Trade Patterns and Environmental Pollution:

(1)

University of Groningen

Faculty of Economics and Business

MSc International Economics and Business

Double Degree with Corvinus University of Budapest

Master Thesis

Trade Patterns and Environmental Pollution:

An analysis of the popular pollution hypotheses

Supervisor: Prof. dr. Robert Inklaar

Co-assessor: dr. Péter Isztin

Barna Hunor Horváth

b.h.horvath@student.rug.nl

(2)

Abstract

The relationship of environmental pollution and international trade is a popular topic both with academics and policy makers. Theoretical models and empirical analyses sometimes lead to controversial results. The following thesis investigates this relationship considering the three popular hypotheses: the pollution haven hypothesis, the pollution halo hypothesis, and the factor endowments hypothesis. The effects are studied through changes in export patterns, taking environmental regulations, pollution abatement costs, capital endowments, and capital intensities into account. Based on theoretical considerations, I develop an empirical estimating equation that allows all three hypotheses to affect trade patterns simultaneously. Three datasets are compiled due to multiple measures of environmental regulations, consisting of 40 countries and 19 sectors for the years 1995 and 2011. The results support a significant pollution halo and factor endowment effect, which is prevalent throughout the years of interest and across different datasets. Furthermore, the results stress the importance of well-designed environmental policies and factor mobility.

(3)

List of abbreviations

BKC Botta and Kozluk Countries CI Capital Intensity

EFC Eliste and Fredriksson countries EU European Union

FDI Foreign Direct Investment FEH Factor Endowments Hypothesis GDP Gross Domestic Product

NIOT National Input-Output Table PAC Pollution Abatement Cost

PAT Pollution Abatement Technologies PHH Pollution Haven Hypothesis

PWT Penn World Table

SIC Standard Industrial Classification

UNCED United Nations Conference on Environment and Development US United States

USD United States Dollar

(5)

1. Introduction

In recent years, the relationship of environmental pollution and trade liberalization has received a great deal of attention from academics and policy makers, which resulted in the emergence of three theories hypothesizing the effects of regulations, production technologies and factor endowments affecting environmental pollution. These effects are known as the

pollution haven hypothesis (PHH), the pollution halo hypothesis, and the factor endowments hypothesis (FEH).

(6)

This study develops an empirical interaction model which allows to examine changes in trade patterns due to the above-mentioned hypotheses’ effects. More specifically, I test how these hypotheses’ forces change the export specialization of countries. The estimating equation works with two interaction terms combining country and sector measures representing the PHH, the pollution halo hypothesis, and the FEH. The FEH is represented by the interaction of countries’ capital endowment and the capital intensity of their sectors: theory suggests that well-endowed countries specialize in capital-intensive sectors. These sectors are typically pollution intensive sectors, as elaborated in the Literature review. The effect of the PHH and the pollution halo hypothesis is identified by the other interaction term, which combines the stringency of environmental regulations in a country as a national measure and its sectoral pollution abatement costs (PACs). Theory implies that in case of the PHH’s dominance, countries with lax regulations export more in sectors where PACs are high, the dirty industries, while in case the pollution halo hypothesis is dominant, stringent regulations entail higher exports in low-PAC sectors, the clean industries. The environmental regulatory measure is relatively problematic to exactly define compared to the other measures considered in the analysis, thus, it is difficult to find data that represents this stringency of regulations well. In this study, three measures are considered to eliminate the potential biases by using only one dataset. The first measure is based on Botta and Kozluk’s (2014) index, who consider policy instruments related to pollution and climate, the second is Eliste and Fredriksson’s (2002) measure, who compile country reports on regulations to build an index, and the third measure is a de-facto stringency index based on pollution intensities. The outlined approach allows all three hypotheses to influence the outcome of the model. Based on the described measures, three datasets are constructed including 40 countries and 19 sectors for the years 1995 and 2011 to examine the effects.

(7)

This thesis strengthens the results of analyses emphasizing the importance of well-designed environmental regulations and advanced production techniques to reduce pollution; furthermore, it also confirms that the FEH cannot be ignored, as its effect is found to be present in this study and all related studies that consider factor endowments. Additionally, the contribution of my thesis to the existing literature is that the approach allows to consider the effects of all three hypotheses simultaneously, which is neglected in the existing literature; moreover, the multi-country setup with a wide set of countries provides robust results, and lastly, earlier evidence is updated and strengthened emphasizing that the effects are still dominant in influencing specialization in 2011.

The rest of the thesis is organized as follows: Section 2 provides a literature review about the research topic, Section 3 introduces the methodology and the hypotheses regarding the empirical investigation, Section 4 provides the data description, Section 5 includes the analysis, Section 6 discusses the results of the empirics, and Section 7 concludes.

2. Literature review

This thesis evaluates the relationship between environmental pollution and international trade through changes in trade patterns. This section provides the literature review about the topic: the first subsection introduces the theoretical considerations regarding the relationship of trade and pollution and thereafter the empirical evidence of relevant studies is reviewed.

2.1. International trade and environmental pollution

(8)

(1977) to analyze the relationship of environment and trade. In their model, the authors assume that the North has a higher level of technology or human capital, which leads to the South having a lower level of income. The income difference under endogenous environmental policy implies that the North implements sterner environmental regulations, which generates an income-induced comparative advantage for the South in producing the dirty goods and for the North in producing the clean goods. This results in the South becoming the dirty goods’ net exporter due to its lower level of income and lax environmental policy. Thus, the model is a detailed example of the PHH explaining the underlying mechanisms: low-income countries specialize in producing dirty goods. The PHH is a popular hypothesis amongst researchers, even though it is heavily debated, as empirical validation depends on the different methodological approaches, the analyzed locations, and the pollutants considered.1_{The empirical research on the topic indicates that the PHH’s effect, among}

others, is dependent on the technological and human capital differences, the stringency of environmental regulation, and the countries’ abundance of national resources. Antweiler et al. (2001) and Cole and Elliott (2005) argue that the effect of the PHH tends to be difficult to identify, since it is simultaneously present with the factor endowment hypothesis’ opposing force. This leads us to the next hypothesis mentioned previously.

Compared to Copeland and Taylor’s (1994) model introduced above, the focus of a later work from the authors shifts towards the FEH (Copeland & Taylor, 2004): they theorize that high-income and capital-abundant nations will be the net exporter of dirty goods. The FEH originates from the well-known Heckscher-Ohlin-Samuelson theorem, which states that the comparative advantage of a country is determined by its relative capital and labor endowments. In our case, we are interested in the impact of endowments on pollution, since the hypothesis states that capital-abundant countries specialize in pollution-intensive production. Cole and Elliott (2005) find that there is a strong correlation between the pollution intensity and the capital intensity of a sector: capital-intensive industries are typically the dirty industries as well. This means that firms producing dirty goods will relocate to capital-abundant countries, since the capital’s relative price is lower there, which is a cost advantage for the firm. Therefore, the FEH predicts that low-income and labor-abundant countries will

(9)

specialize in the production of the clean good, while high-income and capital abundant countries will pollute more due to the production of the dirty good. As mentioned above, the PHH and the FEH have opposing effects; however, the attempted justification of the FEH is less dominant in the empirical literature.

Finally, I elaborate on the third hypothesis: the pollution halo hypothesis. This proposition is based on the Porter hypothesis, which indicates that a well-designed strict environmental regulation can direct investment in cleaner technologies and innovation, thus enhancing competitiveness (Ambec, 2013). Therefore, multinational firms looking to relocate or to partner with another firm to buy their products from abroad oftentimes already comply with stringent environmental regulations, which leads to them exporting cleaner production processes via their knowledge and better practices. Eskeland and Harrison (2003) argue that this may lead to the rejection of the PHH and the rise of the pollution halo hypothesis. Thus, increasing international trade may result in the import of clean technologies and good environmental practices, which in turn, may increase growth rates and income. Based on the above-mentioned reasoning, higher income will result in sterner environmental regulation. Consequently, more awareness and demand towards environmentally-friendly production will be present. The pollution halo hypothesis’ effect, therefore, represents the changing techniques of processes towards cleaner production. The opposing effects of these hypotheses suggest that they need to be assessed together investigating the conditions which make one hypothesis more dominant than the other when analyzing international trade’s effect on pollution.

(10)

different technologies and practices, the more or less environmentally-friendly production’s impact is captured by this effect. Grossman and Krueger (1995), Antweiler et al. (2001), and Copeland and Taylor (1994, 2004) suggest that the relationship of pollution and international trade can also be explained through the three aforementioned effects. Therefore, now that the three hypotheses to consider have been established previously, I will use Grossmann’s (1995) framework to study the effect of these hypotheses systematically.

2.2. Empirical evidence

In this subsection, empirical evidence is reviewed regarding the above-theorized pollution hypotheses. Most of the empirical work is based on the PHH in this area of environmental research. However, its empirical support appears to be weak, possibly because the predictions of the other theories discussed run counter to the PHH prediction. Thus, the focus of this discussion is not the validation of each hypotheses, but the analyses which try to incorporate the opposing forces of the contrasting concepts and link international trade with industrial pollution from a broader perspective.

(11)

significant effect at all regarding the pollution haven component of world trade (Manderson & Kneller, 2012; Rezza, 2013; Ren et al., 2014; and Garsous & Kozluk, 2017).

The ambiguous empirical results from the PHH stem from the assumption that the environmental institutional and regulatory differences are the sole motivation for international trade (Copeland & Taylor, 2004). Furthermore, studies that focus on the trends of the pollution-intensive industries’ output instead of pollution levels or PACs necessarily assume that changes in the composition of a country’s output correspond to changes in environmental quality. However, the techniques of production advance over time; therefore, a larger share of dirty-industry output may be consistent with either higher or lower levels of pollution (Copeland & Taylor, 2004). In the early 2000s, studies started to emerge explaining why the PHH often fails to hold empirically. Eskeland and Harrison (2003) analyze firms’ location of production and technology levels (energy efficiency) using micro data. The authors argue that a fall in PACs, a measure for the technique effect, with the scale of output may lead to a firm producing at home even when facing stricter environmental regulations. The study finds that evidence for the PHH is weak at best; however, foreign firms producing in a country are more energy efficient. A study of Grether et al. (2012), which analyzes the pollution content of imports, states that in reality, developing countries are also importers of dirty products. Therefore, it is necessary to take the technique effect into account, which is embodied in the different relative abatement costs in various countries. Levinson and Taylor (2008) corroborate this view in their paper which examines the effect of environmental regulations on trade flows. The authors find that industries with increased PACs experienced the largest net imports. These empirical findings are consistent with the pollution halo hypothesis and suggest that the technique effect simultaneously influences a country’s industrial specialization and pattern of trade with the country’s stringency of environmental regulations.

(12)

find evidence that both the PHH and the FEH are driving comparative advantage in terms of site-specific pollution concentration. Moreover, their results show that international trade alters the composition of national output and trade-induced technique and scale effects imply a net reduction in pollution. In a closely related study, Cole and Elliott (2003) argue that pollution-intensive sectors are subject to the opposing forces of PHH and FEH, since these sectors are capital intensive, but nations with lax environmental regulations are often the least capital abundant. The authors’ empirical investigation focuses on compositional changes in pollution as a result of increased international trade due to the differing stringency of environmental regulations and the differences in capital-labor endowments. The results implicate that both regulation and endowments affect environmental pollution and the effects tend to cancel one another out. In contrast to Antweiler et al. (2001), Cole and Elliott (2003) cannot fully confirm that the overall trade-induced composition effect is negative based on their analysis, meaning that there is no conclusive evidence of freer trade being good for the environment. Azhar and Elliott (2007) examine trade specialization and aim to distinguish between the PHH and the FEH. The authors apply the aforementioned North-South trade patterns and break down their analysis to subperiods. The results, similarly to the studies above, find that the two hypotheses with opposing forces simultaneously influence trade patterns; however, the empirical evidence is temporary: there are subperiods where the PHH effect is significant and dominant, while in other time periods the FEH effect acts the same way. Lastly, in a recent paper, Grether et al. (2012) study the pollution content of international trade flows and find significant impact of both the PHH and the FEH in a North-South setting.

(13)

The reviewed studies are in line with the theoretical implications discussed in the previous section: the results tend to be ambiguous when studies focus on one of the driving forces and studies incorporating more aspects confirm that all the hypotheses should be considered for an accurate representation of how the environment is affected by international economic activity. However, such inclusive studies have not been performed to establish the relevance of these forces relative to each other. A possible reason may be the absence of accurate and detailed data on production, consumption, and pollution. Furthermore, to accurately study the composition effect in the relationship between international trade and pollution one requires sector-level datasets. The reason why the studies in this area neglect sectoral data is that it requires detailed and consistent information at the industry level for multiple countries, such as trade, output, factor intensities, and emission levels. In conclusion, the effect of the introduced hypotheses is inconclusive due to partial studies focusing on certain aspects and the lack of detailed data. This thesis aims to fill part of this gap by conducting an empirical analysis that allows us to investigate the forces of the introduced hypotheses simultaneously focusing on international trade and the environment and by combining data from multiple sources to build broader datasets.

3. Methodology

(14)

3.1. Estimating equation

As established in the literature review, this thesis argues that the relationship of international trade and environmental pollution may be subject to opposing forces of regulatory, factor endowment, and technological sources of comparative advantage, which causes nations to specialize differently. The empirical evidence regarding the PHH, the FEH, and the pollution halo hypothesis suggests that the export composition of a country is a function of environmental regulations, pollution abatement technologies, and capital-labor endowments:

𝑋𝑖𝑗 = 𝑓( 𝐸𝑅𝑖 , 𝑃𝐴𝑇𝑖𝑗 , 𝐾𝐿𝑖𝑗 , 𝛺𝑖 ) (1)

where subscripts i and j represent the country and the sector respectively, 𝑋 is exports, 𝐸𝑅 is environmental regulations, 𝑃𝐴𝑇 is pollution abatement technologies, 𝐾𝐿 is capital endowment, and 𝛺 captures other determinants at the country level.

(15)

production, meaning it is relatively cheap, will specialize in industries intensively using that abundant factor.

Similarly to factor intensities, the same approach can be used for stringency of environmental regulation and PACs. In this case, the industries are ranked by their PACs, which reflects their technological advancement regarding pollution intensity during the production process, represented by PAT in equation (1). On the one hand, the more stringent a country’s environmental regulation is, the higher the costs of production, since the price of environmental inputs increase, which causes nations with lax environmental policies to have a comparative advantage in dirty industries’ production. However, if we only looked at environmental regulation, an immediate limitation would arise: it would be assumed that changes in the export composition of a country correspond to changes in environmental regulation. But if technologies advance and techniques of production change as a result of international trade over time, the effect of regulation is not straightforward. On the other hand, thus, well-designed environmental policies may induce technological progress and better pollution abatement techniques, as elaborated in the literature review previously, which reduces the PAC of a sector.

To supplement the conclusion of the previous paragraph, trade shares are determined by relative product prices, similarly as before, which are a function of factor intensities and relative factor prices. However, regarding the PHH and the pollution halo hypothesis, factor prices depend on environmental policies and factor intensities depend on pollution abatement technologies (reflected in PACs). The theoretical considerations make these effects inseparable. Furthermore, Grether (2012) notes that trade shares depend on the interaction of factor costs and intensities, which also suggests that the PHH and the pollution halo hypothesis should be jointly evaluated. Thus, the effect of the comparative advantage of regulations and technologies is whether a country which has a more stringent regulatory system will specialize in the production of dirty or clean industries.

(16)

theoretical and methodological considerations above, the estimating equation to measure the effect of the PHH, the FEH and the pollution halo hypothesis is the following:

Xij = δi + γj + ß1 ( ERi * PACj ) + ß2 ( KLi * CIj ) + εij (2)

where subscripts i and j represent the country and the sector respectively, Xij is the exports of

a sector in a country, ERi is the stringency of environmental regulations in a country, PACj is

the pollution abatement cost of an industry, KLi is the capital-labor ratio representing the

factor endowment of a country, CIj is the capital intensity representing the industry measure

for factor intensity. Terms δi and γj are country- and industry-fixed effects respectively, and εij

is the error term.

The interaction terms in the estimating equation allow for the sectoral component to be influenced by the country component. The interaction between a country’s strength of environmental regulations and the sectoral abatement costs allows for the environmental stringency to affect PACs, while the interaction between a country’s factor endowment and the sectoral factor intensities allows for the resource abundance to affect the sectoral capital-labor ratios. Thus, the first term measures the opposing effects of the PHH and the pollution halo hypothesis and the second term measures the impact of the FEH. As this thesis is focused on the effects of the elaborated pollution hypotheses, country and time fixed effects are included in the estimating equation to avoid erroneously attributing variations in specialization to the three effects of interest: δi captures all other time-invariant

country-specific characteristics, while γt captures all other time-invariant sector-specific

characteristics.

(17)

instruments in other countries, because PACs and CIs are unaffected by changes caused by national exports. The relevance of using US sectoral measures for all countries is also discussed in Section 4. Now that the empirical approach is introduced, the next section guides through the hypotheses of this study, which reflect the expected results implied by the theoretical considerations and set a clear view about the quantification of their impact.

3.2. Hypotheses

Equation (2) serves as guidance to the hypotheses of this thesis. The first term, the interaction of environmental regulations and PACs, captures the PHH’s and the pollution halo hypothesis’ effect and the second term, the interaction of factor endowments and sectoral factor intensities, captures the FEH’s effect.

The pollution haven effect suggests that countries with laxer environmental regulation export more in pollution-intensive industries. This force is represented by the first interaction term in equation (2), which includes the environmental regulation and PACs. PAC represents the pollution intensity of a sector, since the dirtier the sector, the more abatement cost arises in connection with cleaning up the impacted environment (note that this can be influenced by the reducing technique effect of more advanced pollution abatement technologies, as described by the next hypothesis). Thus, should higher-PAC sectors export more where environmental regulation is lax, the PHH is relevant and the ß1 coefficient is positive. Based

on these considerations, Hypothesis 1a is stated as follows:

H1a: Less stringent environmental regulation leads to higher volumes of exports in sectors where the PACs are higher.

The pollution halo effect suggests that more stringent environmental regulation induces more technologically advanced production processes, which are less harmful for the environment. This force is represented by the first interaction term in equation (2) as well, which includes both environmental regulation and PACs (which also indicates the advancement of pollution abatement technologies, since the better the technology, the lower the cost): should lower-PAC sectors export more where environmental regulation is strong, the pollution halo hypothesis is relevant and the ß1 coefficient is negative. Thus, Hypothesis 1b is stated as

(18)

H1b: More stringent environmental regulation leads to higher volumes of exports in sectors where the PACs are lower.

The factor endowments effect suggests that capital abundant countries export more in capital-intensive industries. As established previously, capital-intensive sectors are typically the most polluting ones: there is a strong correlation between capital and pollution intensity. This force is represented by the second interaction term in equation (2), which includes the capital endowment of a country and the capital intensity of the sectors: should higher capital-intensity sectors export more in capital abundant countries, the FEH is relevant and the ß2

coefficient is positive. Thus, Hypothesis 2 is stated as follows:

H2: More capital abundant countries exhibit higher volumes of exports in sectors where the capital intensities are higher.

Table 1: Hypothesized effects of the introduced pollution theories

Hypothesis / Coefficient ß1 ß2

H1a (Pollution haven effect) Positive (+) H1b (Pollution halo effect) Negative (-)

H2 (Factor endowments effect) Positive (+)

Source: own illustration

In summary, the coefficients of equation (2) are the measure for the effect of the opposing pollution theories and the signs indicate the respective significance of their impact. Table 1 illustrates the above-described effect for a clear view of the hypotheses described above.

4. Data

This section introduces the concepts and measurements regarding the underlying data for the empirical analysis. Three datasets are constructed as the result of the different dimensions of available data at the country level. The first group covers the 40 countries included in the World Input-Output Database2_{(WIOD), which is a database containing essential information}

for the analysis of this study (Timmer et al., 2015). The second group contains 30 countries

(19)

from Botta and Kozluk’s (2014) dataset which are common with the WIOD, and similarly, the last group includes 29 countries from Eliste and Fredriksson’s (2002) dataset which are common with the WIOD. The purpose of tailoring the country list according to the studies mentioned before is that these studies contain essential data for the analysis, as it is presented in the following paragraphs. Thus, using data from the aforementioned papers allows me to use only the common countries between these studies and the WIOD. The list of countries in the different datasets can be found in Appendix 1 with the differences highlighted. The required data for the analysis in the case of less-developed nations is scarce, especially regarding the availability of environmental accounts, which limits the scope of countries that can be observed: the WIOD includes 27 nations of the European Union (EU) and 13 other major economies, several developing countries amongst them; however, the number of emerging countries omitted is still high. Nevertheless, I am forced to use the mentioned set of countries, as the WIOD provides sectoral data on economic activity with the respective environmental accounts, which is required for the analysis of this study. Furthermore, the sectoral data is organized into 19 two-digit Standard Industrial Classification (SIC) industries, which is also determined similarly as the set of countries: this is the highest number of sectors with the required data available. Moreover, this thesis’ years of interest are 1995 and 2011, selected according to the earliest and latest complete dataset with the necessary information. The results of the two different years will allow us to gain insights about the changes during the time elapsed.

(20)

technologies. Table 2 gives an overview of the of the PACs with the respective SIC industries ranked from highest to lowest. The table also presents the values for sectoral capital intensity (CI), which is defined as non-wage value added per worker expressed in 1990 USD. Similarly, CIs are compiled by Cole and Elliott (2005) based on data from the US Census Bureau.

Table 2: Industries with their respective PACs and CIs

Pollution Abatement Costs (% of value added) Capital Intensity (per worker) Petroleum and coal products 9.9% Chemicals and allied products $129 400 Primary metal industries 3.5% Petroleum and coal products $125 900 Paper and allied products 2.7% Primary metal industries $82 100 Chemicals and allied products 2.4% Paper and allied products $80 800

Tobacco products 2.3% Tobacco products $64 400

Leather and leather products 1.5% Instruments and related $57 400 Stone clay and glass 1.4% Printing and publishing $48 800 Fabricated metal products 0.9% Stone clay and glass $45 600 Lumber and wood products 0.8% Food and kindred products $45 100 Textile mill products 0.8% Electronic equipment $44 600 Food and kindred products 0.7% Industrial machinery $41 800 Transportation equipment 0.7% Transportation equipment $39 200 Rubber and misc. plastics 0.7% Fabricated metal products $33 100 Furniture and fixtures 0.5% Rubber and misc. plastics $32 100 Electronic equipment 0.5% Misc. manuf. industries $29 700 Misc. manuf. industries 0.4% Leather and leather products $27 700 Industrial machinery 0.3% Textile mill products $24 400 Instruments and related 0.3% Furniture and fixtures $24 200 Printing and publishing 0.2% Lumber and wood products $23 100

Source: Cole and Elliott (2005)

(21)

(22)

Figure 1: The relationship between sectoral PACs and CIs

Source: Cole and Elliott (2005)

The available data for stringency of environmental regulation provides three alternatives. The first measure originates from Botta and Kozluk (2014), who develop an environmental policy stringency index based on 14 policy instruments primarily related to climate and air pollution, which provides data for the years of interest of this study. The index is a country-specific measure defined as the degree to which an explicit or implicit price on pollution or environmentally damaging behavior is set. The index takes values from zero to six and the countries included from the authors’ database in this study are shown in Appendix 1.

(23)

The third option is a relative de-facto measure of stringency in pollution control based on pollution intensities. As for the sectoral data the US is determined the reference country above, it is wise to make the same country the benchmark in this case as well. The index is the difference of sectoral pollution intensities from each country in the sample and the US weighted by the sectoral output shares in the US, summed on the country level. Since relatively more pollution than the US is assumed to imply laxer regulations, I take the negative value of this index. The calculation of the index described above is shown by equation (3):

ERi = -Σi wUSj [Pij / Yij – PUSj / YUSj] (3)

where subscripts i and j represent the country and the sector respectively, wUSj is the weight

of industries in the US economy, Pij is the carbon emission of a sector in a certain country, Yij,

is the output of a sector in a certain country. The second term in the brackets represents the US pollution intensities. The sectoral carbon emission data is measured in kilotons and extracted from the environmental accounts of the WIOD along with the output data from the national input-output tables (NIOTs), measured in millions of USD. Developing this relative environmental regulation index based on pollution intensities allows me to include all the countries present in the WIOD. However, it is not a direct measure of the actual regulatory stringency, as in the previous cases; therefore, the results are interpreted carefully keeping this point in mind.

The capital endowment data is derived from the Penn World Table3_{(PWT): the database}

contains historical information on capital stock and number of persons engaged (Feenstra et al., 2015). The capital endowment in this study is measured by the capital stock relative to the active workforce in a country. Thus, the variable is calculated by taking the capital stock of a nation and divide it by the number of laborers. The PWT provides data on a wide range of countries for the past decades; the capital-endowment variable can be constructed for the nations and years of interest. The number of persons engaged are in millions, and the capital stock at current PPPs is expressed in 2011 USD. The relationship between the capital endowment and the stringency of environmental regulations is also worth to observe, since as mentioned in the literature review, richer countries tend to have better environmental regulations. Figure 2 presents the capital endowment and the regulatory index from Botta and

(24)

Kozluk (2014) for each country, which is the most reliable source on regulation out of the three described options. It is visible that the least well-endowed countries typically have the lowest scores for their regulatory stringency, as Brazil, Russia, India, Indonesia, and China lag behind. Although the regulation is significantly better in European countries, there is variation amongst these countries as well: typically, the eastern and southern countries received a lower, while the northern nations received a higher regulatory index. However, it would be hazardous to state that well-endowed countries have better environmental stringency based on solely Figure 2.

Figure 2: Countries with their respective capital endowment and stringency of environmental regulation index

Table 3 presents how the different environmental regulatory measures compare to each other and supplements Figure 2 by providing a formal test on whether the data corresponds to a typical North-South country setting. The ER measures have a statistically significant strong relationship: the first two measures from the mentioned studies have a correlation coefficient of 0.718, while the constructed WIOD measure has slightly lower coefficient with the other two regulatory indices. Furthermore, the North-South distinction holds well: there is a strong relationship between the natural logarithm of capital endowment and the various ER

0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 0 50 000 100 000 150 000 200 000 250 000 300 000 350 000 400 000 450 000 500 000 550 000 Indi a C h in a Indon es ia P ol an d B ra zi l R us si a Tur key Sl ov ak ia K or ea H ung ar y Au st ra lia U ni ted K ing dom Sl ov eni a C ana d a Ja pa n Ge rm any C ze ch R ep ubl ic U ni ted S ta tes Sw ed en Gr eece N et he rl ands Fi nl and P or tug al Au st ri a Fr ance _Spa in D en m ar k Ir el an d It al y B el gi um R eg ul at o ry Str ing enc y Index US D / wo rker

(25)

measures and a moderate-strong relationship between the natural logarithm of gross domestic product (GDP) per capita and the ER measures.

Table 3: Correlation between ER measures, capital endowment and GPD per capita ER - Botta & Kozluk ER - Eliste & Fredriksson ER - WIOD measure ln Capital endowment ln GDP per capita ER - Botta & Kozluk 1 - - - - ER - Eliste & Fredriksson 0.718*** (0.006) 1 - - - ER - WIOD measure 0.627** (0.026) 0.643*** (0.004) 1 - - ln Capital endowment 0.766*** (0.001) 0.712*** (0.001) 0.677*** (0.002) 1 - ln GDP per capita 0.575** (0.017) 0.562* (0.063) 0.526** (0.027) 0.656* (0.068) 1 Where *, **, and *** indicate significance at 10%, 5%, and 1% respectively.

P-values are in parentheses.

Source: own calculation

The gross export values, the dependent variable in the estimating equation, are obtained from the WIOD: the database includes NIOTs which provide, among others, the sectoral export volumes in current millions of USD for every included country for the years of interest. Summary statistics of the discussed datasets and variables can be found in Appendix 2. This concludes the concepts and measurements regarding the data used in this thesis.

(26)

exception from this is the index from Botta and Kozluk (2014), which was initially constructed for the period 1990-2012 but has been updated since, although not for all countries. However, the dataset only covers six developing nations. Furthermore, data on PACs is similarly rare, especially with details on different industries, which is the reason why the sectoral data is based on US measures based on a dataset constructed for another study including 19 sectors. Nevertheless, the construction of the relative PAC measure from equation (3) allows to extend the country sample for all 40 WIOD countries.

Regarding data transformation in the estimating equation (2), the natural logarithm values are considered for exports, capital endowment, and capital intensities based on the standard in the empirical literature and the variable values' magnitude to conform to the estimating technique. The rest of the valuables, the PAC as percentage and the ER as an index, remain unchanged. Furthermore, as implicated by the Breusch-Pagan test, the variance of the measures on the country level is not homoscedastic, thus, heteroscedasticity robust standard errors are used clustered by countries.

5. Results

(27)

Table 4: Results for 1995 1995

BKC measure EFC measure WIOD measure

Variables lnGross exportsij lnGross exportsij lnGross exportsij

(1) (2) (3) ERi * PACj -0.118*** (0.005) -0.027** (0.017) -0.051** (0.030) lnKLi * lnCIj 0.343*** (0.001) 0.128** (0.029) 0.255** (0.023) Nr. of observations 570 551 760 Adjusted R-squared 0.748 0.721 0.812 Nr. of countries 30 29 40 Country fixed

effects Yes Yes Yes

Sector fixed effects Yes Yes Yes

Where *, **, and *** indicate significance at 10%, 5%, and 1% respectively. P-values are in parentheses.

Table 4 presents the results for the year 1995. The coefficients of the first interaction term (ERi*PACj) measuring the pollution halo and the pollution haven effect show negative values

in all three cases, indicating the dominance of the pollution halo hypothesis, meaning more stringent environmental regulation leads to higher volumes of exports in sectors where the PACs are lower. For the BKC measure in column (1), the coefficient is -0.118 and it is highly significant at the one-percent level. Column (2) shows the results for the EFC measure, where the pollution haven effect is still dominant, albeit with smaller impact, as the coefficient is closre to zero, namely -0.027 with a five-percent significance level. The WIOD measure in column (3) produces a coefficient of -0.051, somewhat milder than the BKC measure and also significant at the five-percent level. The coefficients of the second interaction term (lnKLi*lnCIj)

measuring the factor-endowment effect have positive signs, meaning that more capital abundant countries exhibit higher volumes of exports in more capital-intensive industries. In column (1), the coefficient is 0.343 with a significance level of one percent. Coefficients in column (2) and (3) are lower but still positive, 0.128 and 0.255 respectively with a significance level of five percent.

(28)

period. The first interaction term for the BKC measure remains negative and significant at the five-percent level with a value of -0.039. Interestingly, the same ERi * PACj term for the WIOD

measure in column (2) turns positive with a value of 0.259, significant at the ten-percent level, indicating the dominance of the PHH in this case. The coefficients of the second term measuring the FEH are positive in both columns (1) and (2), as in 1995, with a higher value of 0.479 for the BKC measure and a slightly lower value of 0.237 for the WIOD measure, significant on the one- and five-percent levels respectively.

Table 5: Results for 2011 2011

BKC measure WIOD measure

Variables lnGross exportsij lnGross exportsij

(1) (2) ERi * PACj -0.039** (0.016) 0.259* (0.075) lnKLi * lnCIj 0.479*** (0.006) 0.237** (0.030) Nr. of observations 570 760 Adjusted R-squared 0.750 0.804 Nr. of countries 30 40

Country fixed effects Yes Yes

Sector fixed effects Yes Yes

Where *, **, and *** indicate significance at 10%, 5%, and 1% respectively. P-values are in parentheses.

In order to test whether the changes in the effects of the pollution halo hypothesis and the FEH between 1995 and 2011 are statistically significant, equation (4) is estimated:

Y = ß1X + ß2YR + ß3 ( X * YR ) + ε (4)

where X is the variable of interest (ER*PAC in case of estimating the changes of the pollution halo effect and lnKL*lnCI for the factor endowment effect), YR is an indicator variable that takes the value zero for year 1995 and one for 2011.

(29)

whether the effect of the pollution halo hypothesis or the FEH is significantly smaller or larger in 2011 than in 1995. Equation (4) is estimated for the BKC and WIOD measures, since the EFC measure is only available for the first period. The ß3 coefficient from estimating equation (4)

for the BKC measure is positive and significant for the pollution halo effect at the five-percent level and also positive and significant for the factor endowment effect at the one-percent level. The same estimation for the WIOD measure does not result in significant ß3 coefficients,

meaning that the change is not significantly different from zero.

The changes in the coefficients between the years 1995 and 2011 are summarized in Table 6 for the BKC dataset, as the differences are only statistically significant in that case. On the one hand, the positive difference for ER*PAC indicates that the pollution halo effect became less dominant relative to the PHH, albeit still dominant: the coefficient remained negative for 2011, shown in the previous Table 5. On the other hand, the difference for lnKL*lnCI indicates that the factor endowment effect became more dominant over the examined years: the coefficient increased and the positive change is statistically significant, as mentioned before. In summary, the FEH remained dominant and further increased its effect, while the pollution halo hypothesis also remained dominant, albeit losing some of its effect to the PHH.

Table 6: Changes in coefficients between 1995 and 2011 Difference between 1995 and 2011

BKC measure

Variables lnGross exportsij Δ (ERi * PACj) 0.079

Δ (lnKLi * lnCIj) 0.136

In order to put the magnitude of the interaction terms’ effects into perspective, a calculation is performed as follows. The sector at the 75th_{percentile of CI is Tobacco products (high capital}

intensity) and the sector at the 25th_{percentile is Miscellaneous manufacturing (low capital}

intensity). The country at the 75th_{percentile of capital endowment is Finland (high capital}

endowment) and the country at the 25th_{percentile is Slovakia (low capital endowment). I set}

(30)

the 25th_{and 75}th_{percentile values for lnKL*lnCI in the BKC dataset for 1995, the coefficient}

estimate predicts that the exports in the sector of Tobacco products should be 1016 mio. USD higher than the exports in the sector of Miscellaneous manufacturing, and in real terms, in Finland as compared to Slovakia. This difference is 13.39% of the sectoral average exports in 1995; therefore, the effect of the FEH is fairly large. The same computation is made for each interaction term, dataset, and year. Evidently, the countries at the 75th_{and 25}th_percentile

vary with measure and year, as do the industries with PAC and CI. The results are reported in Table 7 for 1995 and in Table 8 for 2011.

Table 7: Magnitude of results in 1995

Differentials relative to average exports in 1995

BKC measure EFC measure WIOD measure

ERi * PACj 2.82% 0.97% 1.51%

lnKLi * lnCIj 13.39% 9.43% 11.91%

The results show that the magnitude of the FEH’s effect is similar across datasets and years with a value ranging from 9.43% to 15.51%. The magnitude of the interaction term measuring the pollution halo and the pollution haven effect is also similar across specifications with a value between 0.97% and 3.07%. Overall, the magnitude for the FEH interaction term is quite large, while the magnitude of the ER*PAC term is relatively low compared to the average sectoral exports.

Table 8: Magnitude of results in 2011

Differentials relative to average exports in 2011

BKC measure WIOD measure

ERi * PACj 1.09% 3.07%

lnKLi * lnCIj 15.51% 10.77%

(31)

in one case in 2011 for the WIOD measure: the coefficient of the first interaction term in equation (2) measuring the pollution haven and the pollution halo effect turns positive only in this instance and remains significant at the ten-percent level with a coefficient of 0.259. In the rest of the cases, coefficients are significant at least at the ten percent and support the pollution halo hypothesis and the FEH. Moreover, it is reassuring that most of the results are not only on the edge of being significant: the BKC measure, which works with the preferred indicator for environmental regulation has coefficients significant at the one-percent level in 1995, a ß1 coefficient significant at the five-percent level and a ß2 coefficient at the

one-percent level in 2011. Nevertheless, nine out of ten coefficients are significant at the five-percent level, three out of which are also significant at the one-five-percent level. In conclusion, it is convincing that the coefficients are significant in all cases and the sign of the coefficients only changes once out of all cases.

6. Discussion

The results and robustness of the empirical investigation allow to draw conclusions regarding the hypotheses of this thesis elaborated in Section 3.2. The preferred specification is the BKC measure with gross exports, as the data quality is complete over both years and most reliable in that case; the EFC measure does not provide data on ER for the year 2011 and the WIOD measure does not use a variable directly measuring stringency of regulation, instead it uses a measure based on sectoral intensities. However, it is reassuring that all regressions based on different datasets produce similar results, as discussed in the previous section. Hypothesis H1a regarding the PHH is stated as follows:Less stringent environmental regulation leads to higher volumes of exports in sectors where the PACs are higher. Thus, the significant positive sign of the ß1 coefficient should confirm this statement. However, the results lead to the rejection of

the hypothesis, which directs to hypothesis H1b, representing the pollution halo hypothesis, stated as follows: More stringent environmental regulation leads to higher volumes of exports in sectors where the PACs are lower. Based on this hypothesis, the expected sign of the ß1

coefficient is negative; therefore, H1b cannot be rejected, confirming the effect of the pollution halo hypothesis. Lastly, hypothesis H2 regarding the FEH is stated as follows: More capital abundant countries exhibit higher volumes of exports in sectors where the capital intensities are higher. The significant and positive ß2 coefficients corroborate this statement;

(32)

explain that the specialization of nations amongst polluting sectors is dominated by well-regulated countries having better technologies to produce with less emissions and by the wealthy nations in capital goods attracting dirty industries.

(33)

for the PHH may lie in the neglect of the pollution halo hypothesis by the authors and the specific region pairs that are designed to amplify the PHH effect according to theory, similarly as in the study of Cole and Elliott (2003), where the PHH was insignificant until North-South approach was implemented. Levinson and Taylor (2008) examine trade flows and environmental regulations to search for evidence of the PHH with the inclusion of PACs. The authors use North-South country pairs with a panel dataset from 1977-1986. They find that net imports can be negatively correlated with sectoral PACs, which undermines the pollution haven effect. This is some evidence for the pollution halo effect, even though the authors continue with the analysis to search for the PHH effect, which they find in a different setting with an instrumental variable approach. This partial evidence regarding the PACs supports the finding for the halo effect of this thesis, while the significant PHH effect may be because the authors neglect the FEH, which is found significant in my analysis. Grether et al. (2012) examines the PHH and the FEH by decomposing the pollution content of imports. The authors perform an analysis for 1987 involving 10 different pollutants, 48 countries, and 79 sectors in a North-South setting. The analysis finds some significant effects for both hypotheses, although not systematic across regressions. Moreover, the authors note that the effects may be overestimated because of the opposing force of the technique effect. Table 9 summarizes these studies with their most relevant details in order for the reader to easier put the results of this thesis into context.

Table 9: Summary of related studies

All evidence considered, with its multi country setting that covers 40 countries, this thesis strengthens the findings of studies that emphasize the importance of advanced production

Study Investigated

period Country sample Pollution component

Examined outcome Grossman framework effect Investigated hypotheses Results Antweiler et al. (2001) 1971-1996 43 developed and developing countries Pollution concentration (SO2) Trade intensity scale, composition, technique PHH, FEH no evidence for PHH, evidence for FEH Cole & Elliott

(2003) 1995

60 developed and developing countries

EFC measure, energy

intensity Exports composition PHH, FEH

partial evidence for PHH, evidence for FEH Azhar & Elliott

(2007) 1969-1996

US–Asia, Japan–Asia,

UK–Asia, US–Lat. Am. Pollution intensity Exports composition PHH, FEH

partial evidence for PHH and FEH Levinson & Taylor

(2008) 1977-1986

US-Canada, US-Mexico

Measure derived

from PAC Imports

composition, technique

PHH, pollution halo hypothesis

partial evidence for PHH and the pollution halo

hypothesis Grether et al. (2012) 1987 48 developed and developing countries Emission (for 10

different pollutants) Imports composition PHH, FEH

(34)

techniques to reduce polluting activity, which can be supported by well-designed environmental regulations and confirms the significance of the FEH, of which the effect is found to be present in all related studies which consider factor endowments.

7. Conclusion

This thesis provided an inclusive analysis investigating the three main hypothesis regarding the relationship of international trade and environmental pollution: the PHH, the pollution haven hypothesis, and the FEH. According to theoretical considerations, these hypotheses have opposing effects. However, the empirical literature has neglected the importance of the pollution halo hypothesis and the FEH, which is why often ambiguous results were found examining theories without the inclusion of all main influencing factors. This study aimed to overcome this limitation by developing a framework that includes the factors of all three hypotheses. The research question asked was whether these hypotheses had a significant effect in influencing export patterns.

The empirical approach was derived from theory based on comparative advantages influencing the specialization of a country. Previous empirical evidence suggested that comparative advantage was driven by the stringency of environmental regulations, pollution abatement technologies, and capital-labor endowments. Thus, an interaction estimating equation was developed to be able to examine specialization in exports by employing sectoral data including the effects of the PHH, the pollution haven hypothesis, and the FEH. The investigation was conducted based on three datasets for the years 1995 and 2011, including 40 countries and 19 sectors. The different datasets were constructed in order to check the robustness of results, since there are several data availability issues.

(35)

technology revolution allowed new techniques and better processes to be transmitted from developed innovative countries to their emerging peers easily, it is important for the destination country to be able and willing to adapt. Thus, policy makers play an essential role in incentivizing and facilitating this adoption process. The findings of this thesis show that well-designed environmental policies can indeed help shape the export pattern of a country towards cleaner production. Therefore, policy makers should not engage in a race to the

bottom type competition regarding environmental regulations to attract economic activity

with companies operating in dirty industries, since the PHH’s effect is subdued by the pollution halo effect. In contrast to the halo effect, the significance of the FEH points to the prevalence of capital mobility issues, whether it means entities are not willing to move their capital or the move itself is exogenously restricted. This study finds the effect of FEH to be significant, even more so than the pollution halo effect, which means that well-endowed countries specialize towards dirty exports. However, in this case it is hard to say what is to be done by policy makers, since relocating production simply induces pollution elsewhere. Even though factor mobility has improved significantly in the last century, the FEH is still prevalent and not losing significance, as the significant positive change coefficient between the examined years showed previously from equation (4). Nevertheless, the mobility of capital is important to further improve, as it makes it easier to optimize the locations of factories, possibly lowering pollution levels by reducing emissions related to freight. Thus, the results stress the importance of policymakers to facilitate environmental pollution control and the enhancement of capital mobility.

(36)

sectors included and for the same reason, only US measures for sectoral data could be considered. Due to this reason, the constructed datasets included 19 sectors, for which all necessary data was available. The assumption that the US sectoral measures are relevant for other countries cannot be formally tested because of the missing data; however, using the data this way is reasonable as elaborated in Section 4. Thus, without this formal test, the reverse causality concern mentioned in Section 3.1. cannot be fully eliminated with a statistical method in this setting, for example, with a relevant tested instrument. Nevertheless, it is reassuring that the related literature on the subject suggests that causality runs as it is in the setup of this study’s model. Furthermore, omitted variables are unlikely, but it is possible that a variable is omitted that varies both across countries and industries, thus its effect is not captured by the country- or industry-fixed effects. In general regarding data availability, the main limitation is the few developing countries included. Although they represent most of the economic activity in the developing world, the results may be different with a full country sample for the world: the PHH effect may be more dominant as the developing world has less advanced regulatory system, and the effect of FEH may be less dominant as these countries are mainly labor abundant. Regarding the emitted matters considered, this thesis employed carbon emissions for the construction of pollution intensities. It may be beneficial to consider other major environmentally damaging substances that are emitted due to economic activity. Lastly, the empirical model focused on the composition effect of the Grossmann framework. The scale and technique effect were not directly tested, albeit the dominance of the pollution halo hypothesis suggests that the technique effect being good for the environment is present.

(37)

The contribution of this thesis to the existing literature is that it conducted an inclusive simultaneous investigation regarding the PHH, the pollution haven hypothesis, and the FEH, as at least one of them are neglected in existing papers. Furthermore, the multi-country setup allowed to test the hypotheses on a wide sample of countries to provide more robust results than the usual one-country or country-pair studies in the empirical literature. Lastly, the previous evidence was updated with new findings: the results for 2011 emphasize the previous findings and establish that the effects are still dominant in influencing export patterns.

(38)

Appendix

Appendix 1: Country List WIOD

countries

Botta & Kozluk datasets' countries

Eliste & Fredriksson datasets' countries

Australia Australia Australia

Austria Austria Austria

Belgium Belgium Belgium

Brazil Brazil Brazil

Bulgaria - Bulgaria

Canada Canada Canada

China China China

Cyprus - -

Czech Republic Czech Republic Czech Republic

Denmark Denmark Denmark

Estonia - -

Finland Finland Finland

France France France

Germany Germany Germany

Greece Greece Greece

Hungary Hungary Hungary

India India India

Indonesia Indonesia -

Ireland Ireland Ireland

Italy Italy Italy

Japan Japan Japan

Republic of Korea Republic of Korea Republic of Korea

Latvia - -

Lithuania - -

Luxembourg - -

Malta - -

Mexico - Mexico

Netherlands Netherlands Netherlands

Poland Poland Poland

Portugal Portugal Portugal

Romania - -

Russia Russia -

Slovak Republic Slovak Republic Slovak Republic

Slovenia Slovenia -

Spain Spain Spain

Sweden Sweden Sweden

Taiwan - -

Turkey Turkey Turkey

(39)

Appendix 2: Summary Statistics BKC measures for 1995 Nr. of

observations Mean

Standard

deviation Minimum Maximum Exports 570 7 588.84 14 875.01 0.00 127 920.40 ln Exports 570 7.79 1.58 0.00 11.76 ER 570 0.94 0.49 0.33 1.98 PAC 570 1.61 2.16 0.20 9.90 KL 570 179 060.90 85 515.45 7 280.61 309 548.80 ln KL 570 11.82 0.98 8.89 12.64 CI 570 52 600.00 30 795.99 23 100.00 129 400.00 ln CI 570 10.73 0.51 10.05 11.77

EFC measures for 1995 Nr. of

observations Mean

Standard

WIOD measures for 1995 Nr. of

observations Mean

Standard

(40)

Appendix 2: Summary Statistics (continued) BKC measures for 2011

Nr. of

observations Mean

Standard

WIOD measures for 2011 Nr. of

observations Mean

Standard

(41)

References

Ambec, S., Mark A. Cohen, M. A., Elgie, S., Lanoie; P. (2013). The Porter Hypothesis at 20: Can Environmental Regulation Enhance Innovation and Competitiveness? Review of

Environmental Economics and Policy, 7(1), 2–22, https://doi.org/10.1093/reep/res016

Ang, J. B. (2009). CO2 Emissions, Research and Technology Transfer in China. Ecological

Economics, 68(10), 2658-2665. https://doi.org/10.1016/j.ecolecon.2009.05.002

Antweiler, W., Copeland, B., & Taylor, M. (2001). Is Free Trade Good for the Environment?

The American Economic Review, 91(4), 877-908.

https://dx.doi.org/10.1257/aer.91.4.877

Azhar, A., & Elliott, R. (2007). Trade and Specialisation in Pollution Intensive Industries: North-South Evidence. International Economic Journal,21(3), 361-380.

https://doi.org/10.1080/10168730701529926

Birdsall, N. & Wheeler, D. (1993). Trade Policy and Industrial Pollution in Latin America: Where are the Pollution Havens? Journal of Environment and Development, 2(1), 137– 149. https://doi.org/10.1177/107049659300200107

Botta, E. & T. Kozluk (2014). Measuring Environmental Policy Stringency in OECD Countries: A Composite Index Approach. OECD Economics Department Working Papers, No. 1177, OECD Publishing, Paris. http://dx.doi.org/10.1787/5jxrjnc45gvg-en

Brunnermeier, S., & Levinson, A. (2004). Examining the Evidence on Environmental

Regulations and Industry Location. The Journal of Environment & Development, 13(1),

6-41. https://doi.org/10.1177/1070496503256500

Cai, X., Che, X., Zhu, B., Zhao, J., & Xie, R. (2018). Will Developing Countries Become Pollution Havens for Developed Countries? An Empirical Investigation in the Belt and Road.

Journal of Cleaner Production, 198, 624-632.

https://doi.org/10.1016/j.jclepro.2018.06.291

Candau, F., & Dienesch, E. (2017). Pollution haven and corruption paradise. Journal of

Environmental Economics and Management, 85(1), 171-192.

(42)

Chor, D. (2010). Unpacking Sources of Comparative Advantage: A Quantitative Approach.

Journal of International Economics, 82(2), 152-167.

https://doi.org/10.1016/j.jinteco.2010.07.004

Cole, M., & Elliott, R. (2003). Determining the Trade-Environment Composition Effect: The Role of Capital, Labor and Environmental Regulations. Journal of Environmental

Economics and Management, 46(3), 363-383.

https://doi:10.1016/S0095-0696(03)00021-4

Cole, M., & Elliott, R. (2005). FDI and the Capital Intensity of “Dirty” Sectors: A Missing Piece of the Pollution Haven Puzzle. Review of Development Economics, 9(4), 530-548.

https://doi:10.1111/j.1467-9361.2005.00292.x

Copeland, B., & Taylor, M. (1994). North-South Trade and the Environment. Quarterly

Journal of Economics, 109(3), 755-755. https://doi.org/10.2307/2118421

Copeland, B., & Taylor, M. (2004). Trade and the Environment: Theory and Evidence. Princeton, NJ: Princeton University Press.

Dornbusch, R., Fischer, S., & Samuelson, P. (1977). Comparative Advantage, Trade, and Payments in a Ricardian Model with a Continuum of Goods. The American Economic

Review, 67(5), 823-839. Retrieved from: https://www.jstor.org/stable/1828066

Eliste, P., & Fredriksson, P. (2002). Environmental Regulations, Transfers, and Trade: Theory and Evidence. Journal of Environmental Economics and Management, 43(2), 234-250.

https://doi.org/10.1006/jeem.2000.1176

Eskeland, G., & Harrison, A. (2003). Moving to Greener Pastures? Multinationals and the Pollution Haven Hypothesis. Journal of Development Economics, 70(1), 1-23.

https://doi:10.1016/S0304-3878(02)00084-6

Feenstra, R. C., Inklaar, R., & Timmer, M. P. (2015). "The Next Generation of the Penn World Table" American Economic Review, 105(10), 3150-3182.

(43)

Garsous, G., Kozluk, T. (2017). Foreign Direct Investment and the Pollution Haven Hypothesis - Evidence from Listed Firms. OECD Economics Department Working Papers, No. 1379. OECD Publishing, Paris. https://dx.doi.org/10.1787/1e8c0031-en

Grether, J., Mathys, N., & De Melo, J. (2012). Unravelling the Worldwide Pollution Haven Effect. The Journal of International Trade & Economic Development, 21(1), 131-162.

https://doi.org/10.1080/09638190903552040

Grether, J., Mathys, N., & Melo, J. (2006). Unraveling the Worldwide Pollution Haven Effect.

Policy Research Working Paper Series, No. 4047, The World Bank.

https://doi.org/10.1080/09638190903552040

Grossman GM (1995) Pollution and Growth: What Do We Know? In: The Economics of Sustainable Development. Cambridge: Cambridge University Press, 19-47.

Grossman, Gene M & Alan B. Krueger (1995). Economic Growth and the Environment.

Quarterly Journal of Economics, 110(2), 353-77. https://doi.org/10.2307/2118443

Halicioglu, F., & Ketenci, N. (2016). The Impact of International Trade on Environmental Quality: The Case of Transition Countries. Energy, 109(C), 1130-1138.

https://doi.org/10.1016/j.energy.2016.05.013

Kozluk, T. & Timiliotis, C. (2016). Do Environmental Policies Affect Global Value Chains?: A new Perspective on the Pollution Haven Hypothesis. OECD Economics Department

Working Papers, No. 1282, OECD Publishing, Paris,

https://doi.org/10.1787/5jm2hh7nf3wd-en

Levinson, A. (2008) Pollution Haven Hypothesis. London: The New Palgrave Dictionary of Economics.

Levinson, A., & Taylor, M. (2008). Unmasking the Pollution Haven Effect. International

(44)

Lucas, R.E.B., Wheeler D. & Hettige, H. (1992) Economic Development, Environmental Regulation and the International Migration of Toxic Industrial Pollution: 1960-88. Policy

Research Working Paper Series, No. 1062, The World Bank. Retrieved from:

http://documents.worldbank.org/curated/en/673921468765926548/Economic- development-environmental-regulation-and-the-international-migration-of-toxic-industrial-pollution-1960-88

Manderson, E., Kneller, R. (2012). Environmental Regulations, Outward FDI and

Heterogeneous Firms: Are Countries Used as Pollution havens? Environmental and

Resource Economics, 51(3), 317-352. https://doi.org/10.1007/s10640-011-9500-z

Mani, M. & Wheeler D. (1998) In search of pollution havens? Dirty Industry in the World Economy, 1960-1995. Journal of Environment and Development, 7(3), 215–247.

https://doi.org/10.1177/107049659800700302

Millimet, D. L., & Roy, J. (2016) Empirical Tests of the Pollution Haven Hypothesis When Environmental Regulation is Endogenous. Journal of Applied Econometrics, 31(4), 652– 677. https://doi.org/10.1002/jae.2451

Peters, G.P., Weber, C.L., Guan, D., Hubacek, K. (2007). China's Growing CO2 Emissions - A Race between Increasing Consumption and Efficiency Gains. Environmental Science &

Technology. 41(17), 5939-5944. https://doi.org/10.1021/es070108f

Ren, S., Yuan, B., Ma, X., Chen, X. (2014). International Trade, FDI (Foreign Direct Investment) and Embodied CO2 Emission: A Case Study of Chinas Industrial Sectors. China Economic

Review. 28, 123-134. https://doi.org/10.1016/j.chieco.2014.01.003

Rezza, A.A. (2013). FDI and Pollution Havens: Evidence from the Norwegian Manufacturing Sector. Ecological Economics. 90, 140-149.

https://doi.org/10.1016/j.ecolecon.2013.03.014

Romalis, J. (2004). Factor Proportions and the Structure of Commodity Trade. The American

Economic Review,94(1), 67-97. https://doi.org/10.1257/000282804322970715

(45)

Sorsa, P. (1994). Competitiveness and Environmental Standards: Some Exploratory Results.

Policy Research Working Paper Series, No. 1249. The World Bank. Retrieved from:

http://documents.worldbank.org/curated/en/520381468764426037/Competitiveness-and-environmental-standards-some-exploratory-results

Stern, D. (2017). The Environmental Kuznets Curve After 25 Years. Journal of Bioeconomics, 19(1), 7-28. https://doi:10.1007/s10818-017-9243-1

Taylor, M. (2005). Unbundling the Pollution Haven Hypothesis. Advances in Economic

Analysis & Policy, 3(2), 1-28. https://doi:10.2202/1538-0637.1408

Taylor, M. (2005). Unbundling the Pollution Haven Hypothesis. Advances in Economic

Analysis & Policy, 4(2), 1408-1408. https://doi.org/10.3171/foc.2005.19.5.1

Timmer, M., Dietzenbacher, E., Los, B., Stehrer, R., & De Vries, G. (2015). An Illustrated User Guide to the World Input–Output Database: the Case of Global Automotive Production.

Review of International Economics, 23(3), 575-605. https://doi.org/10.1111/roie.12178

Trade Patterns and Environmental Pollution:

University of Groningen

Faculty of Economics and Business

MSc International Economics and Business

Double Degree with Corvinus University of Budapest

Master Thesis

Trade Patterns and Environmental Pollution:

An analysis of the popular pollution hypotheses

Supervisor: Prof. dr. Robert Inklaar

Co-assessor: dr. Péter Isztin

Barna Hunor Horváth

b.h.horvath@student.rug.nl

Abstract

Table of Contents

List of abbreviations

1. Introduction

2. Literature review

3. Methodology

4. Data

5. Results

6. Discussion

7. Conclusion

Appendix

References