Mapping CO2 emissions for the production of automobiles

1

Mapping CO2

emissions for

the production

of automobiles

Supervisor: C.J.J. Jepma Co-assessor: H.W.A. Dietzenbacher

2 Table of Content

Abstract ... 3

1. Introduction ... 4

2. Methodology ... 7

3. Data – World Input-Output Database ... 10

4. Carbon footprint by country ... 13

5. Trends in international fragmentation and corresponding CO2 emissions ... 18

6. Trends in CO2distribution generated for production ... 22

7. Trends in CO2emissions by region ... 29

7.1 Hypotheses and model ... 29

7.2 Variables and data sources ... 31

7.3 Estimation results ... 32

8. Conclusion ... 34

9. Limitations ... 36

10. References ... 37

3 Abstract

The aim of this paper is to examine whether the automobile industry has been successful in decreasing the carbon dioxide emissions for the production of automobiles. Furthermore, it analyses whether countries are offshoring “dirty” work, meaning whether countries are offshoring those production activities that emit relatively more carbon dioxide than they produce themselves. This analysis is undertaken for 40 countries in the automobile industry between the years 1995 and 2009. Overall, it is found that domestic as well as foreign countries have been decreasing its emissions of carbon dioxide generated for the production of automobiles. Also, countries increasingly offshore “dirty” work. Furthermore, it is found that European Union and NAFTA countries emit much less carbon dioxide per dollar of production than Asian countries do.

4 1. INTRODUCTION

Global warming has been happening since the industrial revolution and has had and still does have major implications on the world’s agriculture, the economy, the environment and on human health. Furthermore, climate scientists state that any positive impacts of global warming are far outweighed by the negative impacts (IPCC, 2007). For instance, rising temperatures cause natural snow-covered land such as in Greenland to melt, which then can be used for agricultural purposes to feed us human beings. On the other side, rising temperatures also causes stable sea ice to melt, causing the natural habitat of animals such as polar bears and penguins to shrink. It also causes water shortages, expanding deserts, more frequent and more intense wildfires, hurricanes and storms to become stronger and floods and droughts to become more common. Therefore, it is very important that we take global warming seriously.

There is no doubt that the emissions of carbon dioxide have been one of the major causes of global warming. Global warming is mainly caused by the emissions of greenhouse gases (GHG) created through human activities. Of the total emitted greenhouse gases, 72% is carbon dioxide (CO2), while this is only 18% and 9% for methane (CH4) and nitrous oxide (NO2) respectively. Carbon dioxide is mainly produced by burning fossil fuels and causes 60% of the anthropogenic (i.e. human-emitted) greenhouse effect (Achtnicht, 2012).

The transport sector1 is one of the main emitters of carbon dioxide due to its intensive use of fossil fuels. In 2013, the transport sector contributed approximately 19% of total carbon dioxide emissions within the EU-282. Road transport causes most of the emissions in this sector. Between 1990 and 2013, carbon dioxide emissions from road transportation increased by 14% in the EU-28. This increase in CO2 emissions is due to intensive fossil fuel consumption, which increased by 22% between 1990 and 2013 (EEA, 2015). Moreover, passenger cars account for more than half of the total carbon dioxide emissions caused by the transport sector.

Consequently, the individual transport vehicles play a major role in the political debate on climate change. The European Commission (EC) has set the goal of reducing GHG by 20% by 2020, compared to the emission values of 1990 (EC, 2008). To ensure that EU members

1

The transportation sector includes the movement of goods and people by cars, trucks, trains, ships, airplanes and other modes of transport.

2

5

will achieve those climate goals, the European Parliament approved the EU’s energy and climate package. A part of this package is a regulation that addresses emissions performance standards for new passenger cars registered in the EU. This regulation stipulated that the whole car industry had to comply with an average of 140 grams of carbon dioxide per km in 2008 (Launay, 2013). In reality, the EU countries did not achieve this target. In 2008, new passenger cars were emitting an average of 153 grams of CO2 per km (T&E, 2009). The long-term target of the EC for 2020 is an average of 95 grams of carbon dioxide per km.

This matter concerning the reduction of carbon dioxide emissions in the automobile industry has been well studied. Previous studies mostly focus on the reduction of CO2 emissions that are generated while driving transport vehicles and the effects of the CO2 reductions. Alessandrini et al. (2012) for example concluded that if drivers had adopted the new driving style, called eco-driving, CO2 emissions would have been up to 30% lower than the measured average. Samaras and Meisterling (2008) find that plug-in hybrid electric vehicles reduce GHG by 32% compared to conventional vehicles. Wegener (1996) showed that a combination of policies to improve the quality of public transport and to increase the costs of car travel would result in a significant reduction of energy use and CO2 emissions of urban transport. King (2007) state that when car manufacturers optimally use the newest vehicle and fuel technologies, almost complete decarbonisation of road transport is possible by 2050. Woodcock et al. (2009) noted that more active travel (walk, run, bike, etc.) and the use of lower-emission motor vehicles would give major health benefits. Huo et al. (2014) conclude that if China does not take matters into hands to reduce the current CO2 emissions, the uncontained growth in motor vehicles will have severe consequences for oil use, CO2 emissions and in turn human health in China. The World Bank (1997) estimated that in 1995 178,000 premature deaths in urban China were caused by China’s air pollution.

6

This study will try to investigate how much carbon dioxide is emitted corresponding the production in the automobile industry and whether the level of produced carbon dioxide has decreased over time. More specifically, I will try to investigate how much carbon dioxide emissions are generated per dollar of production and analyze whether this has changed over time. To do so, the production levels for each country and the corresponding carbon dioxide emissions have to be determined, as well as what amount is produced domestically and abroad, which is known as ‘fragmentation’3. Therefore, I will use the same method as Los et al. (2015), whom for each final product define its value chain as the total of all value-added activities that are needed in its production. Los et al.’s method also makes it possible to trace the location of these value-adding activities. In the case of a car, value is added when the car is assembled in for example Germany, but also intermediates that are delivered to the assembly factory generate value added, which could partly be done domestically and partly abroad. By using an input-output model of the global economy, the value of a car can be fully decomposed into domestic value added and value added abroad, the latter will be referred to as “foreign value added”. The corresponding levels of carbon dioxide emissions can be obtained using the same method as Los et al., however with a small modification. Furthermore, I will try to find out whether countries are offshoring “dirty” work, meaning offshoring those production activities that emit relatively more carbon dioxide than the activities that are produced by themselves.

The remainder of this paper is structured as follows. In section 2, I introduce the methodology to test whether countries have been successful in decreasing the carbon dioxide emissions. Section 3 covers a brief description of the data used. For this analysis,data from the World Input-Output Database (WIOD) is used, because WIOD provides systematic data of the production in the automotive industry as well as its corresponding CO2 emission values. Data concerning production that is done in a country itself as well as what is imported by that country is available from 1995 until 2001. Its corresponding carbon dioxide data however are only available until 2009. Therefore, the timespan from 1995 until 2009 will be used. The WIOD covers 40 countries and provides the opportunity to analyse developments in three regions: the European Union (EU) (incorporating the 27 countries that were EU members in 2011), NAFTA (incorporating Canada, Mexico and the United States), and South East Asia (incorporating China, India, Indonesia, Japan, South Korea, and Taiwan). In section 4, it is found that the demand for automobiles increased considerably over the years. Furthermore,

3

7

almost all countries have been responsible for less CO2 emissions generated for the production of an average car. Section 5 shows that the automotive value chains became more internationally fragmented and increasingly offshored “dirty” work since 1995. In section 6, it is found that the automotive value chains have been decreasing its average carbon dioxide emissions domestically as well as abroad. Armed with these insights, I investigate whether distance matters concerning offshoring “dirty” work and whether there is a bias to specific regions. My longitudinal analysis shows that countries have higher CO2 values corresponding foreign production activities when distance decreases. Also I find that the average emissions generated for the final demand of automobiles are increasingly lower in EU countries as well as NAFTA countries compared to Asian countries. Section 8, concludes.

2. METHODOLOGY

To estimate international fragmentation, the part of the production process that is produced abroad has to be calculated. Therefore, I used the same method as Los et al. (2015). Their disentanglement of production processes is based on value chains of finalized products identified by the last stage of production, which they define as the country-of-completion. Intermediate inputs are needed to produce the final product, which can be produced domestically, generating domestic value added, or imported from foreign firms. The firms producing those intermediate inputs in turn add value (first-tier suppliers), but because the first-tier suppliers also require parts and components for their production, second-tier suppliers also add value by providing the required parts and components.

8

𝑮𝑮 = 𝒗𝒗� ∗ (𝑰𝑰 − 𝑨𝑨)−𝟏𝟏_{∗ (𝑭𝑭𝑭𝑭) (1)}

This equation has been a standard tool in input-output analysis for over decades.4 G represents the vector of value added for each industry that has contributed to the production of finalized products in a specific value chain. v represents value added coefficients expressed as a ratio over gross output for each of the industries and countries.5 Value added is defined as the difference between the amounts for which products are sold in each stage minus total intermediate inputs used in that stage. The Leontief inverse matrix (I-A)-1 is incorporated to

ensure that value added contributions for first-tier-suppliers as well as other-tier-suppliers are accounted for. The value chain that is considered is determined by the choice for a specific final demand or final output matrix F. Final demand is the final output delivered for household consumption, government consumption and investment demand for domestic as well as foreign use. e represents a summation vector.

Equation (1) makes it possible to decompose the value of a final product into value-added contributions from all industries and countries incorporated in the input-output table that is used. Since I will use the world input-output table from the World Input-Output Database containing 35 industries and 40 countries all over the world, the decomposition is exhaustive. The final output value of product (i,c), denoted as FO(i,c), can be decomposed into value-added contributions by industry and country, denoting country and its production by k and VA(k)(i,c) respectively. The vector G contains the matching value added levels (VA(k)(i,c)) for each product group (i,c). When summed over all countries, the final output value of product (i,c) will be equal to the value-added contributions to the production of (i,c), such that

𝐹𝐹𝐹𝐹(𝑖𝑖, 𝑐𝑐) = ∑ 𝑉𝑉𝑉𝑉(𝑘𝑘)(𝑖𝑖, 𝑐𝑐)𝑘𝑘 (2)

Foreign value added is measured by summing all value added activities outside the country-of-completion (c):

𝐹𝐹𝑉𝑉𝑉𝑉(𝑖𝑖, 𝑐𝑐) = ∑𝑘𝑘≠𝑐𝑐𝑉𝑉𝑉𝑉(𝑘𝑘)(𝑖𝑖, 𝑐𝑐)= 𝐹𝐹𝐹𝐹(𝑖𝑖, 𝑐𝑐) − 𝑉𝑉𝑉𝑉(𝑐𝑐)(𝑖𝑖, 𝑐𝑐) (3) And its share is expressed as a share of all value added in the production of (i,c):

𝐹𝐹𝑉𝑉𝑉𝑉𝐹𝐹(𝑖𝑖, 𝑐𝑐) = 𝐹𝐹𝑉𝑉𝑉𝑉(𝑖𝑖, 𝑐𝑐)/𝐹𝐹𝐹𝐹(𝑖𝑖, 𝑐𝑐) (4)

4

For a more detailed and technical description of Equation (1) see Miller and Blair (2009). 5

9

To find out how much carbon dioxide is emitted due to the production of a particular product group (i,c), the matching carbon dioxide levels need to be found. These can be obtained in the same way as Equation (1), but with a small modification. First, G will be modified by EM, representing the vector of emissions of carbon dioxide created in each industry that has contributed to the production of finalized products in a specific value chain. Next, instead of using value added coefficients v, emission coefficients d will be used expressed as a ratio over gross output for each of the industries and countries. This results in the following formula:

𝑬𝑬𝑬𝑬 = 𝒅𝒅� ∗ (𝑰𝑰 − 𝑨𝑨)−𝟏𝟏_{∗ (𝑭𝑭𝑭𝑭) (5)}

The vector EM contains the emissions of carbon dioxide corresponding the value added levels for each industry (i,c). Denote the emissions of carbon dioxide by country (k) and its production by CD(k)(i,c). The foreign generated emissions will be expressed by ∑𝑘𝑘≠𝑐𝑐𝐶𝐶𝐶𝐶(𝑘𝑘)(𝑖𝑖, 𝑐𝑐) and its corresponding share will be expressed as a share of total emissions generated in the production of (i,c):

𝐹𝐹𝐶𝐶𝐶𝐶𝐹𝐹(𝑖𝑖, 𝑐𝑐) = ∑𝑘𝑘≠𝑐𝑐𝐶𝐶𝐶𝐶(𝑘𝑘)(𝑖𝑖, 𝑐𝑐)/ ∑ 𝐶𝐶𝐶𝐶(𝑘𝑘)(𝑖𝑖, 𝑐𝑐)𝑘𝑘 (6)

To measure the importance of carbon dioxide emissions due to the production of a specific product (i,c) and to make it comparable over countries, emissions are expressed as a ratio over value added for each country (CDR represents carbon dioxide ratio).

𝐶𝐶𝐶𝐶𝐶𝐶(𝑐𝑐) = 𝐶𝐶𝐶𝐶(𝑐𝑐)(𝑖𝑖, 𝑐𝑐) 𝑉𝑉𝑉𝑉(𝑐𝑐)(𝑖𝑖, 𝑐𝑐)⁄ (7)

10

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶(𝑖𝑖, 𝑐𝑐) = 𝐶𝐶𝐶𝐶(𝑐𝑐)(𝑖𝑖, 𝑐𝑐) 𝑉𝑉𝑉𝑉(𝑐𝑐)(𝑖𝑖, 𝑐𝑐)⁄ (8)

Then, the ratio of emissions over value added by foreign countries (FDCR) are determined by all emissions generated abroad for the production of good (i,c) divided by all value added generated abroad:

𝐹𝐹𝐶𝐶𝐶𝐶𝐶𝐶(𝑖𝑖, 𝑐𝑐) = ∑𝑘𝑘≠𝑐𝑐𝐶𝐶𝐶𝐶(𝑘𝑘)(𝑖𝑖, 𝑐𝑐)⁄∑𝑘𝑘≠𝑐𝑐𝑉𝑉𝑉𝑉(𝑘𝑘)(𝑖𝑖, 𝑐𝑐) (9)

The measures used above have a number of important characteristics. First, the shares are bounded between zero and one. The share of domestic value added and emissions will never be equal to zero as the final stage of production by definition takes place in the country-of-completion, and therefore must involve some value added contribution which is associated with carbon dioxide emissions. Second, the value added and emissions contributions of countries are based on value added, which means that it does not depend on the stages of production or the sequence of those stages. Therefore, the sequence of those stages in the production chain is inconsequential for measuring the domestic and foreign shares. Third, value added and emissions are not measured based on the ownership of production factors, but on the location of production. It therefore does not necessarily measure the geographical distribution of income, but that of value added. Finally, ratios provide a standardized method to compare countries, because ratios put all countries on a relatively equal playing field.

3. DATA - WORLD INPUT-OUTPUT DATABASE

The computation of value added contributions relies on Equation (1), which as said before requires the availability of global input-output tables (GIOTs). Furthermore, to investigate how much carbon dioxide is emitted during production activities in the automobile industry, corresponding environmental accounts are needed. Recently, databases have become available containing GIOTs, such as those from the Economics Co-operation and Development (OECD), Eora, Asian Development Bank (ADB), etc., but some of which do not have systematic data available or do not have corresponding environmental accounts.6 In this research the newly developed World Input-Output Database (WIOD) is used, because it is the first public database that contains new information on the consequences of the location of

6

11

production due to international trade and it provides the opportunity to analyse the consequences of fragmentation on employment, value added, investment patterns and most importantly environmental issues (Timmer, 2012). The WIOD provides time-series of GIOTs and corresponding socio-economic and environmental accounts covering 35 industries in 40 countries worldwide plus an estimation for the remaining countries that are not included in the WIOD named “Rest of the World”.7 The database became available in 2013 and covers the period 1995-2009.

But what exactly is a global input-output table? A global input-output table, also known as a world input-output table (WIOT), is a combination of national input-output tables (NIOTs), with the addition that the use of products is broken down according to their industry and country origin. A WIOT is able to show whether (intermediate) products are produced by its domestic industries or by foreign industries, and in particular in which foreign industries the (intermediate) products have been produced. So, a WIOT can be seen as a description of the internationally fragmented production processes in the world. In Figure 1 the structure of a WIOT is displayed with n countries, representing the world economy. Each row cell in the WIOT indicates the value of output deliveries from a particular industry in a country. This can be for intermediate use (blocks labelled Z) or final use (blocks labelled F). In both cases it can be for domestic or foreign use. Each column contains information on the technology of production of a particular industry in a country, since each column indicates how much intermediate inputs are needed for its production. Besides importing and exporting, industries also add value when they use domestic factors to produce (intermediate) output. The

FIGURE 1: A stylized World Input-Output Table Intermediate deliveries (s

columns per country)

Final demand (c columns per country) Total 1 … n 1 … n s industries, country 1 Z11 Z1. Z1n F11 F1. F1n x1 … Z.1 Z.. Z.n F.1 F.. F.n x. s industries, country n Zn1 Zn. Znn Fn1 Fn. Fnn xn Value added (v1)’ (v.)’ (vn)’ Output (x1)’ (x.)’ (xn)’ Emissions (d1)’ (d.)’ (dn)’ Note: An apostrophe indicates a row vector.

7

12

difference between intermediate inputs and final output (blocks labelled as x) is value added (blocks labelled v’). An important accounting identity in a WIOT or any other input-output table (IOT) is that total use of output (Z+F) in a row of a particular industry in a country equals total value of input including intermediate products and value added such as labour and capital (Z+v) in a column of that same industry. Furthermore, production is most often associated with the emissions of carbon dioxide, which are labelled as d.

The WIOD’s WIOTs are constructed by combining national supply and use tables8(SUTs) with detailed bilateral international trade statistics. The benefit of using SUTs over IOTs is that SUTs provide information on both products and industries, while IOTs are either product or industry based. This is crucial when linking it to international trade data and socio- economic and environmental accounts, because international trade data is mainly product-based, while socio economic and environmental data is mainly industry-product-based, which makes linking data more natural in a SUT framework.

The next challenge was to ensure consistency and comparability of the WIOTs over time. Typically, national SUTs are only available for a limited amount of years and once released revisions are rare, which is in contrast with National Accounts Statistics. Through this, substantial discrepancies can occur between the latest version of the National Accounts for a particular year and the published SUT for that same year. Therefore, the SUTs used for the WIOD are benchmarked on the National Accounts. To achieve time consistency, time series for intermediate inputs, gross output and value added by industry, total imports, total exports and final use from the NAS were used as constraints with the so called SUT-RAS method when generating times series of SUTs.9

Another challenge was the allocation of imports of goods to a use category in SUTs and the allocation to industry and country origin. The standard assumption in most databases is to apply the import proportionality assumption, which assumes that every sector imports in the same proportion as its economy wide-use. Studies have found that this assumption can lead to biases (Winkler and Milberg, 2012). Therefore, the same method that the UN COMTRADE database applies to allocate imports is used. The UN COMTRADE database provides trade data for about 5000 products and allocates imports to three use categories: “intermediate consumption”, “final consumption” and “capital goods”. The share of use category is used to

8

SUTs are a combination of supply and use tables. Supply tables provide information on products that are produced by each domestic industry and use tables indicates the use of each product by industry or final user. SUTs are the core statistical sources from which national input-output tables are derived by the national account statistics (NAS).

9

13

split up total imports across the three categories, which generates an import table. Then, each cell in the import table is split up to the country of origin, which unlike in the case of using the standard proportionality assumption might cause different import shares across end-use categories.

The data in the WIOD’s WIOTs are expressed in millions of dollars and in current U.S. dollars using official exchange rates from the IMF to convert SUTs, since values in SUTs are originally in national currencies. All tables are valued at basic prices, which means that prices are valued as the amount received by the producer from the purchaser for a good, unit or service produced as output (including intermediate products). It excludes net taxes and trade and transportation margins. These margins are recorded to the respective trade and transport industries. The main reason to opt for basic prices is that the basic price concept reflects best the underlying cost structures of industries, because the use of trade and transport services and the use of goods are clearly separated. The WIOD’s environmental account tables, containing the total carbon dioxide emissions emitted per industry and country, are expressed in kilotons (Gg) of CO2 emissions.

4. CARBON FOOTPRINT BY COUNTRY

In this section, the average carbon footprint of a car will be investigated for 40 countries for the years 1995 and 2009. The countries are grouped by the following 3 regions: The EU10 (including all European Union members as of 2011), South East Asia (including China, India, Indonesia, Japan, South Korea and Taiwan), NAFTA (including Canada, Mexico and the United States) and a remaining category labelled other (including the remaining countries). This grouping of countries will be used throughout the paper. As said before, a carbon footprint is defined as the total amount of greenhouse gases which were induced for a particular individual, event, activity, organization or a product. It has become a widely used tern to map the responsibility of produced carbon dioxide for a specific event or product, which is in this case the production of an average car. To determine the average carbon footprints for each country, the average manufacturing price of a car is needed for both years.

10

14

According to DailyFinance11 the average selling price of a car were $17,518 and $19,395 in 1999 and 2009 respectively. Since I need the average price of a car for the year 1995, I assume a price inflation rate of 1%, which results in a price of around $16,834. Market Realist12 decomposed the cost of a car into cost components such as raw materials, labor, research and development, advertising and logistics. On average 70% of the selling price is necessary for the production of a car. Therefore, I assume that the production costs of an average car are around $11,784 and $13,577 in 1995 and 2009 respectively.

I estimate the average carbon footprint by a country-of-completion as the total emissions generated for the final demand (i,c) of a country-of-completion k divided by the total value added created in each of the industries s and countries n times the price of an average car. This results in the following formula:

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐹𝐹𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝐶𝐶𝐶𝐶 = ∑ 𝐶𝐶𝐶𝐶(𝑘𝑘)(𝑖𝑖, 𝑐𝑐)𝑘𝑘 / ∑ 𝑉𝑉𝑉𝑉(𝑘𝑘)(𝑖𝑖, 𝑐𝑐)𝑘𝑘 ∗ 𝑃𝑃𝐶𝐶𝑖𝑖𝑐𝑐𝑃𝑃 (10)

Table 1 shows the average carbon footprints of a car for 40 countries for the years 1995 and 2009, based on equation (10). The first three columns present the final output value of automobiles for the years 1995 and 2009, and the change between those two years respectively. The next three columns show the total carbon dioxide emissions for which each country-of-completion is responsible for as well as the change over time. The last three columns presents the average carbon footprints for each country-of-completion. The results are grouped by region and sorted by the change in the average carbon footprint in column (9), from highest to lowest change, with the corresponding final demand and total carbon dioxide emissions in the previous columns.

15

relative terms. This can be seen by the major decline in the average carbon footprint of Romania (87.0%, column 9). This is mostly related to the fact that Romania faced stringent environmental regulations from 2007 onwards, due to its entrance to the European Union. Therefore, Romania became more aware of the emissions generated for the production of a car.

A number of observations clearly stand out from Table 1. Column 3 reveals that the final demand for automobiles increased considerably. Virtually all countries-of-completion show major increases in the total final demand for automobiles. This is obviously related to economic growth. At first sight, column 6 reveals that more than half of all the incorporated countries have decreased the total emissions in absolute terms. When comparing the change in total final demand and the change in total CO2 emissions, the final demand increased more for almost all countries, indicating that virtually all countries have been responsible for less carbon dioxide in relative terms. This can also be seen when looking at the differences in the average carbon footprints (column 9). Almost all countries have decreased their average carbon footprints, indicating that almost all countries-of-completion have been responsible for less carbon dioxide emissions generated during the production processes of a car. This is no surprise, since people all over the world are becoming more and more aware of the devastating effects of carbon dioxide emissions. Hence, it can be concluded that the final demand for automobiles increased considerably over the years. Furthermore, almost all countries have been responsible for less carbon dioxide emission generated for an average car over the period 1995 and 2009. These general trends are found for all three regions, however there are also some clear differences, which will be addressed next.

16

Table 1: Average carbon footprints of an average car by country

Total final demand Total CO2 emissions Average carbon footprint

1995 2009 1995-2009 1995 2009 1995-2009 1995 2009 1995-2009 (1) (2) (3) (4) (5) (6) (7) (8) (9) European Union Romania 927 6989 653.8 3778 3224 -14.7 48,027 6,263 -87.0 Bulgaria 206 434 110.4 1818 630 -65.3 103,965 19,734 -81.0 Slovakia 508 7407 1358.2 869 2169 149.6 20,165 3,977 -80.3 Lithuania 44 387 776.0 78 125 60.3 20,761 4,378 -78.9 Poland 3746 15550 315.1 8314 7111 -14.5 26,157 6,209 -76.3 Latvia 36 166 358.3 66 66 -0.1 21,572 5,421 -74.9 Czech Republic 1783 15504 769.8 2667 5390 102.1 17,634 4,720 -73.2 Estonia 1 135 9361.7 4 97 2494.5 30,996 9,793 -68.4 Hungary 1430 5929 314.7 1403 1741 24.1 11,562 3,986 -65.5 Cyprus 19 36 93.2 102 77 -24.5 64,151 28,869 -55.0 Portugal 2927 4370 49.3 1332 942 -29.3 5,363 2,925 -45.5 Italy 27843 42856 53.9 14462 11345 -21.6 6,121 3,594 -41.3 Luxembourg 22 103 373.5 9 23 161.5 4,679 2,977 -36.4 Spain 29156 49785 70.8 13186 12634 -4.2 5,329 3,445 -35.4 Austria 3394 7694 126.7 1250 1672 33.8 4,338 2,950 -32.0 Ireland 548 1493 172.4 307 495 61.2 6,600 4,499 -31.8 Finland 2141 2601 21.5 1040 752 -27.7 5,723 3,926 -31.4 Netherlands 8729 9070 3.9 3657 2276 -37.8 4,937 3,407 -31.0 Slovenia 863 2121 145.9 455 677 48.8 6,210 4,331 -30.3 Denmark 1656 1269 -23.4 567 265 -53.2 4,038 2,840 -29.7 Greece 1035 1799 73.8 708 753 6.2 8,063 5,680 -29.6 Great Britain 39317 49705 26.4 17658 14205 -19.6 5,293 3,880 -26.7 France 75543 115478 52.9 24059 23555 -2.1 3,753 2,769 -26.2 Germany 127727 215384 68.6 51572 57619 11.7 4,758 3,632 -23.7 Belgium 19715 16839 -14.6 7469 4433 -40.7 4,464 3,574 -19.9 Sweden 12914 15802 22.4 3676 3470 -5.6 3,355 2,981 -11.1 Malta 89 108 21.2 53 54 1.1 7,000 6,725 -3.9

South East Asia

China 23740 243033 923.7 83481 279455 234.8 41,439 15,612 -62.3 Indonesia 17663 45150 155.6 49708 67244 35.3 33,163 20,221 -39.0 Korea 36425 74429 104.3 24010 44944 87.2 7,768 8,198 5.5 Taiwan 10540 7320 -30.5 6048 4135 -31.6 6,762 7,669 13.4 Japan 170542 159946 -6.2 43654 47898 9.7 3,016 4,066 34.8 India 13383 18266 36.5 3987 7820 96.1 3,511 5,812 65.6 NAFTA Canada 45686 68018 48.9 25361 21652 -14.6 6,542 4,322 -33.9 Mexico 19338 39178 102.6 11486 13914 21.1 6,999 4,822 -31.1 United States 266543 313784 17.7 156796 116374 -25.8 6,932 5,035 -27.4 Other Russia 8223 22882 178.3 25346 27340 7.9 36,323 16,223 -55.3 Australia 6478 14051 116.9 4212 5462 29.7 7,662 5,278 -31.1 Turkey 7853 11036 40.5 4049 4515 11.5 6,076 5,555 -8.6 Brazil 23567 68307 189.8 6644 15752 137.1 3,322 3,131 -5.8

Note: Total final demand for transport equipment (in millions of US $), total corresponding emissions (in Gg of CO2 emissions) and average carbon footprint (in Gg of CO2 emissions), by country-of-completion. Shares are in percentages and rounded to one decimal place. Entries are grouped by region and sorted on the change in the average carbon footprint.

17

no surprise that the newly entered EU member states show such significant differences in their average carbon footprints. When looking at the carbon footprints of the core EU members and the newly entered EU members, it can be seen that the core EU members show significantly lower values than the new EU members. This indicates that the new EU members are on the right track regarding the reduction of CO2 emissions, however they still have a long way to go.

Also, it should be noted that in 2009 the average carbon footprint of the core EU member such as Austria, Denmark, France, Portugal, and Sweden are significantly lower than the other countries. Even NAFTA countries show significantly higher carbon footprint values. This might be related to the fact that the GHG performance of the U.S. cars lags considerably behind most other nations (An and Sauer, 2014).

The South East Asian region seem to be the only region that is responsible for more emissions generated for the production of an average car with the last production stage taking place in these Asian countries. 4 out of 6 South East Asian countries show an increase in their average carbon footprints (column 9). This should not be surprising since the production processes in Asia are often more carbon-intensive as the same production processes in Europe. The Chinese and Indonesian automotive value chain show a reduction in their carbon footprints, indicating that these two value chains have been able to reduce total CO2 emissions generated for an average car for which the last production stage takes place in China or Indonesia. However, when the carbon footprints of the South East Asian countries are compared to each other, it can be seen that the Chinese and Indonesian carbon footprints are significantly higher than those of Korea, Taiwan, Japan and India. This might be related to the fact China’s and Indonesia’s automotive industry has only been developing itself recently. For example China’s automobile industry started only in the 1950s and continued to have small volumes over the first 30 years (Zhang, 2014). Japan, on the other hand, already produced the first manufactured bus in 1904.13 Therefore, Japan has been developing itself long before China started producing cars. Also Japan faces the one of the most stringent environmental regulations.

At last, it should be noted that the European Union countries still have to take measures in order to reduce the GHG emissions. As said in the Introduction of this paper, the EC has set the goal of reducing GHG by 20% by 2020, compared to 1990. When looking at the difference in total emissions of the EU-27, it can be seen that GHG in the 27 EU member

13

18

states only decreased by a total of 3.0% in 2009, compared to 1995. This change is very little over a time span of 14 years. However, when looking at the difference in total emissions of EU-15 (EU members as of 1995), it can be seen that this change is almost 2% higher, indicating that the newly entered EU member states as of 2004 have not been on the same pace as the core EU members.

5. TRENDS IN INTERNATIONAL FRAGMENTATION AND CORRESPONDING CO2 EMISSIONS

This section investigates the trends in international fragmentation and the corresponding emissions values of the transport equipment14 value chains for 40 countries over the period 1995 and 2009. Table 2 shows the foreign value added shares and foreign generated carbon dioxide emissions shares as well as the change over the period 1995 and 2009 for the transport equipment value chains, based on Equation (4) and (6) respectively. The results are grouped by region and sorted by the change in the foreign value added shares in the first 3 columns, from high to low by country-of-completion. The corresponding foreign carbon dioxide emission shares are presented in the next 3 columns and the percentage point difference between the foreign value added shares and the carbon dioxide emissions shares are presented in the last 3 columns.

The first row, for example, refers to the foreign value added shares, the corresponding foreign emissions shares and the percentage point difference between those shares for which Poland is the country-of-completion. The results indicate that over the period 1995-2009, a larger share of the value in the Polish automotive value chain was added outside the country-of-completion. It increased by 16.2% to be more specific (column 3). More and more auto companies are outsourcing those activities that are labour-intensive to Eastern Europe countries such as Poland (Krassnin and Henning, 2014). Companies such as Volkswagen and Opel are building entire factories in Eastern Europe to systematically utilise low wages to slash wages across the entire continent. The large increase in foreign value added shares might be due to the fact that Poland is importing those activities that require in-depth knowledge, which have a high value added contribution (Lemke, 2015). Also, Poland is responsible for a huge increase in carbon dioxide emissions (29,9%, column 6) due to more fragmentation. Furthermore, the trend towards foreign emissions shares is higher than the trend towards foreign value added shares, indicating that Poland is increasingly offshoring

19

“dirty” work. This could be because Poland is specialized in producing automotive parts and accessories, and imports heavier machinery, which has a higher value added contribution and higher corresponding emissions of carbon dioxide (OECD, 2005).

Another feature that stand outs when looking at the foreign emissions shares is Slovakia. The foreign carbon dioxide emissions of Slovakia have increased considerably (39.3%, column 6), while foreign value added shares increased considerably less (10.3%, column 3). Slovakia constructs the latest automotive plants with the latest technologies, which has by definition less generation of emissions of carbon dioxide compared to older technologies15. Also R&D activities and results in the field of electric vehicles took place at Slovak Universities, which may indicate that Slovakia is one of the front-runners in decreasing CO2 emissions. Therefore, the increase in foreign emissions shares may be because compared to its own production, foreign countries emit considerably more CO2 because they use older technologies to produce automotive parts and components that Slovakia imports.

A number of observations clearly stand out from Table 2. Column 3 reveals that the automotive value chains for more than 30 out of 40 countries became more internationally fragmented between 1995 and 2009. This means that more than 75% of the incorporated countries increasingly offshored production activities in 2009 compared to 1995. There is no clear pattern when looking at the countries, which means that fragmentation takes places all over the world and not just in a specific region. This increase in offshoring may be the result of the fragmentation process, the slicing-up of the value chain, and the creation of global supply chains. This process has increased final imports but more importantly also intermediate imports (Cadarso, López, Gómez and Tobarra, 2010). Column 6 reveals that virtually all countries are responsible for an increase in emissions of carbon dioxide shares. Only the countries India, Taiwan, Malta and Cyprus are the exception. These countries produce relatively more carbon dioxide emissions domestically, probably because these countries do not have the newest technologies available yet and producing parts and components themselves are cheaper then importing them. While both shares were increasing for almost all countries-of-completion, foreign emissions shares increased more for 29 out of 40 countries, as shown in column 9. This indicates that almost all countries contributing in the automotive industry increasingly offshored those production activities that emit relatively a large amount of carbon dioxide. This is no surprise, since people and businesses have been buying an increasing proportion of manufactured goods from overseas, especially in those

15

20

TABLE 2: FVA and FCD shares in output of the automotive industry by country

Foreign value added shares (FVAS)

Foreign carbon dioxide emissions shares (FCDS)

Difference between FVAS and FCDS (%-points) 1995 2009 1995-2009 1995 2009 1995-2009 1995 2009 1995-2009 (1) (2) (3) (4) (5) (6) (7) (8) (9) European Union Poland 24.0 40.2 16.2 10.6 40.5 29.9 -13.5 0.3 13.7 Hungary 40.5 53.8 13.4 51.9 73.8 21.8 11.5 19.9 8.5 Czech Republic 38.7 49.1 10.4 31.5 64.2 32.6 -7.2 15.1 22.3 Slovakia 45.6 56.0 10.3 36.4 75.7 39.3 -9.2 19.7 29.0 Greece 17.7 27.8 10.1 30.7 39.6 8.9 13.0 11.8 -1.2 Luxembourg 44.9 54.9 10.0 78.7 91.4 12.7 33.8 36.5 2.7 Germany 21.0 30.6 9.6 50.7 59.3 8.6 29.7 28.7 -1.0 Ireland 41.6 49.1 7.5 43.2 56.9 13.7 1.6 7.7 6.2 France 27.0 34.4 7.4 61.0 68.7 7.8 34.0 34.4 0.4 Sweden 33.5 40.8 7.4 69.5 76.1 6.6 36.1 35.3 -0.8 Austria 38.6 44.9 6.4 70.8 80.2 9.4 32.3 35.3 3.0 Bulgaria 31.3 37.5 6.3 7.9 19.4 11.4 -23.3 -18.2 5.1 Finland 28.1 34.1 6.0 54.4 60.5 6.1 26.2 26.3 0.1 Great Britain 28.1 33.7 5.6 45.2 56.2 11.0 17.2 22.5 5.3 Romania 23.6 28.4 4.8 8.0 34.3 26.3 -15.7 5.9 21.6 Latvia 30.3 33.7 3.4 46.5 55.7 9.2 16.2 22.0 5.8 Spain 30.8 34.1 3.3 46.1 50.3 4.1 15.3 16.2 0.8 Italy 22.5 25.3 2.8 41.4 46.8 5.4 18.9 21.5 2.6 Belgium 54.9 57.4 2.5 75.4 83.1 7.7 20.6 25.8 5.2 Estonia 29.0 31.2 2.1 21.3 26.7 5.4 -7.7 -4.4 3.3 Malta 34.7 36.1 1.5 42.1 30.6 -11.5 7.5 -5.5 -13.0 Denmark 31.4 30.8 -0.6 70.6 72.4 1.8 39.2 41.6 2.3 Netherlands 45.6 44.7 -0.9 74.1 75.6 1.5 28.5 30.9 2.4 Lithuania 28.6 27.7 -0.9 32.9 56.9 24.1 4.2 29.2 25.0 Portugal 43.6 42.6 -0.9 56.1 67.3 11.2 12.5 24.6 12.2 Slovenia 52.3 50.7 -1.5 69.5 70.8 1.3 17.3 20.1 2.8 Cyprus 32.1 26.8 -5.3 7.5 7.1 -0.4 -24.6 -19.7 4.9

South East Asia

Korea 22.0 31.0 8.9 29.6 34.7 5.1 7.6 3.7 -3.9 Japan 5.6 12.6 7.0 29.1 31.8 2.7 23.5 19.2 -4.3 Indonesia 13.4 19.7 6.3 4.7 8.0 3.3 -8.7 -11.7 -3.0 China 16.2 19.0 2.7 3.7 7.6 3.9 -12.6 -11.4 1.2 Taiwan 31.9 33.4 1.5 52.5 37.8 -14.7 20.6 4.4 -16.2 India 23.7 16.9 -6.7 48.2 23.5 -24.7 24.6 6.6 -17.9 NAFTA United States 16.0 19.5 3.4 22.3 32.6 10.3 6.3 13.1 6.9 Mexico 34.6 32.8 -1.8 41.7 50.7 9.0 7.1 17.9 10.8 Canada 41.8 38.3 -3.5 50.9 55.6 4.6 9.1 17.2 8.1 Other Turkey 18.6 28.9 10.3 34.9 41.3 6.4 16.3 12.4 -3.9 Russia 16.3 23.2 7.0 4.3 8.1 3.8 -11.9 -15.1 -3.2 Brazil 11.4 16.7 5.4 28.2 36.5 8.2 16.9 19.7 2.9 Australia 17.1 20.7 3.6 18.9 37.1 18.2 1.8 16.4 14.6

Note: weighted foreign value added shares and foreign carbon dioxide emissions shares of the transport equipment product group in final output of value chains by country-of-completion. Shares are rounded to one decimal place.

21

countries where regulations on carbon dioxide emissions are weaker than within the EU, such as China (Harvey, 2012). When goods are manufactured in France or any other European country, the companies producing the manufactured goods are subject to strict emissions controls and regulations. To give one example: European countries have to pay for the carbon dioxide emissions they produce, and subsidise renewable forms of generation by paying a surcharge on energy. But other countries outside the EU such as China and India are not subject to such stringent controls and regulation. Also their manufacturing processes and energy generation are often more carbon-intensive than the same processes in Europe. Hence it can be concluded that the automotive value chains got more fragmented internationally, but most of the value chains increasingly offshored “dirty” work over the period 1995 and 2009. These general trends are found for all three regions, but there are also some clear differences, which will be addressed next.

22

those goods that have a higher value added contribution that are associated with more emissions of carbon dioxide (Economy Watch, 2010).

The South East Asian region appears to be the only region to import less carbon-intensive products. 5 out of 6 Asian automotive value chains show both increases in foreign value added and foreign carbon dioxide emission shares, however they show smaller increases in the former. The Chinese automotive value chains shows a higher increase in foreign value added shares, however only to a small extent (1,2%, column 9). These observations are not surprising. As said before, Asia does not have stringent controls and regulations in place yet such as in the EU or NAFTA and their production processes are often more carbon-intensive than the same processes in other EU or NAFTA countries. Therefore, when Asian countries import goods from for example European or American countries they import goods that emit relatively less carbon dioxide as they would produce themselves.

The NAFTA region’s automotive value chains appear to offshore fewer activities since 1995. For both the Mexican and Canadian automotive value chains the foreign value added shares decreased, indicating that they became less reliant on other countries. The U.S. automotive value chains show an increase in foreign value added shares, however only a small increase (3,4%, column 3). When comparing the FVAS by region, it can be concluded that the region NAFTA became the least integrated with the rest of the world in 2009.

While more than 75% of the incorporated countries increasingly offshored production processes, it should be noted that there is still a large home bias in production. Look at, for example Poland, famous of its low-cost parts and components, shows that foreign value added shares indeed increased rapidly (increase of 16.2%, column 3). But in 2009, 59.8% of total value added in the Polish automobile value chains was still added domestically. Even in the case of Germany, well-known for its offshoring of parts and components to Eastern European countries in the automobile value chains, 69.4% of the final output value is still added in Germany itself.

6. TRENDS IN CO2 DISTRIBUTION GENERATED FOR PRODUCTION

23

dioxide generated per dollar of production. To be more specific, I will investigate whether countries-of-completion have been able to reduce carbon dioxide emissions corresponding the whole production process for which these countries are responsible for, as well as domestic and foreign production processes separately.

Table 3 has been constructed to show whether the automotive industry of 40 countries have been successful in reducing the average emissions of carbon dioxide corresponding the production of automobiles over the period 1995 and 2009, based on equation (7) – (9). The first three columns present the average amount of generated emissions per dollar of production including both production which is done in the country-of-completion and which is done abroad. This ratio will be referred to as ‘total emissions ratio’. The next three columns show the average amount of generated carbon dioxide per dollar of production that is produced in the country-of-completion. This ratio of domestic generated carbon dioxide to its corresponding domestic production will be referred to as ‘domestic emissions ratio’. The last three columns display the average amount of generated emissions per dollar of production produced abroad. This ratio of foreign generated carbon dioxide corresponding its foreign production will be referred to as ‘foreign emissions ratio’. The results are sorted by the change in total emissions ratio as defined above (column 3), from highest change to lowest change by country-of-completion, with the corresponding domestic and foreign emissions ratios in the next columns respectively. The first row, for example, refers to the ratios of the automotive value chain for which Romania is the country-of-completion. The results indicate that over the period 1995-2009, the average generated carbon dioxide emissions per dollar of production of Romanian cars have decreased with 88.7% (column 3), indicating that Romania has significantly decreased overall emissions for the production of its cars. The first row also shows that the domestic emissions ratio has a higher change than the foreign emissions ratio, indicating that Romania’s domestic production have been decreasing the average carbon dioxide emissions to a higher extend compared to the emissions generated for Romania’s imports (91.4%, column 6 against 59.5%, column 9). In 1995, Romania was not a member of the European Union yet. European Union countries face highly regulated environmental rules, especially on carbon dioxide emissions since this affect global warming enormously (Hincu, 2014). Since, Romania only joined the EU in 2007, it only faced these stringent environmental regulation from 2007 onwards. Therefore the enormous decline in the domestic emissions ratio might be an effect of its entrance to the EU.

TABLE 3: Total, domestic and foreign generated emissions over value added ratios for the automotive industry

25

EU-15 region 0.2870 0.1505

South East Asia

China 3.5165 1.1499 -67.3 4.0444 1.3112 -67.6 0.7911 0.4603 -41.8 -3.2533 -0.8509 25.8 Indonesia 2.8143 1.4893 -47.1 3.0972 1.7072 -44.9 0.9895 0.6029 -39.1 -2.1077 -1.1043 5.8 Korea 0.6592 0.6038 -8.4 0.5953 0.5715 -4.0 0.8853 0.6759 -23.7 0.2900 0.1044 -19.7 Taiwan 0.5738 0.5648 -1.6 0.3998 0.5275 31.9 0.9453 0.6394 -32.4 0.5455 0.1119 -64.3 Japan 0.2560 0.2995 17.0 0.1924 0.2338 21.5 1.3256 0.7535 -43.2 1.1332 0.5197 -64.7 India 0.2979 0.4281 43.7 0.2021 0.3940 94.9 0.6072 0.5959 -1.9 0.4051 0.2019 -96.8 Asian region 1.4219 0.7909 NAFTA Canada 0.5551 0.3183 -42.7 0.4684 0.2293 -51.0 0.6758 0.4614 -31.7 0.2074 0.2321 19.3 Mexico 0.5939 0.3552 -40.2 0.5298 0.2608 -50.8 0.7150 0.5484 -23.3 0.1852 0.2876 27.5 United States 0.5883 0.3709 -37.0 0.5442 0.3103 -43.0 0.8189 0.6212 -24.1 0.2747 0.3109 18.9 NAFTA region 0.5141 0.2668 Other Russia 3.0824 1.1949 -61.2 3.5213 1.4301 -59.4 0.8218 0.4176 -49.2 -2.6995 -1.0125 10.2 Australia 0.6502 0.3887 -40.2 0.6360 0.3084 -51.5 0.7192 0.6961 -3.2 0.0832 0.3877 48.3 Turkey 0.5156 0.4091 -20.6 0.4122 0.3377 -18.1 0.9687 0.5850 -39.6 0.5565 0.2473 -21.5 Brazil 0.2819 0.2306 -18.2 0.2282 0.1759 -22.9 0.7012 0.5032 -28.2 0.4730 0.3273 -5.3 Note: average total, domestic and foreign emissions ratios of the transport equipment product group in final output of value chains by country-of-completion. The ratios are presented in kilograms of carbon dioxide per dollar of value added and rounded to 4 decimal places. The changes are presented in percentage changes and rounded to 1 decimal place.

means that virtually all countries have been successful in decreasing the average carbon dioxide emissions generated during the production activities for the final demand of automobiles. The total emissions ratio does not specifically tell whether this is due to reductions in the domestic production processes or in the production activities for imported products. Column 6 reveals that for 36 out of 40 countries the domestic emissions ratios have decreased, indicating that virtually all countries have been successful in reducing its average emissions generated during domestic production activities. Also, it can be seen that for all 40 countries the foreign emissions ratios have decreased (column 9), indicating that all countries have been importing less carbon intensive (intermediate) products in 2009 compared to 1995. These trends are clearly related to the fact that nowadays people are becoming more and more aware of the dangers of carbon dioxide emissions for our health and its effects on global warming, such as melting glaciers, rising sea levels and more dangerous weather (Ghose, 2016). Furthermore, it can be seen that the reduction in total emissions ratios are mainly due to major reductions in the domestic emissions ratios. While both domestic and foreign emissions ratios show major reductions for almost all countries-of-completion, the domestic emissions ratios decreased more for 30 out of 40 countries, as shown in column 13. This means that 75% of the countries have been importing those products that are associated with more carbon dioxide emissions than domestically produced products are. This can also be seen in column 10 and 11. These columns also reveal that more than 30 out of 40 countries increasingly offshored “dirty” work for the years 1995 and 2009 respectively. While more than 75% of the countries still outsource “dirty” work to other countries, it can be seen that the average emissions generated for these “dirty” production activities have decreased considerably for almost all countries (column 9). Hence, we can conclude that the automotive value chains have been decreasing its average carbon dioxide emissions domestically as well as abroad, but are still outsourcing those production activities that emits relatively more carbon dioxide than the countries-of-completion produce themselves. These general trends are found for all three regions, but there are also some clear differences, which will be addressed next.

27

than 70% or higher in its domestic emissions ratios, column 6).16 This is clearly related to the fact that new EU member states were quickly integrating with the old EU member states. EU member states face highly regulated environmental controls and regulations, which these new EU member states must comply with. In column 6, it can be seen that many new EU member states show the largest changes in domestic emissions ratios. In particular, Romania, Bulgaria, Slovakia, Lithuania, Poland, Latvia, Czech Republic, Estonia and Hungary show the largest reductions the average emissions generated during domestic production activities in the automobile value chains. At the same time, it can be seen that most of the domestic emissions ratios of newly entered EU member states are still higher compared to the old EU member states, indicating that these new EU member states are on the right track concerning the EU environmental regulations, but still have a long way to go.

Furthermore, it can be seen that the EU members as of 1995 (core EU members) and Japan emit significantly less carbon dioxide per dollar of production in 2009. This can be seen by the significant differences in the domestic emissions ratios of the EU members as of 1995 (except for Greece) and Japan (column 6) compared to the other countries. This can be explained by the fact the environmental regulations for European Union countries and Japan are the most stringent in the world.

The South East Asian region appears to be only region to show increases in the average emissions generated during domestic production activities in the automobile value chains between the period 1995 and 2009. In particular, Taiwan, Japan and India show increases in its domestic emissions ratios (column 6). This may be related to the fact that these countries do not have the newest technologies available yet and most of these countries do not face stringent environmental regulations. Therefore, they only produce in a way that seems efficient for them, without taking the emissions of carbon dioxide into account. Japan is a surprise, since Japan faces the most stringent environmental regulation in the world. It might be because Japan is trying to reduce the CO2 emissions when driving a car so bad that it shifted the emissions to the production of automobiles. Furthermore, China and Indonesia seem to be one of the few countries that import less-carbon intensive products. Both value chains show a higher value for the foreign emissions ratio than for the domestic emissions ratio for both the years. This can be explained by the fact that the production processes of Asian countries are often more carbon-intensive than the same processes in other regions and therefore when these countries import goods from other regions they import goods that emit

16

28

relatively less carbon dioxide as they would produce themselves. These facts also explains why 4 out of 6 Asian countries show bigger changes in their foreign emissions ratio than in their domestic emissions ratio (column 12).

The NAFTA region has been very successful in decreasing its average emissions generated during domestic production activities for the automobile sector between the years 1995 and 2009 (NAFTA countries show a decrease in domestic emissions ratios of around 50%, column 6). This could be surprising, since America is known for its enormous carbon dioxide emissions generated from driving transportation vehicles. In total, the transportation sector of the US produced around 30 percent of all US global warming emissions, which is more than most countries generate.17 Also carbon dioxide emissions have increased by about 17% since 1990, which is mainly caused by increased demand for travel as well as the limited gains in fuel efficiency.18 Many may expect that the production of American cars is associated with a large amount of carbon dioxide emissions, because Americans have a bigger appetite for bigger cars compared to rest of the world and bigger cars emit more emissions. The NAFTA’s average domestic emissions ratios are higher than the core EU member states, however only to a small extent. NAFTA countries have an average value of 0.5141 and 0.2668 in 1995 and 2009 respectively compared to core EU countries values of 0.2870 and 0.1505. This can be supported by the fact that the GHG performance of the U.S. cars lags behind most other nations (An and Sauer, 2014). It might also be related to the fact that Americans have a bigger appetite for bigger cars compared to the rest of the world. According to data from IHS Automotive, of total US sales, larger vehicles accounted for 63% in 2013, while this is only 25.4% in the rest of the world (Salomon, 2015). Salomon (2015) also pointed out that Americans buy bigger cars, because the roads and highways in the U.S. have grown large enough to accommodate them. This is not always the case in other regions. In Europe, for example, there are very old infrastructures and small streets, which make it impossible to get around with large vehicles. Furthermore, the U.S. environmental standards are less stringent than the EU standards. Therefore, the American automotive companies are probably not pushing that hard to decrease the generated emissions just as European countries. American automobile value chains do emit less than the Asian countries, indicating that the expectation that American cars emit more carbon dioxide is false (Asian countries have an average value of 1.4219 and 0.7909 respectively).

17

Source: Union of Concerned Scientists, Science for a healthy planet and safer world 18

29

Regarding the EU environmental package of the EC, only a subset of the EU countries has met the target goal of the EC. The European Parliament approved an energy and climate package, to ensure that EU members will achieve the goal of reducing GHG. One of the rules stipulated that the whole car industry has to comply with an average of 140 grams of CO2 per km in 2008. However, almost all EU countries did not meet this goal. Only Portugal met the goal with 138.2 g CO2/km (see Appendix A). New EU passenger cars were emitting an average of 153 grams of CO2 per km in 2008 (T&E, 2009). To make this data comparable to my data, I assume that 140 grams of CO2 per km is comparable to 140 grams of CO2 per dollar of production. I only look at the domestic emissions ratios, because this reflects best how each country-of-completion are producing for the automotive value chains. Column 5 reveals that in 2009, only 8 EU countries were generating less than 140 grams of CO2 per dollar of production, indicating that only 8 countries have met the target goal. It should also be noted that all 8 countries were core EU members, indicating that the core EU members are on the right track, and that all newly entered EU member states have to work that much harder to reduce carbon dioxide emissions.

It should also be noted that EU countries were emitting more carbon dioxide during the production processes, than what the finished car emits while driving it. 19 out of 27 EU countries were emitting more carbon dioxide per dollar of production than the average CO2 what is emitted while driving the new car (see Appendix A). This might indicate that 70% of the countries have been substituting the emissions of driving a car to the production processes in order to achieve the goal of the EC. Only 3 out of 27 countries have been able to keep the gCO2/$ of production and gCO2/km below 140, of which are all core EU members.

7. TRENDS IN CO2 EMISSIONS BY REGION

This section investigates whether there is a patterns in the emissions generated for production processes which are imported by countries-of-completion. More specifically, it investigates whether distance matters in the distribution of CO2 for foreign production activities and whether there is a bias to specific regions.

7.1 Hypotheses and model

30

activities has great advantages such as greater management control, quick response to changing market conditions by producers, facilitation of just-in-time inventory and production schedules, and communication among managers, suppliers and customers happens with more speed (Clark, 2006). Also, engineering activities like manufacturing automobiles have separable stages of production, such as parts and components, with differing skill, scale and technological needs and can therefore be located at different locations. However, many auto-manufacturing components are heavy, making its production only suitable for relocation in proximate areas due to its low value-to-weight ratio (Lall, Albaladejo and Zhang, 2004). In the section above, I concluded that almost all countries are increasingly offshoring “dirty” work. Therefore, I hypothesize that the closer countries are located to each other, the higher the ratio of emissions per dollar of foreign production. Arguing the other way around results in the following hypothesis:

Hypothesis 1: The ratio of CO2/$ of foreign production decreases when distance increases.

Furthermore, countries located close to each other often have the same levels of per capita income. Therefore, I hypothesize that the lower the differences between GDP, the higher the ratio of emissions per dollar of foreign production. This results in the following hypothesis:

Hypothesis 2: There is a negative relationship between differences in per-capita income and the ratio of emissions per dollar of foreign production.

Over the years, more and more people have become aware of the detrimental (short- and long term) effects of the emissions of carbon dioxide. The most well known effect is global warming. Global warming has major effects on our weather. It causes extreme weather, such as increased rainfall, colder winters, hotter summers, hurricanes to become more frequent and more intense, and more lightning strikes. Therefore, I hypothesize that over the years the ratio of carbon dioxide emissions per dollar of foreign production has decreased.

Hypothesis 3: The ratio of CO2/$ of foreign production decreases over time.

31

these are not as strict as those of the EU. Also, Asia does not have environmental regulations just as Europe or NAFTA. Therefore, I assume the following hypotheses:

Hypothesis 4a: The ratio of CO2/$ of foreign production is lower for EU countries than NAFTA and Asian countries.

Hypothesis 4b: The ratio of CO2/$ of foreign production of NAFTA countries is higher than that of EU countries and lower than that of Asian countries.

Hypothesis 4c: The ratio of CO2/$ of foreign production are higher for Asian countries than for NAFTA and EU countries.

This results in the following model:

𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝐶𝐶𝑖𝑖,𝑐𝑐,𝑡𝑡 =∝ +𝛽𝛽1𝐶𝐶𝑖𝑖𝐷𝐷𝐶𝐶𝐶𝐶𝐶𝐶𝑐𝑐𝑃𝑃𝑖𝑖,𝑐𝑐,𝑡𝑡+ 𝛽𝛽2𝐺𝐺𝐶𝐶𝑃𝑃𝑖𝑖,𝑐𝑐,𝑡𝑡+ 𝛽𝛽3𝐹𝐹𝑖𝑖𝐶𝐶𝐶𝐶𝐹𝐹𝐶𝐶𝑃𝑃𝐹𝐹𝐶𝐶𝐶𝐶𝐹𝐹𝑖𝑖,𝑐𝑐,𝑡𝑡 + 𝛽𝛽4𝐸𝐸𝐸𝐸𝑖𝑖,𝑐𝑐,𝑡𝑡 + 𝛽𝛽5𝑁𝑁𝑉𝑉𝐹𝐹𝑁𝑁𝑉𝑉𝑖𝑖,𝑐𝑐,𝑡𝑡+ 𝛽𝛽6𝐹𝐹𝐹𝐹𝐸𝐸𝑁𝑁𝑆𝑆𝐸𝐸𝑉𝑉𝐹𝐹𝑁𝑁𝑉𝑉𝐹𝐹𝑆𝑆𝑉𝑉𝑖𝑖,𝑐𝑐,𝑡𝑡+ 𝜀𝜀𝑖𝑖,𝑐𝑐,𝑡𝑡

Where Ratio is the dependent variable, α is the intercept parameter, β’s are the coefficients of the different independent variables and ε is the error term. Different countries and different time periods are measured by i,c and t respectively.

7.2 Variables and data sources

Annual data was gathered from the years 1995 to 2011 for 40 countries located all over the world. For my dependent variable Ratio, I used the ratio of foreign emissions to foreign value added for each country that contributes to the final demand of automobiles of countries-of-completion. This is calculated by using the value added and emission values of equation (1) and (5) created for the final demand of transport equipment for each country-of-completion. For each country-of-completion the emissions created abroad are divided by its corresponding foreign value added levels. More specifically, this process is done for each country separately that contributes to the final demand of automobiles. The ratio is displayed in kilograms of carbon dioxide emissions over value added in dollars (kgCO2/$). The data for the explanatory variable Distance is sourced from DistanceFromTo19, which measures the shortest average distance between trading countries in thousands of kilometers. The data for the explanatory variable GDP is sourced from the World Bank (2016). This variable measures the difference

32

in GDP per capita between the trading countries in thousands of dollars. The dummy variable EU shows a value of 1 if the country is part of the European Union. The dummy variable NAFTA has the value 1 when countries-of-completion are trading with Canada, Mexico or the United States and the dummy variable SouthEastAsia has the value 1 for the trading countries China, India, Indonesia, Japan, South Korea and Taiwan. I left observations with the country distinction Rest of The World (RoW) out, because RoW is an estimation of several countries incorporated in the data set.

7.3 Estimation results

Table 5 presents summary statistics for each variable. It shows the mean, median, standard deviation, the minimum and maximum of each variables and the amount of observations. The log function of the variables Ratio and GDP are used because then the variables are much more normally distributed. GDP and Years appear to have only little differences in their means and medians, implying a more symmetrical distribution. However, this is not the case for Ratio, Distance, EU, NAFTA and SouthEastAsia. With regard to the number of observations for each variable, it can be seen that the explanatory variable GDP has less observations, because the log function is not defined when the difference in GDP is zero.

TABLE 5: Summary statistics

Variable Mean Median SD Min Max Observations

Ratio -0.28642 -0.46229 0.94258 -3.22715 3.26203 23400 GDP 2.37883 2.68335 1.20285 -4.58536 4.71697 23367 Distance 5.25292 3.66614 4.22992 0.14101 18.7457 23400 Years 2002 2002 4.32058 1995 2009 23400 EU 0.485 0 0.49978 0 1 23400 NAFTA 0.075 0 0.26339 0 1 23400 SouthEastAsia 0.15 0 0.35707 0 1 23400

33

domestically, which does not include those emissions which are emitted during foreign production activities. Therefore, countries-of-completion may decrease the total domestic emissions by offshoring the “dirty” production activities. As said before, automobile value chains tend to locate close to each other, because many auto-manufacturing components are heavy which makes production only suitable for relocation in proximate areas due to its low value-to-weight ratio. Countries located close to each other often have the same levels of per capita income. The significant and negative coefficient of GDP also tells us that indeed countries-of-completion mainly offshore to proximate sources.

TABLE 6: Regression results

VARIABLES Ratio Expected sign

GDP -0.0675*** (-) (0.00772) Distance -0.0119*** (-) (0.00411) Years -0.0545*** (-) (0.00116)

EU -0.545*** lower than NAFTA and

SouthEastAsia (0.0176)

NAFTA -0.489*** higher than EU, lower

than SouthEastAsia (0.0401)

SouthEastAsia 0.261*** higher than EU and

NAFTA (0.0555) Constant 109.4*** (2.325) Observations 23,367 Number of entities 1,558

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

34

people have become more aware of the effects of carbon dioxide emissions and have been able to reduce the generated emissions for the production of automobiles.

Hypothesis 4a can be accepted, because the variable EU is significant and has a lower value than NAFTA and SouthEastAsia. The coefficient indicates that European Union member countries have significantly lower ratios of kgCO2/$ of foreign production compared to countries which are not EU members. This can be supported by the fact that the EU has the most stringent environmental regulations compared to the rest of the world. Hypothesis 4b can be accepted due to the fact that the significant coefficient of NAFTA is higher than that of the variable EU and lower than the SouthEastAsia variable. This indicates that the ratio of kgCO2/$ of foreign production of NAFTA countries is higher than that of EU countries and significantly lower than those of Asian countries. This can be supported by the fact that the NAFTA does have environmental regulations, but these are not as strict as those of the EU. Hypothesis 4c can also be accepted, because the variable SouthEastAsia is significant and the coefficient is higher than that of EU and NAFTA, indicating that the ratio of kgCO2/$ of foreign production of Asian countries are significantly higher than countries which are not located in South East Asia. Most Asian countries do not face very stringent environmental regulations. Therefore, they do not worry about any consequences of CO2 emissions. Look at for example the extreme smog levels of China. This is not only caused by the Asian countries itself, but also due to countries which offshore to these Asian countries. Since Asian countries do not face stringent environmental regulations, it incentivizes Eastern countries to offshore production processes that emit relatively a lot of carbon dioxide emissions to these Asian countries. This increases the carbon dioxide levels of Asian countries even more than they would if Eastern countries would not offshore production processes to Asia. Hence, it can be concluded that there is indeed a CO2 bias in specific regions.

8. CONCLUSION