows that, throughout the period, there has been an increase in foreign value. Contributions from China and Eastern Europe increased, as evidenced by the literature, together with non-manufacturing activities taking place in the macro-region itself. On the other hand, the textile industry of the region in 2006 seems to have played a less important role in the overall value-added relative to 1995 levels. This reflects the fragmentation of production, as discussed in the case studies mentioned above. As the stages of production were relocated abroad, the value-added of foreign countries increased, whereas the value-added by the apparel sector in the region itself decreased.

Turning to the measure of GVC jobs instead, the same approach is used for the analysis of employment patterns. In this case, I replaced the value-added by labor inputs, as in equation (5), in order to track the changes in the amount of jobs that are directly and indirectly involved in the production of manufacturing final products, at every stage of the production process.

7_{The sum can be computed as e’gvci where e’ is a summation vector taking value one for the elements corresponding to the Northeastern} apparel industries and zero otherwise.

(25)

Figure 10. GVC Income Shares of North-East Textile and Leather Products (in %)

Notes. Each block represents the shares of value added in the textile industry in Northeast contributed by specific countries and regions, as described by the legend. Figures for Northeast are decomposed in the value added by other manufacturing, nonmanufacturing activities and the textile industry in the region. Foreign captures the overall value added contributed by foreign

countries, excluding China and Eastern European countries. The Eastern European countries considered are Bulgaria, Czech-Republic, Estonia, Hungary, Latvia, Lithuanian, Poland and Romania, Slovakia and Slovenia.

Figure 11 plots the number of workers required for the Northeastern textile and leather production in 1995, 2001 and 2006. It is noticeable that the number of foreign workers increased over time, especially Eastern European workers between 1995 and 2001 and Chinese workers from 2001 until 2006. Simultaneously, the graph shows a decrease in the amount of domestic workers, from Northwest and the Southern regions in particular.

Figure 11. GVC Workers of North-East Textile and Leather Products (in thousands)

Author’s computation with data from Cherubini and Los (2015) and the WIOD database (2013 Release).

Notes. Each block represents the shares of workers involved in the textile industry in Northeast contributed by specific countries and regions, as described by the legend.

The comparison between figures 10 and 11, reveals seemingly contrasting results, in that e.g. Chinese contribution of value-added in figure 10 is much smaller than the contribution of Chinese workers in

(26)

figure 11. The apparent contradiction is due to differences in workers productivity levels; the large amount of workers displaced in the regions may simply reflect the fact that e.g. Chinese workers were less productive than Italian workers were. In other words, the share of Chinese workers over the period may have increased due to i) intensive relocation of production to China, which increased the supply of jobs, or ii) low labor productivity, which meant that a numerous labor force was required. As it is not possible to distinguish between these two factors, looking at GVC jobs would give a biased picture. Therefore, my analysis concentrates predominantly on the measure of GVC income and provides implications for GVC workers.

In order to isolate the impact of Chinese and Eastern European countries, I computed the amount that was lost as (ℎ_{𝑁𝐸𝑁𝐸}𝑡𝑒𝑥𝑡.𝑡𝑒𝑥𝑡._{(1995))(∑ ∆𝑔}

𝑖𝑁𝐸𝑛𝑡𝑒𝑥𝑡.)

𝑛 , with ∑𝑛∆𝑔𝑖𝑁𝐸𝑛𝑡𝑒𝑥𝑡. calculated as in equation (6), “text.”

standing for the textile industry of Northeast and i corresponding to China and Eastern Europe. The first factor represents the share of the Northeastern apparel industry in the GVC of the region’s apparel industry, whereas the second factor reflects the total increase in Chinese (Eastern European) contributions in the same GVC between 1995 and 2006.

As mentioned in section 3.3.2, the gains in value-added should also be taken into account. To do so, I consider the contributions of the Northeastern apparel industry in the GVCs having the emerging economies’ as countries-of-completion, and then I evaluate the net impact (table 2). Reasoning in terms of the GVC income matrix G, one should sum the elements in the matrix that correspond to the rows of each region and to the column of Chinese and Eastern European industries for each year. By taking the average over the three years, one would have the average amount of value-added that each region contributed to the GVCs completed in emerging economies.

Table 2. Net Impact of GVC Income between 1995 and 2006

GVC income (in million US$) (I) _{Northeast textile’s contributions in GVCs of Chinese industries} ₇₂ (II) _{China’s contributions in GVC of Northeastern textile industries} _-94

(I-II) _{Net impact} _-22

(I) _{Northeast textile’s contributions in GVCs of Eastern European industries} ₁₃₁ (II) _{Eastern Europe’s contributions in GVC Northeastern textile industries} _-60

(I-II) _{Net impact} ₇₁

Notes: The table reports information about the impacts in the first determinant of competitiveness as described in figure 6 and the net impacts. Rows (I) reflect the average GVC income gained by Northeast textile industry between 1995 and 2006.

(27)

GVC income. This means that the China shock had more detrimental consequences on the competitiveness of the Northeastern textile industry relative to the Eastern European shock.

The third step represents the second determinant of competitiveness, that is the competitiveness of the GVC of Northeastern apparel sector. This is calculated as described in section 3.3.2, that is by multiplying the elements in the b’ vector9_{corresponding to Northeastern textile industry, with the total}

changes of the sizes of GVCs completed in China (Eastern Europe), i.e. (𝑏_𝑁𝐸𝑡𝑒𝑥𝑡._{(1995))(∑ ∆𝑦} 𝑗𝑠.) 𝑠 , with

∑ ∆𝑦_𝑗𝑠.

𝑠 being calculated following equation (7) and j representing China and Eastern Europe.

Table 3 shows the value lost due to size increases in GVCs having China and Eastern Europe as countries-of-completion. It appears that, had China and Eastern Europe not increased their sizes of GVCs, the GVC of Northeastern Italy textile products would have retained 21 million US$ in GVC income. This means that the increase in prominence of emerging economies in the world market had detrimental impacts on the relative amount of textile final products completed in the region.

Table 3. Total Value Lost by GVC of Northeastern Apparel Products, 1995-2006

Caused by China shock Caused by Eastern

Europe shock Total

-17.08 -3.92 -21

Notes: The table reports information about impacts in the second determinant of competitiveness as described in figure 6. Figures are in million US$.

The last step is about the last determinant of competitiveness, i.e. the contributions in GVCs having European countries as countries-of-completion. This was computed by looking at the shares of China and Eastern European countries’ value-added in the GVCs of German, French and British products. The equation used is thus equal to equation (6), but it considers the columns of the European countries’ industries, rather than the ones of the regions’ industries, and the rows correspond to Chinese and Eastern European industries. The increase in the shares of the emerging countries was then redistributed to Northeastern textile industry, proportionally to its contribution in 1995. Results are shown in table 4 and represent the amounts that the apparel industry in Northeast would have retained had China and Eastern Europe not expanded trade to the Western world.

Table 4. GVC Income of Northeastern Textile Industry Lost in GVCs of European Countries’ Products,

1995-2006 France columns in GVC income matrix Germany columns in GVC income matrix UK columns in GVC

income matrix Total

-2.10 -7.95 -1.10 -9.15

Notes: The table reports information about impacts in the third determinant of competitiveness as described in figure 6. Figures are in million US$.

(28)

It appears that the increase in contributions by Chinese and Eastern European value-added meant that the GVC of Northeast apparel industry lost approximately 9.15 million US$ in GVC income. The major loss was to Germany. This signal a trend of the Western world increasingly preferring to involve Chinese and Eastern European countries in the production process of their final goods and services as it is cheaper to undertake activities in those countries.

3.3.4 Limitations of the Methodology

Besides the advantages highlighted in section 2.3.1, taking an input-output approach has also two main limitations, which are: i) the level of product aggregation, and ii) the assumptions of homogeneity in productivity levels. The first one refers to the fact that input-output tables contain aggregated information on final products. For, instance it is not possible to differentiate a pricey piece of clothing from a cheaper apparel final product in the textile and leather industries, even though the latter have very different GVCs. This entails that input-output tables provide only average statistics without high degrees of detailed information.

The second aspect refers to the assumptions that, within an industry, firms use homogeneous production methods, thus, have the same level of productivity and the same input mix. Indeed, a column in an input-output table represents the average of the production structure of all firms that operate in a particular industry (Timmer et al., 2015). This assumption is not realistic as many scholars revealed. For instance, Chen et al. (2012) and Koopman et al. (2012) studied Chinese firms and found significant heterogeneity in the production structures of firms that operate only domestically and firms that export in the international market. Similarly, Altomonte et al. (2013) provide evidence of the differences in technologies used by exporters and non-exporters. De la Cruz et al. (2011) found similar results when comparing Mexican maquiladoras with non-processing firms of the same country. In general, these studies give evidence to the fact that the import content of the exports differs significantly between exporters and firms that are involved in domestic production only. Not considering the differences in input mix between sectors can generate aggregation errors that bias the sectors’ contributions of value-added. Nomaler and Verspangen (2014) argue that these biases result in an overestimation of the global distribution of value-added generation, and thus misrepresent the average value-added generated in GVCs10.

(29)

Chapter 4 Changes in Regional Competitiveness

4.1 Preliminary Findings

The framework underlined in the previous section can be expanded in order to include all manufactures final products and all macro-regions of the country. Table 5 reports the relative shares of the regions’ GVC incomes to national and worldwide GVC income. Firstly, it is important to note the large gap between the shares of value-added of Northern regions, especially in Northwest, relative to the South. For instance, in 1995, the former accounted for 41% of national GVC income and 1.7% of worldwide GVC income, whereas the latter generated only the 15% and 0.6% of total national and worldwide GVC income. The second aspect to note is the decrease in every region of their share contributing to total GVC income generated in the world, i.e. a decrease in regional competitiveness that confirms the first hypothesis proposed in section 2.3.2. Interestingly, this decrease was particularly pronounced in Northern regions, casting doubts on their ability to cope with competitive pressures.

Table 5. Regional GVC Income (GVCI)

Share of Italian GVCI Share of worldwide GVCI 1995 2001 2006 1995 2001 2006 Northwest 41 39 41 1.7 1.3 1.2 Northeast 26 27 26 1.1 0.9 0.8 Central 18 18 18 0.7 0.6 0.5 South 15 16 15 0.6 0.5 0.4 Total 100 100 100

Notes: Figures represent the shares of GVC income that each macro-region contributed to the production of final manufactured products, where the last stage of production took place anywhere in the world. The table represents the measure of competitiveness as

depicted in figure 6 and as computed for the Northeastern apparel industry in table 1.

The Effects of Trade Shocks on Regional