The impact of economic development on the
environment in emerging markets
Bachelor thesis by Eva Maria Langeler, Faculty of Economics and Business, University of Amsterdam
Abstract
Emerging markets show a pattern of economic development through industrialization, liberalization and a growing free-market system leading to changes in income distribution and employment patterns. This economic development can lead to environmental degradation since industrialization and agricultural modernization increase pollution in terms of emission. The conventional Environmental Kuznets Curve (EKC) describes how pollution will rise in the early stages of economic development, but will turn down due to environmental awareness, policy and technology after a certain income value. The monotonic EKC states that traditional pollutants and GDP might show an inverted U-shaped relationship, but “new toxics” will monotonically increase in volume because they are not substituted. I estimate the impact of economic development on environmental degradation in terms of CO2,
CH4, N2O and F-gas emissions, taking into account the possible effect of population
density, country and time fixed effects. I find some evidence for a significant effect of GDP per capita on emissions, but population density, country and time specific factors may play an essential role as well.
Name: Eva Maria Langeler Student number: 10218645 Faculty: Economics and Business Field: Macroeconomics
Supervisor: Dr. Koen Vermeylen Date: 20 January 2015
Table of Content
1. Introduction 2
2. Theoretical framework
2.1 Economic development in emerging markets 3
2.1.1 Benefits of economic development at an environmental cost 4 2.1.2 An economic pattern in emerging markets 5
2.1.3 Population density and increasing GDP 6
2.2 Emission as a cost to the environment 7
2.2.1 Overview of greenhouse gases 8
2.2.2 A determinant of environmental quality: population density 11 2.2.3 Country specific factors affecting emission 12 2.2.4 International environmental policy and technological change 13 2.2.5 Environmental degradation in emerging markets 14 2.3 Models for the relationship between pollution and
economic development 15 2.3.1 The conventional Environmental Kuznets Curve 16
2.3.2 A contemporary model: a monotonic EKC 17
3. Methodology
3.1 Introduction 19
3.2 Data 20
3.3 Data descriptive: dependent, independent and dummy variables 20
3.4 Empirical model and procedure 22
4. Results 4.1 Results CO2 23 4.2 Results CH4 25 4.3 Results N2O 26 4.4 Results F-gas 27 5. Conclusion 29 Bibliography 33 Tables 36 Graphs 48
1. Introduction
“The disruption of the world's ecological systems — from the rise of global warming and the consequent damage to our climate balance to the loss of living species and the depletion of ocean fisheries and forest habitats — continues at a frightening rate. Practically every day it becomes clearer to us that we must act now to protect our earth while preserving and creating jobs for our people.” Al Gore’s statement shows that environmental awareness is growing whilst keeping the simultaneous importance of economic development in mind. Emerging economies see their industries grow, which can be at a cost to the environment in terms of emission (Stern, 2004).
Selden and Song (1994) state that pollution is negatively impacted by economic development through industrialization and agricultural modernization, but that market forces and regulation can turn this effect down thereby improving the environment at a certain degree of income per capita. The relationship is regarded as the conventional Environmental Kuznets Curve (EKC). Recent research on the other hand does not always support this argument as economic development can bring about a continuous increase in greenhouse gas emission when pollutants have not yet been replaced, which refers to a monotonic EKC (Stern, 2004). This ambiguous relationship can additionally be affected by country specific factors such as population density, since a higher density creates a larger incentive to diminish emission based on health and monetary reasons (Selden & Song, 1994). Time related effects include modern environmental policies such as the Kyoto Protocol and cleaner technologies, which (together) help reduce environmental degradation (Jaffe, Newell, & Stavins, 2002).
There is however a gap within existing literature about the relationship between emission and economic development as it has mainly addressed substituted pollutants such as nitrogen oxides and sulfur (dioxide) and these studies seem outdated by 10 years. Furthermore primarily developed countries have been studied whilst emerging markets seem important due to their global economic rise since 2000. To investigate the impact of economic development on environmental quality in these markets this paper will focus on the research question: “Does economic development impact environmental degradation in terms of emission in emerging economies and what specific characteristics in these countries could be major drivers?”
In order to answer this question literature will be studied and empirical research based on an econometric model of the EKC shall be performed. The literature will outline assumptions and mechanisms concerning economic development and environmental pollution. Furthermore the empirical research will test the impact of economic development on the environment in terms of emission. The underlying variable is emission and will be represented by new pollutants (carbon dioxide and F-gas) and traditional pollutants (methane and nitrous oxide). The explanatory variable depicting economic development is GDP per capita and population density is used to show the degree of urbanization. Tests prove with some certainty that both new and traditional pollutants are significantly affected by GDP per capita. Furthermore there is some proof for the significant impact of population density, country fixed and time fixed factors.
The rest of this paper proceeds as follows. Section 2 will discuss relevant literature about pollution from emission and the impact of economic development. Section 3 involves the methodology that describes the data and the empirical model used in this research. The results of the empirical research are presented in section 4, and section 5 contains the conclusion (with an integrated discussion). The bibliography, tables and graphs follow these sections.
2. Theoretical framework
2.1 Economic development in emerging markets
Long-run economic trends since the industrial revolution have been broadly studied. Economic development refers to the increasing value of output in terms of goods and services within country borders over time thereby boosting living standards (Kravis, Heston, & Summers, 1978). It is usually measured by real (inflation adjusted) gross domestic product (GDP hereafter) per capita. Factors causing economic development are productivity growth, investment and availability of resources. Productivity can grow through technological innovation, for example portrayed by the 19th century industrial revolution that was based on mechanization. Mechanization took place through innovative production of energy from water and coal and the use of
modernized tools thereby enhancing production power. This led to real wage increases and higher life expectancies (Lucas, 2002). Investments in human and physical capital reduce costs of production by enhancing knowledge and production capacity (Lucas, 2002). Additionally when more natural resources are available, the cost of using them will diminish thereby reducing production costs. This can potentially create a competitive advantage, which possibly results in an economic boost. Furthermore the historical pattern of economic development in the world between 1990 and 2013 is depicted in figure 1. This shows a long-term trend of economic development interrupted by the 2008 financial crisis, which was characterized by financial market failure and bailout plans (Shah, 2009). During the period between 1980 and 1990 the United States of America, the European Union and Japan have been leaders in economic expansion (World Bank, 2013b). This was followed by a 10-year period during which the United States dominated in economic development, but between 2000 and 2010 the development of emerging economies prevailed (World Bank, 2013b).
Figure 1: World GDP per capita. Source: http://data.worldbank.org/indicator/NY.GDP.PCAP.PP.KD
2.1.1 Benefits of economic development at an environmental cost
As stated before economic development is based on the real incremental market value of national output. The key advantage of this is the improvement of living standards due to increasing consumption possibilities (Komlos, 1994). This concept is often linked to utility and prosperity increases. Living standards might also be boosted due
0 2000 4000 6000 8000 10000 12000 14000 16000 GD P p er c ap it a (c on st an t 2 01 1 in ter na tio na l US $) Year
World GDP per capita trend
World GDP per capita trend
to improved public services from tax increases collected from incremental production. These public services include education and healthcare. The main drawback of economic development is environmental degradation (Brock & Taylor, 2005).
The theoretical relationship between economic development and
environmental degradation is based on the use of fossil fuels and toxics (e.g. fertilizers) in industrial and agricultural activities creating (air) pollution. This can be harmful to the earth along with human health and the ecosystem (Grossman & Krueger, 1994). This implies “limits to growth” (Brock & Taylor, 2005) which does not only refer to natures limited resources, but also the finite capacity of the earth to act as a sink for human waste. This waste can take the shape of both garbage and toxic chemicals from industrial activity. When economic development takes place and output value increases, these harmful byproducts also expand in size. This in turn causes environmental quality to fall from the point where nature cannot absorb waste anymore (Brock & Taylor, 2005). These externalities will then also reduce social welfare in terms of health.
2.1.2 An economic pattern in emerging markets
Between 2000 and 2010 emerging markets have dominated economic development in the world (World Bank, 2013b). These low-income countries show fast growth through economic liberalization and can be defined as either developing countries or transition economies (Hoskisson, Eden, Lau, & Wright, 2000). Developing countries include Asian, Latin American, African and Middle Eastern markets whilst transition economies are China, the Soviet Union and countries situated in Western Europe. According to the list compiled by the International Monetary Fund (2012) emerging markets include Argentina, Brazil, Bulgaria, Chile, China, Colombia, Estonia, Hungary, India, Indonesia, Latvia, Lithuania, Malaysia, Mexico, Pakistan, Peru, Philippines, Poland, Romania, Russia, South Africa, Thailand, Turkey, Ukraine and Venezuela. Their typically positive economic development pattern is shown in figure 2. This graph includes all IMF listed emerging economies and shows a steep increase between 2000 and 2010 since these countries took the lead in sustaining and recovering from the credit crisis (World Bank, 2013b). Brazil and India were major contributors to world output enlargement after 2010 with respectively 9 and 5 percent (International Monetary Fund, 2010).
Figure 2: GDP per capita for emerging markets. Source: http://data.worldbank.org/indicator/NY.GDP.PCAP.PP.CD
Features of economic development in emerging markets identified by Kuznets (1955) primarily includes changes in income distribution due to mutations in national output shares between the agricultural, industrial and service sector. Furthermore employment patterns change moving human capital from the agricultural to the industrial and service sector, which in turn brings about increased urbanization as people prefer to live closer to their job (Kuznets, 1955). These changes come about by industrialization, liberalization and stimulating a free-market system (Hoskisson et al., 2000). Industrialization refers to a period in time characterized by modernization in terms of social and economic change led by technological innovation and large-scale energy production (Rosenstein-Rodan, 1943). Liberalization focuses on government policy becoming less restricted in terms of social, economic and political affairs thereby stimulating openness and freedom (Hoskisson et al., 2000). Free trade emerges when openness is taken one step further into international markets thereby promoting import and export (Antweiler, Copeland, & Taylor, 1998). As mentioned earlier: a side effect of this development into economic prosperity is environmental pollution (Brock & Taylor, 2005).
2.1.3 Population density and increasing GDP
Economic development thus features increased urbanization as workers migrate 0 2000 4000 6000 8000 10000 12000 14000 16000 GD P p er c ap it a (c on st an t 2 01 1 in ter na tio na l US $) Time
GDP per capita over time
"Emerging markets total average"
towards industrial areas for job opportunities. This increases population density in the urban areas. In this subsection the relation between economic development and population density (the amount of people per square kilometer of land) will be analyzed.
A higher fertility rate leads to population growth and thereby increased population density. This increase in population adds to the potential workforce thereby creating a possible competitive advantage, which in turn could lead to economic development. However according to Coale and Hoover (1958) an increase in population density through population growth does not guarantee an increasing GDP, because the growing potential work force could solely add to the pool of people who are unemployed because of competition. Furthermore a fast increase of consumers will constrain a rise in output value as current resource consumption restricts the use of resources for economic development in the future. It is thus suggested that societies with a lower population density are less burdened by consumers thereby creating the opportunity for more effective and extensive economic development plans (Coale & Hoover, 1958). These findings advocate a negative relationship between population density and GDP. However this research assumes that mortality rate is independent of economic development, which does not necessarily hold, especially not in developing countries since insufficient funds for primary needs (e.g. nutrition) can cause death. Conclusions should therefore be drawn carefully in terms of the discussed relationship.
2.2 Emission as a cost to the environment
In the above paragraph on economic development and emerging economies it has been stressed that economic development negatively affects the environment. This is brought about by greenhouse gases, which are gases in the atmosphere that emit radiation and absorbs heat thereby storing it into the atmosphere (United States Environmental Protection Agency, 2013b). The past and current increase in pollution concentrations cause ocean warming, changing temperatures and extremes: climate change (Intergovernmental Panel On Climate, 2007). These past 100 years there has already been an atmospheric temperature increase of about 0.50C (Wuebbles & Edmonds, 1991). It is expected that within a few decades average temperatures will
rise between 30C and 100C indicating significant climate change (Intergovernmental Panel On Climate, 2007). Main greenhouse gases include the traditional pollutants methane and nitrous oxide and new pollutants like carbon dioxide and fluorinated gases that have replaced other pollutants (Stern, 2004). Since 2004 the main sources of these pollutants on a global scale are energy supply (26%), industry (19%), the clearing of land (17%), agriculture (14%) and transportation (13%) (United States Environmental Protection Agency, 2013a). In the following paragraphs these main greenhouse gases will be discussed.
2.2.1 Overview of greenhouse gases
The first main gas to be discussed is carbon dioxide (CO2): a greenhouse gas that
emerges through the use of natural gas and other fossil fuels such as coal and oil (Intergovernmental Panel On Climate, 2007). Furthermore this gas is emitted when burning objects made out of wood or through chemical reactions (United States Environmental Protection Agency, 2013b). Major sectorial sources of CO2 related to
burning of fossil fuels are electricity (38%), transportation (32%) and industry (14%) (United States Environmental Protection Agency, 2013b). The industrial sector also involves emission from transformation processes in terms of minerals and chemicals. Between 1990 and 2008 emission has globally incremented by approximately 50% and the corresponding CO2 emission accounts for 74% of global greenhouse gas
emissions (United States Environmental Protection Agency, 2013a). The pattern of CO2 emission is shown in figure 3, which stresses the extensive increase in volume.
One of the reasons for this is that over time the emission of sulfur and nitrogen oxides (SO2 and NOX) has decreased, which was replaced by CO2 thereby keeping aggregate
emission high (Stern, 2004). This pollutant is recognized as the most important greenhouse gas because it is extensively emitted in every sector that burns fossil fuels (Intergovernmental Panel On Climate, 2007).
Figure 3: World emission by gas. Source: http://cait2.wri.org/
The second primary pollutant is methane (CH4), which is the essential
chemical component of natural gas that can be found underneath the surface of the earth and is naturally emitted by wetlands (United States Environmental Protection Agency, 2013b). The lifetime of CH4 emission is about 10 years and lower than that
of CO2. However it is more damaging in the long run because it is highly efficient in
trapping heat (Wuebbles & Edmonds, 1991). Anthropogenic activities causing emission include agriculture (enteric fermentation in cattle raised and rice production, 43%) and industrial activities involving natural gas, petroleum systems and coal mining (39%) (Wuebbles & Edmonds, 1991). The U.S. emission trend between 1990 and 2012 shows a decrease of about 11% that is mainly related to the implementation of upgraded storage systems for processing natural gas and petroleum, which prevents major leakage (United States Environmental Protection Agency, 2013b). However on a global scale emission has increased by about 14% between 1991 and 2010, which is possibly based on incremental use of fertilizers in the agricultural sector. Figure 3 shows this long-term increase in methane emission.
About 150 years ago concentrations of the third main greenhouse gas, nitrous oxide emission (N2O), began to increase and it has 300 times the strength of CO2 in
terms of trapping heat into the atmosphere (United States Environmental Protection Agency, 2013b). Anthropogenic sources are only responsible for 40% of the
emission. This pollutant is essentially present in the stratosphere as it is part of the nitrogen cycle and reacts with ozone. Like methane nitrous oxide has several sources such as agriculture (75%), fossil fuels burnt in transportation (9%) and it is a byproduct of synthetic products in industrial activities (6%) (United States Environmental Protection Agency, 2013b). In agriculture, nitrous fertilizers used in soil management or cattle urine is broken down in the soil through which emission emerges. In figure 3 the approximate 8% global increase of N2O in the atmosphere
between 1991 and 2010 is depicted. The main reason for this increase is the incremental disruption of soil in agricultural activities, but it is also explained by the use of electrical power in transportation (United States Environmental Protection Agency, 2013b).
Fourth, fluorinated gases are strictly anthropogenic emissions that have an atmospheric lifetime of thousands of years, which is extreme compared to other greenhouse gases. Gases included in this category are hydrofluorocarbons (HFCs), perfluorocarbons (PFCs) and sulfur hexafluoride (SF6). The main component of F-gas
is HFC, which is a substitute for the ozone depleting substance chlorofluorocarbon (CFC) that is emitted by refrigerators, air-conditioners, foams and aerosols for example. After emission the excess that cannot be removed by the earth reacts with ozone thereby harming the environment through depletion of the ozone layer (United States Environmental Protection Agency, 2013b). The substitution of CFCs by HFCs took place after forcing the Montreal Protocol on Substances that Deplete the Ozone Layer in 1989 thereby phasing out CFCs in order to save the ozone layer (Montreal Protocol, 1987). This is also the reason why the global trend of F-gases in figure 3 shows an increase of over 150% during 1991 and 2010 (World Resources Institute, 2014). During the first 6 years the increase was only 10% due to the fact that old products such as refrigerators are gradually substituted. The effect is much stronger in the long run.
As stated earlier there is an impact of economic development on pollution in terms of emission. Analyzing the key sources of the above mentioned main greenhouse gases provides some proof for this as both energy supply and industry are key components within economic development and industrialization (Lucas, 2002). However another general determinant of environmental degradation in terms of emission is population density, which will be discussed in the following subsection.
2.2.2 A determinant of environmental quality: population density
The Environmental Protection Agency (2010) published a report on the relationship between air pollution and population density in the main metropolitan areas of the United States of America. This report concludes that higher population densities result in larger congestion, which in turn produces more pollution in terms of nitrogen oxides, carbon monoxide, volatile organic compounds and sulfur dioxide. It is argued that urbanization intensifies population density thereby increasing traffic density in the city. This argument can be extended in terms of energy use and the size of the industrial sector. A shortcoming of this research is that it focuses solely on congestion issues in the urban area. Selden and Song (1994) however firstly state that an increase in population density will lead to overall diminished transportation due to the fact that people live closer together. This would then reduce emission portraying a negative relationship. Furthermore they argue that countries with low population densities care less about environmental damage in terms of emission at every level of income compared to densely populated areas. The affirmed reason for this is that societal pressure for public health is more severe in densely populated areas, which can cause stricter regulations due to the political and monetary argument that preventing illness is presumed to be less costly than treating it. The following hypothesis is thus formed:
Hypothesis 1:
Sparsely populated countries are likely to be less concerned about reducing per capita emissions, at every level of income.
Furthermore Selden and Song (1994) assume that the relationship between population density and pollution is of linear nature for all greenhouse gases. There is some evidence in literature however that is in favor of a quadratic relationship for methane and nitrous oxide (Fernandes, Styger, & Costello, 2008). This implies a turning point in the decreasing trend where an increase of one unit of population density causes a smaller decrease in emission than before. The first possible reason for this is that as stated above, CH4 and N2O are mainly produced by agricultural
activities. Emission may decrease since rural area diminishes when population density increases because of urbanization thereby reducing agriculture (Fernandes et al., 2008). The turning point, where an increase of population density by one unit
decreases emission less than before, appears because a vast amount of agriculture is needed for consumption. Even though population density is increasing, this amount of agriculture cannot decrease beyond a certain point due to a minimum demand for agricultural products by the population (Cropper & Griffiths, 1994). Though this argument is significant, caution is advised because the effect depends on the severity of emission in other sectors, which can be country specific.
2.2.3 Country specific factors affecting emission
As mentioned above population density can be a country specific factor influencing the degree of environmental pollution in terms of emission. Other country fixed factors that vary across entities but stay the same over time are local culture, climate, geographic location and country size.
Local culture is a determinant of emission because attitudes towards the environment can influence polluting behavior and are often linked to local norms and values (Arcury, 1990). Culture can take multiple shapes and personal behavior can range from deciding against intensively using a car to installing solar panels on the roof. This is closely linked to politics as these cultural attitudes can influence policy through public pressure (Irwin, Simmons, & Walker, 1999). Local culture also includes the concept of education. Knowledge and attitude are positively related so when a specific culture does not value (scientific) knowledge about environmental pollution and long lasting effects such as the greenhouse effect, public action will not take place. This way low levels of knowledge can have a devastating effect on the environment (Arcury, 1990).
Another country specific factor influencing the degree of pollution is climate. The pollutants CH4 and N2O are primarily emitted through agricultural activities. These
activities are largely dependent on “the right climate” as this creates a production advantage that differs per country. Furthermore relatively hot countries in terms of temperature will demand more air-conditioning, which in turn will increase F-gas emission levels (United States Environmental Protection Agency, 2013b).
Geographic aspects also differ per country and influences emission through transportation. An example is Australia, which is relatively distanced from other continents and involves long national distances between major cities. This stimulates the use of fossil fuels in travelling.
The size of a nation is a country specific determinant of emission for the following reasons: firstly, size affects the amount of pollution because not only does policy differ per country, but the relative strength of the (policy) impact also varies due to the surface extent (Irwin et al., 1999). Secondly, larger countries can have higher absolute values of emission assuming constant emission density due to the relatively large surface.
Based on the arguments about possible country fixed effects mentioned above the following hypothesis can be formulated:
Hypothesis 2:
Country fixed effect is a determinant of the relationship between economic development and environmental degradation in terms of emission.
2.2.4 International environmental policy and technological change
The fixed effects mentioned above are strictly country specific factors by nature. However environmental policy that affects several countries within the area reviewed in terms of environmental degradation can be interpreted as time fixed since it stays constant across entities but varies over time. Apart from the Montreal Protocol an example of this is the Kyoto Protocol to the United Nations Framework Convention on Climate Change (2014) effective since 2005 that sets binding targets for reducing greenhouse gas emissions in industrialized countries including the entire United Nations and European Union (exceptions were made for Canada and the United States). Developing countries are not obliged to stick to the specific target, but the treaty compels them to reduce greenhouse gas emissions. Clean Development Mechanisms are implemented which allow developed countries to get involved in projects in developing countries that reduce emission. Furthermore international emission trading is now allowed and emission reduction is rewarded by distributing emission reduction units under the Joint Implementation mechanism (United Nations Framework Convention on Climate Change, 2014).
Furthermore Jaffe et al. (2002) state that technological change through invention, innovation and diffusion can aid in diminishing environmental degradation from business activities due to higher degrees of efficiency and cleaner technologies. The Jevons paradox however disproves this as the improvement of efficiency actually
leads to an increase in fuel use through the expansionary tendency that is inherent to these improvements and capitalism (Clark & York, 2005).
Regardless Jaffe et al. (2002) subsequently state that environmental policies such as pollution charges and subsidies stimulate pollution control efforts that can be reached through new technologies. Moreover Jaffe et al. (2002) affirm that technological innovation has an impact on economic development through industrialization and agricultural modernization, which implies that this affects the relationship between emission and economic development. Economic development in turn can stimulate technological progress through the availability of money for investment thereby portraying a reverse effect. More research is advised by Jaffe et al. (2002), so this analysis gives limited proof for the reverse effect and the linkage between technology and environmental policy as a determinant of long-term environmental quality. Since technological innovation changes rapidly over time it is categorized as time fixed. This must be said with caution as resources such as knowledge and money for developing technological change vary across countries.
2.2.5 Environmental degradation in emerging markets
The amount of greenhouse gas emission influences environmental degradation and differs across countries and time. In emerging economies the aggregate emissions of CO2, CH4, N2O and F-gases show the same sequence in terms of emission intensity
compared to global values. Carbon dioxide is the main determinant of emission changes in terms of volume and tons of CO2 change per year. This is followed by
methane, nitrous oxides and lastly F-gases.
It is expected that emerging markets show increasing emission values due to industrialization and agricultural modernization (Selden & Song, 1994). Noticeably for emerging economies the trend of CH4 and N2O differs compared to the global
values as they show an average decrease in emission between 1991 and 2010 of about 8 and 14%, respectively (World Resources Institute, 2014). As shown in figure 4 this emission reduction starts in the period towards 1996-2000. This is the period in which the Kyoto Protocol was signed (United Nations Framework Convention on Climate Change, 2014). The effectiveness of the treaty in having developing countries reduce greenhouse gas emission thus seems supported. These values show some proof for the claim of the United States Environmental Protection Agency (2013b) that methane
and nitrous oxide emission can be managed using upgraded equipment and pollution control technologies.
For F-gases the emission volume has increased during the given time frame with about 143%, which is slightly less compared to the global value. The reason for this could be that developing countries have less income to spend on luxury products that use F-gases such as air-conditioning, which makes the change less severe even though emission values still seem to be (positively) affected by demands of the Montreal Protocol (1987).
Figure 4: Emission of emerging markets by gas. Source: http://cait2.wri.org/
2.3 Models for the relationship between pollution and economic development
It has been mentioned throughout this paper that the major drawback of economic development is that it can lead to environmental degradation. A large amount of studies argue that this is true, but only for the early stages of economic development. Beyond a specific value of income per capita the pattern reverses implying that high-income levels will improve environmental quality thereby creating an inverted U-shaped curve (quadratic function that opens downwards): the conventional Environmental Kuznets Curve (EKC hereafter) (Stern, 2004). However more recent
0 1 2 3 4 5 6 1991-1995 1996-2000 2001-2005 2006-2010 Emission total (tCO2e per capita)
Time
Emission total Emerging Markets
CO2 CH4 N2O F-Gas
studies suggest that this relationship is monotonic of nature. These models will be discussed in the following paragraph.
2.3.1 The conventional Environmental Kuznets Curve
The EKC has experienced emerging consensus and is based on the idea that while initially industrialization and agricultural modernization may lead to environmental deterioration, factors such as scale of production, output mix, input mix and state of technology (based on production efficiency and emissions specific process changes) (Stern, 2004) driven by positive income elasticity for environmental quality, education, environmental awareness and changes in environmental policy will cause an ultimate downturn (Selden & Song, 1994). Pollution patterns would thus represent market forces and government regulatory changes. This model however depends on extensive assumptions about the economy, which differ per study. The standard EKC regression model is based on emissions (E), population (P) and GDP shown in equation A:
(𝐸𝐸𝑃𝑃)𝑖𝑖𝑖𝑖 = 𝛽𝛽0+ 𝛽𝛽1(𝐺𝐺𝐺𝐺𝑃𝑃𝑃𝑃 )𝑖𝑖𝑖𝑖+ 𝛽𝛽2(𝐺𝐺𝐺𝐺𝑃𝑃𝑃𝑃 )𝑖𝑖𝑖𝑖 2+ 𝜀𝜀𝑖𝑖𝑖𝑖
A negative quadratic term captures the downturn and both random and fixed effects models are often estimated (Stern, 2004).
The first researchers to examine the EKC concept were Grossman and Krueger (1991) who studied the potential impact of NAFTA on the environment. They used a cross-country panel data set of urban air pollutions such as suspended particles (SPM), fine smoke and sulfur dioxide. They engaged in looking for the
“turning point” income thereby using a conventional second-order polynomial EKC and a cubic, N-shaped function relating emission to GDP, location dummies, trade intensity and to a time trend. Using fixed effects they found a turning point for SO2
and fine smoke around $4500 and an even lower one for SPM. Dark matter was the only pollutant that showed an N-shaped function with an income value after which pollution would infinitely increase around $12500.
Secondly Panayotou (1993) argued that higher levels of development go hand in hand with information-intensive industries and services, awareness, policy and technology improvements towards a gradual decline of environmental degradation. He used estimates of SO2 and NOX in a cross-sectional analysis that aimed to look for
the EKC turning point. He included nominal GDP, population density and a policy variable. Panayotou (1993) concluded that the turning point is lower than previous studies suggest thereby showing a flatter EKC. He assumes however that the emission factors are the same for each greenhouse gas in all countries studied. The emission factor of a greenhouse gas refers to the average emission rate for a specific source relative to activity. As sources of emission such as industry differ per country in terms of emission rate (e.g. relatively dirty industries in the Soviet Union compared to Western Europe) results of Panayotou seem less reliable.
Third Selden and Song (1994) tested the EKC for SO2, SPM, NOX and carbon
monoxide for developed countries (OECD) thereby including population density in the regression and testing for both random and fixed effects. They assumed production to be the sole source of pollution, exogenous technological change and infinitely lived agents (Stern, 2004). They found an EKC for these emissions, but the turning point found seemed substantially higher compared to previous studies, which is explained by the use of aggregate emission compared to urban emission.
Critique of these studies mainly includes the fact that most studies focus on traditional pollutants like CO, NOX and SO2, which have been substituted over time.
Furthermore data on all countries available is often used and studies seem outdated. General critique of the EKC by Stern (2004) is firstly that it is based on an extensive amount of simplified assumptions such as the absence of a reverse effect between emission and GDP. This ‘reverse effect’ refers to environmental damage decreasing economic activity to the point that the growth process of the economy is blocked. This blockage causes income levels to decline in the future. Secondly the EKC is not applicable to all pollutants even though this is claimed sometimes. Thirdly only the fixed effects model is estimated consistently compared to random effects. Fourthly the EKC relationship might be the result of the trade effect on industry distribution in terms of pollution. These critiques lead to the development of an alternative viewpoint.
2.3.2 A contemporary model: a monotonic EKC
Dasgupta, Laplante, Wang, and Wheeler (2002) have written a critical review of the conventional EKC model. They argue that the EKC shows weakness in presenting evidence for the possibility of environmental progress in developing countries. The
monotonic(cally increasing) Environmental Kuznets Curve (Dasgupta et al., 2002; De Bruyn, van den Bergh, & Opschoor, 1998) is presented that portrays a strictly positive relationship between economic development and environmental degradation. This means that incremental income harms the environment in terms of increasing levels of pollution. Two scenarios exist for a monotonic EKC: the “new toxics” view argues that traditional emissions might show the pattern of the conventional EKC, but that new pollutants that have replaced older ones do not. This implies that aggregate emission will not reduce due to replacing pollutants thereby keeping environmental damage in tact. Regarding equation A this pattern is revealed when 𝛽𝛽1 > 0 and 𝛽𝛽2 = 0 (De Bruyn et al., 1998). The second scenario focuses on the “race to the
bottom” that refers to the difficulty imposed on developing countries in terms of reducing emission when developed countries outsource their dirty businesses because of local environmental policy (Dasgupta et al., 2002). Due to the modern concept of globalization environmental policy in developed countries may become less severe thereby transforming the monotonic EKC to a flat curve over time. This (flat) curve depicts the (economically) efficient level of emission.
For the conventional and monotonic EKC a few assumptions are the same such as theoretical assumptions about the economy, technology and environmental investment (Stern, 2004). An important methodological assumption of any form of EKC is that even though emission levels per capita may differ across countries at a specific level of income, the income elasticity is kept constant in these countries at a given income level. Furthermore it is assumed that the economy is sustainable in terms of handling shocks brought about by poor environmental quality. This implies that there is no reverse effect between incremental income and environmental degradation in terms of emission.
As described earlier in this paper CO2 is a substitute for SO2 and NOX and
F-gases for CFCs, which make them according to the previous definition “new toxics”. This implies a monotonic EKC. On the other hand CH4 and N2O have not been
replaced by other emissions yet and are therefore traditional pollutants, which entails the conventional EKC. Summarizing the above a third and fourth hypothesis can be drawn:
Hypothesis 3:
The relationship between GDP per capita within emerging economies and environmental degradation is a monotonically increasing function when measured in carbon dioxide and F-gases.
Hypothesis 4:
The relationship between GDP per capita within emerging economies and environmental degradation is an increasing quadratic function when measured in methane and nitrous oxide.
According to the previous literature review it becomes clear that the impact of increasing income on environmental quality may be negative. Economic development may result in growing environmental degradation for new pollutants, but for traditional emissions there may be an eventual downturn of pollution. It is still unclear however if these impacts are significant and whether population density, country or time specific factors will drastically influence the impact of GDP on environmental quality. This relationship will be empirically tested in the following of this paper in order to answer the research question: “Does economic development impact environmental degradation in terms of emission in emerging economies and what specific characteristics in these countries could be major drivers?”
3. Methodology
3.1 Introduction
In the following paragraphs the data and methods used in this panel data study will be discussed. The aim is to assess for a set of emerging economies whether there is an impact of economic development on environmental degradation in terms of CO2,
CH4, N2O and F-gas emission and if so to examine what relationship can be found.
Whilst analyzing this relationship several major drivers that could influence the relationship between GDP per capita and emission per capita, such as population density, country effects and time effects, will individually and together be accounted for.
3.2 Data
In order to run the OLS regression three data sets are combined namely Economic Policy Indicators and Urban Development Indicators compiled by the World Bank as well as a climate dataset assembled by the World Resources Institute. All datasets contain a large number of countries but this panel data study is based on the 25 countries labeled as “emerging economies” in July 2012 by the International Monetary Fund. The used data thus includes the countries (n) Argentina, Brazil, Bulgaria, Chile, China, Colombia, Estonia, Hungary, India, Indonesia, Latvia, Lithuania, Malaysia, Mexico, Pakistan, Peru, Philippines, Poland, Romania, Russia, South Africa, Thailand, Turkey, Ukraine and Venezuela. Climate indicating data include CO2, CH4, N2O and F-gas, which are only available for time period
1991-2010. Time periods (T) are brought down to 4 by taking the average of the data used for periods 1991-1995, 1996-2000, 2001-2005 and 2006-2010 in order to show a potential development path across time. This way the number of observations in this panel data study equals 100. This allows testing for the relationship between economic development and environmental degradation for emerging economies.
3.3 Data descriptive: dependent, independent and dummy variables
The dependent variable represents the environment: emission per capita. It is split up in four different gases: CO2, CH4, N2O and F-gas. Carbon dioxide (CO2) emission
refers to “those emissions stemming from the burning of fossil fuels and the manufacture of cement including carbon dioxide produced during consumption of solid, liquid, gas fuels and gas flaring ” (World Bank, 2013a). Secondly methane (CH4) emission is defined as “those emissions stemming from human activities such
as agriculture and from industrial methane production” (World Bank, 2013c). Third nitrous oxide (N2O) emission: “those emissions from agricultural biomass burning,
industrial activities and livestock management” (World Bank, 2013d). The fourth and last emissions, fluorinated gases, include hydrofluorocarbons, perfluorocarbons and sulfur hexafluoride. According to the World Resources Institute (2013) fluorinated gases have no natural sources and only come from human-related activities. F-gas is emitted through a variety of industrial processes including aluminum and
semiconductor manufacturing (United States Environmental Protection Agency, 2013). All varieties of emission are measured in tons of carbon dioxide equivalent per capita (tCO2e per capita). Yearly emission per capita data from 1991 up until 2010 is
used and averaged out in four different time periods (T).
The two main independent variables used to assess the implications of a pollution-GDP relationship are GDP per capita and population density. Other independent variables used when applicable include GDP per capita squared and population density squared. For the measurement of economic development the yearly GDP per capita from 1991 until 2010 is used (and averaged out to get four time periods). GDP per capita is gross domestic product divided by midyear population in this dataset. The data is measured in constant 2011 international U.S. Dollars based on purchasing power parity. Within the World Bank dataset used input on GDP per capita of Argentina is completely missing. Therefore Argentina is left out of this research thereby reducing the number of observations to 96. Furthermore data on the first 2 years for Estonia is missing so the GDP per capita of Estonia for time period 1 equals the average of the years 1993, 1994 and 1995.
The variable population density measures the number of people (midyear population) per square kilometers land (World Bank, 2013e) and has not been scaled in this dataset. The definition of population refers to all residents regardless of citizenship or legal status, except for refugees who are not permanently settled in the country of asylum. They are generally considered part of the population of their country of origin. Land area is the total surface of a country, which excludes area of inland water bodies, national claims to continental shelf, and exclusive economic zones (World Bank, 2013).
The use of independent variables GDP per capita squared and population density squared depends on whether or not the greenhouse gas analyzed shows a quadratic relationship for either or both income and population density.
In order to test the validity of the hypotheses in the panel data regression analysis a fixed effects regression will be performed to control for omitted variables. These variables include exogenous factors that affect emissions thereby varying across countries without changing over time (country fixed effects). If these factors change over time but stay constant across countries they are called time fixed effects. Binary variables for country fixed effects include all 24 emerging economies analyzed in this study except for Venezuela. Furthermore the time fixed effects dummies
include period effect 1 (1991 until 1995), period effect 2 (1996 until 2000) and period effect 3 (2001-2005). Analyzing the influence of fixed effects will assist in determining whether emission per capita change significantly differs per entity or per time period.
3.4 Empirical model and procedure
In order to assess the four hypotheses stated above a multiple ordinary least square (OLS) regression is implemented. The analysis focuses on the relationship between emissions per capita, 𝑚𝑚𝑖𝑖𝑖𝑖, real GDP per capita, 𝑦𝑦𝑖𝑖𝑖𝑖, and population density, 𝑑𝑑𝑖𝑖𝑖𝑖, where i is a country index and t is a time index. A constant, 𝛽𝛽0, will be added as well as the disturbance term 𝜀𝜀𝑖𝑖𝑖𝑖, which has a zero mean and finite variance. The following empirical model is therefore applicable:
𝑚𝑚𝑖𝑖𝑖𝑖 = 𝛽𝛽0+ 𝛽𝛽1𝑦𝑦𝑖𝑖𝑖𝑖+ 𝛽𝛽𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖+ 𝜀𝜀𝑖𝑖𝑖𝑖 Model (1.0)
Emission is said to exhibit a monotonically increasing linear relationship with GDP per capita as expected in hypothesis 3 if 𝛽𝛽1 > 0 in this model. However applying model (1.0) does not disprove the possibility of another functional form for the relationship per gas. The following OLS regression models may apply:
𝑚𝑚𝑖𝑖𝑖𝑖 = 𝛽𝛽0+ 𝛽𝛽1𝑦𝑦𝑖𝑖𝑖𝑖+ 𝛽𝛽2𝑦𝑦𝑖𝑖𝑖𝑖2 + 𝛽𝛽𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖+ 𝜀𝜀𝑖𝑖𝑖𝑖 Model (2.0)
𝑚𝑚𝑖𝑖𝑖𝑖 = 𝛽𝛽0+ 𝛽𝛽1𝑦𝑦𝑖𝑖𝑖𝑖+ 𝛽𝛽𝑑𝑑1𝑑𝑑𝑖𝑖𝑖𝑖+ 𝛽𝛽𝑑𝑑2𝑑𝑑𝑖𝑖𝑖𝑖2 + 𝜀𝜀𝑖𝑖𝑖𝑖 Model (3.0)
𝑚𝑚𝑖𝑖𝑖𝑖 = 𝛽𝛽0+ 𝛽𝛽1𝑦𝑦𝑖𝑖𝑖𝑖+ 𝛽𝛽2𝑦𝑦𝑖𝑖𝑖𝑖2 + 𝛽𝛽𝑑𝑑1𝑑𝑑𝑖𝑖𝑖𝑖+ 𝛽𝛽𝑑𝑑2𝑑𝑑𝑖𝑖𝑖𝑖2 + 𝜀𝜀𝑖𝑖𝑖𝑖 Model (4.0)
For hypothesis 4 regarding an increasing quadratic function the forecast is that 𝛽𝛽1 > 0 and 𝛽𝛽2 < 0. Population density is expected to enter models with a negative sign as sparsely populated countries are likely to be less concerned about reducing per capita emissions at every level of income (Selden & Song, 1994).
In order to find the best model for each gas, models (1.0), (2.0), (3.0) and (4.0) are tested for CO2, CH4, N2O and F-gas both with and without dummies. The results
(significant) values for explanatory variables GDP per capita and population density. This can also be the case for quadratic terms of these variables and country and time dummies.
Furthermore an important aspect within the empirical research described above is that the White estimator for (heteroscedasticity-consistent) standard errors is used in analyzing the impact of economic development on environmental degradation in terms of emission in emerging markets.
4. Results
4.1 Results CO2
To determine the relationship between CO2 emission per capita and economic
development multiple OLS regressions were run (tables 1 and 2). Model (4.0) in table 2 seems to fit the data best as this model using both country and time fixed effects has the highest adjusted R-squared value (0.9742) compared to other tested models. Model (4.0) using both time and country fixed effects provides some evidence for an increasing quadratic function, the conventional EKC. This is explained by the fact that the coefficient of GDP per capita (𝛽𝛽1) is positive (.0003679), the coefficient of (GDP per capita)2 (𝛽𝛽2) is negative (-7.85e-09) and both coefficients are significant at the 5% level. It can thus be concluded that this CO2 dataset does not support
hypothesis 3, as there is no empirical evidence for a monotonic EKC. The first possible explanation for the unexpected conventional EKC is a transition from polluting industrial economies to clean service ones for emerging markets over time. Secondly, as stated before, people with greater income prefer higher environmental quality compared to lower incomes. It should be noted though that it is outside the scope of this research to test for a possible cubic function and therefore conclusions about the conventional EKC should be drawn carefully for all tested greenhouse gases.
An increasing quadratic relationship also seems applicable for population density as both population density and (population density)2 are significant at the 1% level with 𝛽𝛽𝑑𝑑1> 0 (.1205873) and 𝛽𝛽𝑑𝑑2< 0 (-.0001743). This is unexpected because
graph 1 suggests a decreasing monotonic relationship between CO2 emission per
capita and population density. Furthermore literature expects population density to enter the model with a negative coefficient (Selden & Song, 1994). A decreasing trend can still be observed depending on the value of population density analyzed since there will be a point where the function turns downwards. Sparsely populated countries thus seem likely to be less concerned about per capita emission after the turning point, but not when we analyze a location before the turning point (increasing trend). Due to this initial increasing trend there is insufficient empirical evidence for hypothesis 1. The increasing trend of the inverted U-shaped relationship could be explained by traffic congestion, which increases as population density raises thereby boosting CO2 emission. The eventual downturn however could be justified by the
health and monetary arguments of Selden and Song (1994). Conclusions concerning an inverted U-shaped relationship should be carefully constructed as no cubic function is tested in this research.
Time fixed effects appear to improve the model as the adjusted R-squared of model (4.0) using country fixed effects only (0.9635) is lower compared to model (4.0) using both time and country fixed effects (0.9742). The only significant time fixed effects dummy is ‘period effect 1’ at the 1% level with a coefficient of 1.429562, but any conclusions about the importance of time period 1 are futile within this research due to a lack of theoretical evidence.
When country fixed effects are added to model (4.0) with time fixed effects most of the countries in the sample deviate from the general trend shown by the significant country dummies (table 9). Country fixed effects seem to improve model (4.0) since the adjusted R-squared using both time and country dummies (0.9742) is much higher compared to model (4.0) using time specific effects only (0.2407). Emission thus appears to be strongly determined by country specific factors, which was expected. Therefore there is some evidence for hypothesis 2 as country fixed effect seem to be a determinant of the relationship between economic development and environmental pollution in terms of CO2.
Overall there seems evidence for an increasing quadratic CO2-GDP
relationship when the fixed effects model is tested thereby not supporting hypothesis 3. Based on the empirical model tested economic development is likely to impact environmental degradation in terms of CO2.
4.2 Results CH4
The relationship between emission of CH4 and economic development has been
empirically tested and shows some evidence for an increasing quadratic function, the conventional EKC. Table 3 and 4 show that model (4.0) using both time and country fixed effects fits this dataset on CH4 best as it has the highest adjusted R-squared
value of all tested models (0.9783) and much explanatory power. Here 𝛽𝛽1 > 0 (.0000551) and 𝛽𝛽2 < 0 (-1.11e-09) at the 5% and 10% significance level respectively. Therefore there appears some evidence in favor of hypothesis 4 for methane. As a cubic function is not tested so conclusion should be drawn carefully.
As for CO2 there is some evidence for an inverted U-shaped
relationship between emission and population density as 𝛽𝛽𝑑𝑑1 > 0 (.0135231) and 𝛽𝛽𝑑𝑑2< 0 (-.0000181) both at the 1% significance level. Graph 2 shows that this is
unexpected due to a suggested decreasing monotonic relationship between CH4
emission and population density. Furthermore at least a negative coefficient for 𝛽𝛽𝑑𝑑1 is expected by the literature so there is not enough evidence for the validity of hypothesis 1. A decreasing trend can be observed because the significance of the negative 𝛽𝛽𝑑𝑑2 creates a turning point, but sparsely populated countries do not seem to be less concerned about emission at every level of income. Emission levels of denser populated countries are higher compared to those of countries that are less densely populated before the turning point. Possible explanations for the eventual downturn include health and monetary arguments (Selden & Song, 1994) as well as the theory that increasing population density crowds out agricultural activities thereby reducing methane emission. There is a lack of theoretical evidence however for the initial increase in emission before the turning point so conclusions are futile. As for GDP, conclusions regarding an inverted U-shaped relationship should be carefully constructed as no cubic function has been tested.
Time fixed effects seem to contribute to model (4.0) as the adjusted R-squared of model (4.0) with country fixed effects only (0.9691) is lower than this model with both country and time fixed effects (0.9783). The significant time dummy variables are period 1 (.3215552) and 2 (.1542721) at the 1% and 10% level respectively. Any
conclusions about the relevance of these time periods are insufficient within this research due to a lack of theoretical evidence.
Equivalent to CO2 country fixed effects improve model (4.0) and thus have an
impact on the relationship between economic development and environmental pollution in terms of CH4. This can be explained by the adjusted R-squared values for
model (4.0), which is lower using time fixed effects only (0.3127) compared to model (4.0) using both time and country fixed effects (0.9783). There is a substantial difference between these adjusted R-squared values, indicating a large improvement of the explanatory power of the model due to implementation of country fixed effects. The significant country dummies can be found in table 10. Therefore there is some empirical evidence that validates hypothesis 2.
Concluding there appears evidence for an increasing quadratic relationship between methane and GDP when testing fixed effects model (4.0) thereby supporting hypothesis 4. Based on the empirical model tested economic development is thus likely to influence environmental degradation in terms of CH4.
4.3. Results N2O
The results regarding the relationship between N2O emission and GDP are fairly the
same compared to those of CH2, the other traditional pollutant. This means that
empirical tests for nitrous oxide show some evidence for an increasing quadratic function, once again the conventional EKC. Comparing adjusted R-squared values displayed in tables 5 and 6 shows that model (4.0) using both time and country fixed effects fits this data best because the adjusted R-squared value of model (4.0) (0.9402) is highest compared to all others and therefore suggests that this model has most explanatory power for the N2O dataset. The inverted U-shaped relationship is shown
by 𝛽𝛽1 > 0 (.0000533) and 𝛽𝛽2 < 0 (-1.25e-09) at the 5% and 10% significance level respectively. For nitrous oxide hypothesis 4 thus seems validated by the data. No cubic function is tested so conclusions should be drawn conscientiously.
Nitrous oxide and population density are expected to show a decreasing monotonic relationship, as described in hypothesis 1 and shown in graph 3. There appears some evidence however for an increasing quadratic relationship between N2O emission and population density because 𝛽𝛽𝑑𝑑1> 0 (.0081507) and
𝛽𝛽𝑑𝑑2< 0 (-9.94e-06) at the 1% and 5% significance level respectively. Following the
same reasoning as described for methane, there seems insufficient evidence that supports hypothesis 1 for N2O. Explanations for the downturn are equivalent to those
for methane and there is not enough theoretical evidence for disclosing the initial increasing trend. No cubic function has been tested so conclusions regarding an inverted U-shaped relationship should be carefully constructed.
Model (4.0) appears to be improved by time fixed effects as the adjusted R-squared of this model using country fixed effects only (0.9088) is lower compared to model (4.0) using both time and country fixed effects (0.9402). Time fixed effects for period 1 (.2548968), period 2 (.1046549) and period 3 (.0553548) are all significant at the 1%, 5% and 10% level respectively. A possible explanation for this could be that since the 1980s synthetic fertilizers have been used more extensively compared to earlier years in emerging markets as it became financially feasible and agricultural business grew in the former Soviet Union after the Iron Curtain came down in 1991.
As for methane, country fixed effects seem to improve model (4.0) and thus impact the relationship between environmental degradation and economic development measured in nitrous oxide. Empirical evidence can explain this because the adjusted R-squared for model (4.0) using time fixed effects only (0.2916) is substantially lower compared to the model using both time and country fixed effects (0.9402). Therefore, regarding N2O, there seems some evidence for the validity of
hypothesis 2. The significant country dummies can be found in table 11.
Chiefly there seems evidence for an increasing quadratic relationship between nitrous oxide and GDP when examining model (4.0) using both time and country fixed effects thereby validating hypothesis 4. Based on the empirical models tested economic development is likely to impact environmental pollution in terms of N2O.
4.4 Results F-gas
For F-gas multiple regressions were run to determine the impact of economic development on F-gas emission (tables 7 and 8). Model (3.0) using country fixed effects only in table 8 displays the highest adjusted R-squared value (0.8170) of the tested models and thus seems to fit this dataset best. It should however be noted that model (1.0) using country dummies has an adjusted R-squared value of 0.8167, which
is only slightly lower compared to model (3.0). Overall model (3.0) provides some evidence for an increasing monotonic function, the monotonic EKC. A monotonic EKC might be applicable as F-gas is recognized as a “new toxic” and has not been replaced yet. This is supported by the fact that the coefficient 𝛽𝛽1 is positive (8.08e-06) and significant at the 1% level. It can thus be concluded that this F-gas dataset supports hypothesis 3.
Model (3.0) includes the variables population density and its square. However, these variables are not significant for F-gas when using country fixed effects. This is unexpected as graph 4 and the literature suggest a monotonic decreasing relationship between F-gas emission and population density. Therefore there seems insufficient evidence for the validity of hypothesis 1 regarding F-gas, as there does not appear to be a relationship between F-gas emission and population density.
Model (3.0) with an adjusted R-squared of 0.8170 does not include the application of time specific effects. Therefore it can be concluded that these effects do not seem to improve the model. It should be noted however that not including time fixed effects could lead to biased results.
When country fixed effects are added to model (3.0) without fixed effects the adjusted R-squared increases from 0.3247 to 0.8170. All countries in the sample except for Mexico, South Africa and Venezuela deviate from the general trend shown by the significant country dummies in table 12. F-gas emission thus seems to be impacted by country specific factors, which was expected. Therefore there appears some evidence that supports hypothesis 2: country effects seem to determine the relationship between environmental degradation in terms of F-gas and economic development.
Overall there seems some evidence for a monotonic EKC when testing model (3.0) including population density, its square and country fixed effects thereby validating hypothesis 3. Therefore based on the empirical model tested it can be concluded that economic development likely impacts environmental pollution in terms of F-gas.
5. Conclusion
In most parts of the world the emission of greenhouse gases causes serious problems in terms of global warming and human health. Emerging economies have shown rapid economic development over the past decades, which can create negative externalities in terms of air pollution as growing industries cause more polluting waste. Abatement costs can be high and vary per greenhouse gas. Therefore economic development may seriously hamper sustainability. Increasing environmental awareness should thus be an important aim for countries as policy measures and technological incentives can minimize environmental degradation.
This study assessed the relationship between economic development and environmental degradation in terms of emission by analyzing World Bank and World Resources Institute data for a set of emerging markets. The data analyses for carbon dioxide, methane and nitrous oxide suggest that the relationship between GDP per capita and environmental degradation within emerging markets follows a conventional EKC. Population density in these analyses has a significant effect and could impact the relationship between economic development and environmental quality. Empirically testing F-gas emission shows no evidence for a conventional EKC, but for an increasing monotonic relationship, a monotonic EKC. This could be explained by the fact that F-gas is recognized as a “new toxic” and has not been replaced yet. Population density seems to have no significant effect on F-gas emission in this research.
Furthermore several other factors besides from population density may influence the relationship between environmental quality and economic development. The data suggests that country fixed effects are a determinant of the impact of economic development on environmental degradation in terms of emission for all tested greenhouse gasses. This could be explained by the strong dependence of the environment on local human action, which is determined by local culture and geography. The effect of time seems overall less important, however for nitrous oxide there is some evidence that inflated implementation of synthetic fertilizers and the fall of the Iron Curtain gradually increased N2O emissions thereby negatively affecting
environmental quality.
The importance of the influence of population density has been stressed throughout this research. It prevails in determining emission for carbon dioxide,
methane and nitrous oxide. Unexpected results appear for these gases as population density and emission show an inverted U-shaped relationship. Regarding F-gas there seems to be little proof for a relationship between emission and population density. Furthermore this research has not indicated that sparsely populated countries are likely to be less concerned about reducing per capita emissions (Selden & Song, 1994).
As noted in the previous paragraph the empirical research has shown some unexpected results compared to the literature review. The research for CO2, CH4 and
N2O provides some evidence for the existence of an increasing quadratic relationship
between emission and population density and for F-gas there seems hardly any evidence for a relationship, which differs from theoretical arguments. Furthermore the empirical research for CO2 shows little evidence for the theoretically expected
monotonic increasing relationship. Therefore when drawing conclusions the following concerns have to be taken into consideration.
This empirical research consists of 96 observations in terms of emission, GDP per capita and population density. This number is relatively low and therefore causes results to be less accurate. It is however difficult to use more carbon dioxide, methane, nitrous oxide and F-gas data as this has only been collected for the past 20 years.
The general econometric model for the Environmental Kuznets Curve includes only two explanatory variables namely GDP per capita and population (density). This is a relatively low number as the literature suggests that there are several country and time fixed factors that influence the emission volume. These factors however are absorbed into the fixed effects model. A more extensive research could give different and more detailed results as the empirical research in this paper suggests that both country and time fixed effects influence the studied relationship between emission and economic development.
The use of GDP per capita has the ability to generate inaccurate results as GDP is mostly based on the narrow group of well educated people who earn a relatively large part of GDP compared to those living in rural areas. This is especially relevant for the empirical study done for emerging economies as developing countries have a higher income gap between urban and rural areas compared to the developed world. Furthermore both the World Bank (2013b) and the International Monetary Fund (2012) have previously stated that they are having problems assembling
accurate GDP data due to differentiating data gathering methods and the influence of undocumented business. This causes a discrepancy between data of these two institutions and shows the unreliable nature of GDP data.
Using population density as a measure can also cause inaccuracies within the empirical research as emerging economies often have a large rural society. In these areas documentation of the population is less effective. This could result in inaccurate data thereby influencing the outcome of the research.
Another key variable that can yield wrong results is emission. The gathering of environmental data is still under development as scientists fail to accurately determine values. This results in the noisy nature of emission data (Selden & Song, 1994). Moreover a large amount of countries do not gather data at all. Science possibly needs another decade before accurate estimations can be made based on emission data.
This empirical study focuses on emerging markets in total, which includes 25 different countries. The literature and empirical research only focus on the overall result for emerging markets whilst it could prove interesting to look at different countries individually since they show different values and sometimes also different (emission) patterns. This was however outside of the scope of this research.
The last concern is that this empirical research merely looks at emerging markets. Looking at global aggregates could yield a different conclusion as the literature has shown a discrepancy between the pattern of global emissions and carbon dioxide, methane and nitrous oxide emissions for emerging markets. Therefore it could be interesting to look at global emission too in order to see if the nature of the relationship between economic development and environmental degradation in terms of emission differs.
By looking at the relationship between emission and GDP per capita for four different pollutants it can be concluded that there is some evidence for the influence of economic development on environmental degradation. However there does not seem to be a general pattern for all greenhouse gases studied. Analyses of CO2, CH4
and N2O show a conventional EKC while accounting for population density, its
square and fixed effects. Analysis of F-gas shows some evidence for a monotonic EKC when population density, its square and country fixed effects are included. Specific characteristics that could be major drivers in these countries include population density, country fixed effects and time fixed effects, as these influences
seem to impact the relationship between economic development and emission intensity. However conclusions should be drawn carefully as the above mentioned concerns show that the discrepancy between the literature and empirical study could be misleading. More research should be done in order to check the validity of the concerns mentioned above and potentially improve the research.
