THE IMPACT OF ECONOMIC GROWTH AND TRADE ON THE ENVIRONMENT: THE CANADIAN CASE

Transcription

1 THE IMPACT OF ECONOMIC GROWTH AND TRADE ON THE ENVIRONMENT: THE CANADIAN CASE OLAIDE KAYODE EMMANUEL STUDENT #: Memorial University Of Newfoundland, Canada Environmental Economics (Econ6022) Project, April, ABSTRACT There is often the presumption that economic growth and trade liberalization are good for the environment. However, the risk is that policy reforms designed to promote growth and trade liberalization may be encouraged with little consideration of the environmental consequences (Arrow et al., 1995). All economic activity occurs in the natural, physical world. It requires resources such as energy, materials and land. Also, economic activity invariably generates material residuals, which enter the environment as waste or polluting emissions. The Earth, being a finite planet, has a limited capability to supply resources and to absorb pollution. Hence, There is growing recognition that gross domestic product (GDP) produced at the expense of the global environment, and at the expense of scarce and finite physical resources, overstates the net contribution of that economic growth to a country s prosperity. Using mainly four environmental indicators of air pollution (greenhouse gas, carbon dioxide, nitrogen oxide and Sulphur oxide), this paper provides an empirical analysis of the verification of the EKC within the Canadian context, and the causal relationship between economic growth and trade and environmental degradation from the Canadian perspective. The study reveals that the traditional EKC does not hold for Canada, for these indicators of environmental degradation. It also, shows that a long run relationship and a bidirectional Granger-causality exist between economic growth and trade, and these indicators. KEY WORDS: Economic growth, Trade, EKC, ARDL, VECM, Environmental degradation I. INTRODUCTION There is often the presumption that economic growth and trade liberalization are good for the environment. However, the risk is that policy reforms designed to promote growth and trade liberalization may be encouraged with little consideration of the environmental consequences (Arrow et al., 1995). All economic activity occurs in the natural, physical world. It requires resources such as energy, materials and land. Also, economic activity invariably generates material residuals, which enter the environment as waste or polluting emissions. The Earth, being a finite planet, has a limited capability to supply resources and to absorb pollution. Hence, There is growing recognition that gross domestic product (GDP) produced at the expense of the global environment, and at the expense of scarce and finite physical resources, overstates the net contribution of that economic growth to a country s prosperity. 1

2 The early stages of the environmental movement began with some scientists questioning how natural resource availability could be compatible with sustained economic growth (Meadows et al., 1972). Neoclassical economists, however, fiercely defended that limits to growth due to resource constraints were not a problem (e.g. Beckerman, 1974). This marks the beginning of the debate between the so-called environmental pessimists and optimists. This debate which still continues till date is centered on nonrenewable resource availability. However, in the 1980s large issues such as ozone layer depletion, global warming and biodiversity loss began to change the main focus of the debate to the impacts of environmental degradation on economic growth. Hence, interest was shifting away from natural resource availability towards the environment as a medium for assimilating wastes (i.e. from source to sink ) (Neumayer, 2003). Also, following the Brundtland Report (WCED, 1987), the discourse of sustainable development largely embraced the economic growth logic as a way out of poverty, social depravation and also environmental degradation particularly for the developing world. As such, the relationship between economic growth and trade liberalization, and the environment came under increased scrutiny. The empirical literature on the link between economic growth and environmental pollution literally exploded in the 1990s (Stern, 2003; 2004). Much of this literature sought to test the Environmental Kuznets Curve (EKC) hypothesis. The environmental Kuznets curve is a hypothesized relationship between various indicators of environmental degradation and income per capita (Stern, 2003). The EKC posits that in the early stages of economic development environmental degradation and pollution will increase until a certain level of income is reached (called the turning point) and then environmental improvement will occur. Several possible reasons for this hypothesized relationship has been put forward: the income elasticity of demand for environmental quality may exceed one, so that as income rises citizens support initiatives and policy propositions to reduce environmental degradation; rising incomes may be associated with shifts from resource-intensive to research-intensive output in the economy; and rising incomes together with improvements in technology and human capital may help decouple economic growth and environmental degradation. The turning points vary for different indicators of environmental degradation. This relationship between per capita income and pollution is often shown as an inverted U-shaped curve. Hence, the curve is named after Simon Kuznets (1955) who hypothesized that as per capita income increases, economic inequality increases over time and then after some turning point starts declining. In the early 1990s the EKC was introduced and popularized with the publication of Grossman and Krueger s (1991) work on the potential environmental impacts of a North American Free Trade Agreement (NAFTA), and the 1992 World Bank Report (Shafik and Bandyopadhyay, 1992; World Bank, 1992). These studies have, however, being criticized for a variety of reasons (Stern, 2003, 2004). One of these reasons is that, most of the empirical studies concentrated on few pollutants, and this may lead to the incorrect interpretation that all other pollutants have the same relation to income. Also, the relationship between the environment and income growth might vary with the source of income growth, since different types of economic activities have different pollution intensities. One 2

3 implication of this is that the pollution consequences of economic growth are dependent on the underlying source of growth (Antweiler et al. 2001). It was also demonstrated that methodological choices can significantly influence the results (Cavlovic et al. 2000). It has also been argued by several researchers that the simplest form of the EKC does not account for trade patterns (Suri and Chapman, 1998; Antweiler et al., 2001; Cole, 2004). Specifically, they indicate that trade patterns may partially explain a reduction in pollution in high income countries, but with the reverse in low income countries. Hence, trade as a driver of economic growth has also been seen to impact greatly on the environment. In particular, two major hypotheses have been used in the literature, to link trade with the environment; these are the Pollution Haven Hypothesis (PHH) and the Factor Endowment Hypothesis (FEH). The PHH argues that less stringent environmental regulations in developing countries will provide them with a comparative advantage in the production of pollution-intensive goods over developed countries (Cole, 2004). The FEH on the other hand, argues that factor abundance and technology determine trade and specialization patterns, and that such countries relatively abundant in factors used intensively in polluting industries will on average get dirtier as trade liberalizes and vice versa (Mani and Wheeler, 1998). Using mainly four environmental indicators of air pollution (greenhouse gas, carbon dioxide, nitrogen oxide and Sulphur oxide), this paper provides an empirical analysis of the verification of the EKC within the Canadian context, and the causal relationship between economic growth and trade and environmental degradation from the Canadian perspective. The rest of this paper is structured as follow: section II briefly discusses the Canadian environmental policies and progress experience; section III looks at the review of literatures on the economic growth and trade, and environment nexus; section IV focuses on the description of the data and the methodology used; section V presents the empirical analysis of the results; section VI provides conclusions about the study and its implications. II. THE CANADIAN ENVIRONMENTAL POLICIES AND PROGRESS Canada has over the past few years taken some actions, enact and implement some policies aiming at improving the quality of the environment, not only within its geographical jurisdiction, but also globally. Canada ratified the United Nations Framework Convention on Climate Change (UNFCCC) in December 1992, under which it committed stabilizing, greenhouse gas (GHG) emissions at 1990 levels by 2000; and the Convention came into force in March However, in 2000, Canada s absolute GHG emissions were 22% higher than they had been 10 years earlier (Environment Canada, 2014). Canada went to ratify the Kyoto Protocol in 2002, pledging to reduce GHG emissions to 6% below 1990 levels between 2008 and As of 2010, however, absolute GHG emissions remained 17% above 1990 levels. At the 15th session of the Conference of the Parties (COP15) to the UNFCCC in 2009, Canada signed the Copenhagen Accord, under which Canada has committed to reducing its GHG emissions to 17% below the 2005 level by the year Canada s National Inventory is prepared and submitted annually to the UNFCCC by April 15 of each year, in accordance with the December 2005 version of the 3

4 Guidelines for the preparation of national communications by Parties included in Annex I to the Convention, Part I: UNFCCC reporting guidelines for national inventories (Environment Canada 2014). Canada has also ratified the United Nations Convention on Long-range Transboundary Air Pollution (LRTAP Convention), which has three international protocols to reduce Sulphur dioxide emissions. The first, the 1985 Sulphur Protocol, adopted a flat rate target to reduce national annual Sulphur emissions by at least 30 percent between 1980 and The next two, the 1994 Sulphur Protocol and the 1999 Gothenburg Protocol to abate Acidification, Eutrophication and Ground-level Ozone, are based on effects. They aim to reduce Sulphur emissions where environmental effects are more severe. The 1985 Sulphur Protocol called for Canada to cap national Sulphur dioxide emissions permanently at 3.2 million tonnes by Canada met this cap in 1992, with national emissions of 3.1 million tonnes. The 1994 Sulphur Protocol allows emission reductions to be geographically targeted to achieve maximum environmental benefit. Canada has met all its current protocol commitments, including capping Sulphur dioxide emissions in the Sulphur oxide management area at 1.75 million tonnes by 2000 (Environment Canada). The 1988 Canadian Environmental Protection Act (revised in 1999 and proclaimed in 2000) has played an important role in improving waste disposal, and the 1998 Environmental Harmonization Accord set national standards for important air and water pollutants. A 1995 report by the Organization for Economic Cooperation and Development (OECD) provides the first snapshot of Canadian environmental performance in terms of ecosystems, water, waste, air and public policy (OECD 1995). Hayward and Jones (1998) and Devlin and Grafton (1999) provide overviews or syntheses of environmental trends over the two decades prior to Hayward and Jones used 20 separate measures of environmental degradation in the categories of air quality, water quality, natural resources and solid waste and conclude that overall environmental quality improved between 1980 and Devlin and Grafton conclude that in a number of significant areas, particularly air quality, Canada has improved its environmental quality but important challenges remain. The Conference Board of Canada, a body with an overarching goal of measuring the quality of life for Canada and its OECD s peer countries, provides an overview of performance and relative ranking of the environmental performance of Canada and its peer countries in 2014.The Board used fourteen indicators to assess environmental performance across six dimensions. The indicators used include, Greenhouse gas (GHG) emissions, Nitrogen oxides emissions, Sulphur oxides emissions, VOC emissions, PM10 concentration, Municipal waste generation, Water Quality Index, Water withdrawals, Threatened species, Forest cover change, Use of forest resources, Marine Trophic Index, Low emitting electricity production, and Energy intensity; while the dimensions used include, air quality, waste, water quality and quantity, biodiversity and conservation natural resource management, climate change and energy efficiency. According to the Board, Canada receives a C grade on environmental performance and ranks 15th out of 17 peer countries. Compared with the 17 country average, Canada s performance is above average on five indicators: use of forest resources low emitting electricity production, Water Quality Index, threatened species, particulate matter concentration. Canada s performance is below 4

5 average on nine indicators: forest cover change nitrogen oxides emissions Sulphur oxides emissions Marine Trophic Index greenhouse gas (GHG) emissions, water withdrawals volatile organic compound (VOC) emissions Canada s success in improving its environmental performance has been mixed. It has improved air quality, reduced its energy intensity, and increased the growth of forest resources relative to forest harvest, municipal waste generation and energy intensity. Although Canada ranks below the best performing Country on all of the environmental indicators, it does earn A grades for four indicators: Water Quality Index, threatened species, use of forest resources, and low emitting electricity production. It earns a B for Sulphur oxide emissions, forest cover, and PM10 concentration; and receives a C grade on water withdrawals. Canada receives D grades on six indicators: nitrogen oxides emissions, VOC emissions, municipal waste generation, Marine Trophic Index, GHG emissions, and energy intensity But Canada must do more to lower greenhouse gas emissions, to use its freshwater resources more wisely, and to reduce waste. Canada ranks 15th out of 17 countries for greenhouse gas (GHG) emissions per capita. Canada s per capita GHG emissions decreased by nearly 5 per cent between 1990 and 2010, while total GHG emissions in Canada grew 17 per cent. The largest contributor to Canada s GHG emissions is the energy sector, which includes power generation (heat and electricity), transportation, and fugitive sources. Canada ranks 16th out of 17 peer countries. Canada s per capita nitrogen oxides emissions have been decreasing since Canada needs to do more to reduce emissions from the transportation, electricity, and industrial sectors. Canada ranks 16th out of 17 countries. Canada s per capita Sulphur oxides emissions are nearly 17 times that of the best performer, Switzerland. Canada s Sulphur oxides emissions decreased between 1990 and The board maintained that, to improve its overall performance, Canada must promote economic growth without further degrading the environment, partly by encouraging more sustainable consumption. III. LITERATURE REVIEW ON ECONOMIC GROWTH, TRADE AND ENVIRONMENT NEXUS Examination of the empirical relationship between national income and measures of environmental quality began with Grossman and Krueger s (1991) paper on the potential impacts of NAFTA. There, they estimated reduced-form regression models relating three indicators of urban air pollution to characteristics of the site and city where pollution was being monitored and to the national income of the country in which the city was located. This was when the EKC concept emerged. However, a 1992 World Bank Development Report (Shafik and Bandyopadhyay, 1992; World Bank, 1992) made the notion of the EKC popular by suggesting that environmental degradation can be slowed by policies that protect the environment and promote economic development. Panayotou (1993), who actually coined the curve EKC, suggested the addition of other explanatory variables, and estimated the EKC for 55 developed and developing countries using a panel data regression. Selden and Song (1994) estimated the EKCs for four emission series using longitudinal data from World Resources (WRI, 1991). The data used are primarily for developed countries, and the study showed that the turning point for 5

6 emissions was likely to be higher than that for ambient concentrations. The reason for this being that urban and industrial development tends to be more concentrated in a smaller number of cities, at the early stage of economic development, and these cities will also have rising central population densities. The trend will be reversed in the later stages of development; hence, it is possible for peak ambient pollution concentration to fall as income rises even if total national emissions are rising (Stern et al., 1996). Grossman and Krueger (1995) identify three different channels through which economic growth can affect the quality of the environment that shape the EKC: the scale effect, the increase in pollution when the economy grows, the composition, and the technique effect. The composition effect in this context refers to structural changes that occur in the economy, leading to different environmental pressures in the long-term. Furthermore it is assumed that the dominant role is played by public pressure towards more governmental regulation and the use of cleaner production techniques by firms (technique effect). This is based on the assumption that, as income grows income elasticity of the environmental quality increases. Therefore, after a threshold level of income, wealthier countries tend to be more willing and able to channel resources into environmental protection and higher environmental standards. Cole et al. (1997) also confirm the existence of the EKC, using emissions data for the OECD countries. List and Gallet (1999) estimated the EKC for the fifty U.S. states using emissions data for the period 1929 to Neumayer (2003) posited that rich countries may be better able to meet the higher demands for environmental protection through their institutional environmental capacity. However, Martinez-Alier (1995), contested whether rich people care more about the environment than the poor. Beckerman (1992) claimed that there is clear evidence that, although economic growth usually leads to environmental degradation in the early stages of the process, in the end the best and probably the only way to attain a decent environment in most countries is to become rich. The notion of the EKC has however, been challenged for some reasons: first, the EKC has never been shown to apply to all pollutants or environmental impacts alike (Dasgupta et al., 2002; Perman and Stern, 2003); second is the econometric critic of the reduced form model (Day and Grafton, 2001; Stern, 2003). Delvin and Grafton (1999) showed that there is considerable evidence of environmental degradation in a number of critical areas, such as species and habitat loss and depletion of natural resources in Canada, despites its wealth. Day and Grafton (2001) used the cubic specification of the reduced form model, and found out that the EKC does not hold for carbon dioxide emissions; concentrations of carbon monoxide, nitrogen dioxide, ground level ozone, sulphur dioxide, total particulate matter; concentrations of dioxin in herring gull eggs in the St. Lawrence river; concentrations of fecal coliform in the Saskatchewan River; and concentrations of dissolved oxygen in the Saskatchewan and St. John s Rivers. Arrow et al. (1995) and Stern et al. (1996) argued that if there was an EKC type relationship, it might be partly or largely a result of the effects of trade on the distribution of polluting industries. Suri and Chapman (1998) introduced trade explicitly into the EKC model, and consequently, numerous studies have examined the relationship between trade and environment in the last few years. However, the empirical results reported from these studies appear to be 6

7 mixed; the study by Antweiler et al. (2001) shows that trade liberalization reduces pollution, the findings by Dasgupta et al. (2002) appear to be skeptical about the positive environmental effects of trade liberalization. Furthermore, some studies (Grossman and Krueger, 1993; Gale and Mendez, 1998) find empirical support in favor of the factor endowment hypothesis (FEH) and against a significant influence of environmental regulation on trade patterns, while some others find evidence in support of the PHH (Suri and Chapman, 1998; Mani and Wheeler, 1998). IV. METHODOLOGY The reduced-form specification that is commonly employed in the empirical literature to examine the relationship between environmental degradation and per capita income in the context of the EKC is given as: E it = Y it + 2 Y 2 it + 3 Y 3 it + 4 Z it + ε it (1) Where, E it represents environmental degradation, i.e. the specific pollutant that is used for the estimation, Y it is income per capita, and Z it are other covariates, for example population density, population growth, or income inequality. Trade has occasionally been included as an additional covariate in the EKC model. The basic EKC models start from a simple reduced-form quadratic function, whereas most recent studies include the cubic level. The inverted-u shaped curve derived from such a formula requires to be positive 1 and to be 2 negative, and both statistically significant. However, empirical results could display several other variants different from the EKC; if 1 is negative and statistically significant but 2 and 3 are statistically insignificant, then we get pattern downward sloping straight line. These are indicators that show an unambiguous improvement with rising per capita income, such as access to clean water and adequate sanitation. If 1 is positive and statistically significant, but 2 and 3 are statistically insignificant, then we get pattern an upward sloping straight line. These are indicators that show an unambiguous deterioration as incomes increase. If 1 and 2 are statistically significant, but 2 is positive and 3 is statistically insignificant, then a U shaped curve results. There is also the possibility of a second turning point in which case, 3 is statistically significant, and multiple turning points could also result, which may not fit the model. The model could also be estimated in its logarithmic form; this imposes a non-negativity constraint on the values of the variables. As noted by Stern (2003), economic activity inevitably implies the use of resources and by the laws of thermodynamics, use of resources inevitably implies the production of waste. Regressions that allow levels of indicators to become zero or negative are inappropriate except in the case of deforestation where afforestation can occur. This paper estimates the model in a time series form and makes use of both the quadratic and the cubic specifications; both are estimated in both the level and logarithmic forms. The paper estimates the co-integration relationship between economic growth, trade and the environmental degradation indicators using the Autoregressive Distributed Lagged (ARDL) 7

8 model. In Pesaran et al (2001), the co-integration approach, also known as the bounds testing method, is used to test the existence of a co-integrated relationship among variables. The procedure involves investigating the existence of a long-run relationship in the form of an unrestricted error correction model for each variable as follows: Y t = µ 1 + n s=0 γ 1,s Y t s + n s=0 γ 2,s E t s + (2) E t = µ 2 + n s=0 θ 1,s Y t s + n s=0 θ 2,s E t s + (3) n s=0 n s=0 γ 3,s Z t s +δ 1 Y t 1 + δ 2 E t 1 + δ 3 Z t 1 +η 1,t θ 3,s Z t s + π 1 Y t 1 +π 2 E t 1 +π 3 Z t 1 +η 2,t Z t = µ 3 + n s=0 φ 1,s Y t s + n s=0 φ 2,s E t s + +φ 1 Y t 1 +φ 2 E t 1 +φ 3 Z t 1 +η 3,t (4) n s=0 φ 3,s Z t s The F-tests are used to test the existence of long-run relationships. The F-test used for this procedure, however, has a nonstandard distribution. Thus, the Pesaran et al (2001) approach computes two sets of critical values for a given significance level. One set assumes that all variables are I (0) and the other set assumes they are all I (1). If the computed F-statistic exceeds the upper critical bounds value, then the null hypothesis (no co-integration) is rejected. If the F- statistic falls within the bounds set, then the test becomes inconclusive. If the F-statistic falls below the lower critical bound value, it implies no co-integration. When a long-run relationship exists, the F-test indicates which variable should be normalized. The causal relationship between the variables is determined using Vector Error Correction (VEC) model. Following Granger (1988), and Engle and Granger (1987), the VEC representation is as follow: p Y t = µ 1 + r k=1 α 1,k ν k,t p + s=1 γ 1,s Y t s + s=1 γ 2,s E t s + (5) E t = µ 2 + r k=1 α 2,k ν k,t p + s=1 θ 1,s Y t s + s=1 θ 2,s E t s + (6) p p p p s=1 p s=1 γ 3,s Z t s +η 1,t θ 3,s Z t s +η 2,t p Z t = µ 3 + r k=1 α 3,k ν k,t p + s=1 φ 1,s Y t s + s=1 φ 2,s E t s + s=1 φ 3,s Z t s +η 3,t (7) Where Y t is GDP per capita, E t is the environmental degradation indicator, Z t is trade, p is lag length and is decided according to information criterion and final prediction error. The parameters ν k,t p are the co-integrating vectors, derived from the long-run co-integrating relationships regression, and their coefficients α i,k are the adjustment coefficients. The parameters μi, (i=1, 2, 3, 4, 5) are intercepts and the symbol Δ denotes the difference of the variable following it. Using the model in Equations (5 7), Granger causality tests between the variables can be investigated through the following three channels: p p 8

9 (i).the statistical significance of the lagged error-correction terms (ECTs) by applying separate t- tests on the adjustment coefficients. This shows the existence of a long-run relationship (ii). A joint F-test or a Wald χ2-test applied to the coefficients of each explanatory variable in one equation. For example, to test whether environmental indicator Granger-causes growth in Eq. (5), we test the following null hypothesis:h 0: γ 2,1 =γ 2,2 = =γ 2,p 0. This is a measure of short-run causality; (iii). A joint F-test or a Wald χ2-test applied jointly to the terms in (i) and the terms in (ii) V. EMPIRICAL RESULTS AND DISCUSSION V. 1. Data and Variable definition Four environmental degradation indicators are made use of in this empirical work, each of them in per capita (kiloton per capital) and in total (kiloton) concentrations. The indicators include, Nitrogen oxides emissions, sulphur oxides emissions, greenhouse gases emissions (excluding land use and land use charge measure and forestry- LULUCF) emissions, and carbon dioxide emissions; the carbon dioxide emissions was also treated separately after from having been included in the GHG emissions in order to determine its separate impact, since it accounts for over 70 percent of the GHG emissions in Canada for each year under consideration. The data for the variables are taken from the OECD online data bank. The GDP per capita is measured in the constant US$ with 2005 as the base year. The data for the trade is the sum of the export and import as a percentage of the GDP. The data for the GDP per capital and the trade are taken from the World Bank online data bank. All data comprise of data from 1990 to The variables notations and definitions are as follows. GDPPK: Per capita real GDP GHGPK: Per capita greenhouse gas emissions NOPK: Per capita nitrogen oxides emissions SOPK: Per capita sulphur oxides emissions CO2PK: Per capita carbon dioxide emissions TGHG: Total greenhouse gas emissions TNO: Total nitrogen oxides emissions TSO: Total sulphur oxides emissions TCO2: Total carbon dioxide emissions TRADE V.2. Test results for unit roots When working with time series data, the first question to ask is whether or not the series is stationary. A stochastic process is said to be stationary if its mean and variance are constant over time, and if the covariance exists between the two time periods and not the actual time at which the covariance is computed. Since, the VEC specification in Equations (5) (7) requires that some 9

10 or all the variables are integrated of order one, I herein investigate the stationarity status of the variables using both the augmented Dickey Fuller (ADF) and the Dickey-Fuller Generalized Least Square (DF-GLS) tests for unit roots. The null hypothesis tested is that the variable under investigation has a unit root against the alternative that it does not (that is, it is stationary). In the ADF, lag-length is chosen using the Akaike Information Criteria (AIC) after testing for first and higher order serial correlation in the residuals while in the DF-GLS, the optimal lag is determined using Ng-Perron seq t, Schwarz Criteria (SC) and AIC. Table 1 reports the results of testing for unit roots in the level variables as well as in their first difference. The table shows the estimated t- statistics. Based on the ADF test, only NOPK and LNNOPK are stationary in their level; CO2PK, GDPPK, TCO2, LNCO2PK, LNGDPPK, and LNTCO2 are stationary in their first difference, while others are stationary in their second difference. Based on the DFGLS, GDPPK and LNGDPPK are stationary in their level; GHGPK, NOPK, TGHG, TNO, LNGHGPK, LNNOPK, LNTGHG and LNTNO are stationary in their first difference, while others are stationary in their second difference. Table1 Results of unit roots test VARIABLE ADF DFGLS VARIABLE ADF DFGLS VARIABLE ADF DFGLS GHGPK DGHGPK D2GHGPK NOPK DNOPK D2NOPK - - SOPK DSOPK D2SOPK CO2PK DCO2PK D2CO2PK GDPPK DGDPPK D2GDPPK - - TRADE DTRADE D2TRADE TGHG DTGHG D2TGHG TNO DTNO D2TNO TSO DTSO D2TSO TCO DTCO D2TCO LNGHGPK DLNGHGPK D2LNGHGPK LNNOPK DLNNOPK D2LNNOPK - - LNSOPK DLNSOPK D2LNSOPK LNCO2PK DLNCO2PK D2LNCO2PK LNGDPPK DLNGDPPK D2LNGDPPK - - LNTRADE DLNTRADE D2LNTRADE LNTGHG DLNTGHG D2LNTGHG LNTNO DLNTNO D2LNTNO LNTSO DLNTSO D2LNTSO LNTCO DLNTCO D2LNTCO *The critical values of t-statistics for the ADF are and (and that of DFGLS, and-2.89) at 5% and 10% level of significance respectively. 10

11 V.3. EKC ANALYSIS Figure1 (a-f) and Table2 (a-f) shows the result for the EKC analysis. The graphical analysis shows that none of the environmental degradation indicators considered follows the traditional inverted U EKC curve pattern, both in their levels and their logarithmic form. In the quadratic specification of the model, the coefficients on GDPPK and GDPPKSQ in the regressions with GHGPK, NOPK, SOPK, TGHG, TNO, and TSO as the dependent variable, though have the predicted EKC signs, are however, not statistically significant at the 5% level of significance. The coefficients are neither statistically significant nor have the predicted EKC signs when CO2PK and TCO2PK are the dependent variables. The coefficient on TRADE is only statistically significant with GHGPK, TGHG and TCO2 as the dependent variable. Trade is positively related to each of these indicators, suggesting that an increase in trade leads to an increase in them. Also in the logarithmic form, the coefficients on LNGDPPK and LNGDPPKSQ in the regressions with LNGHGPK, LNNOPK, LNSOPK, LNTGHG, LNTNO, and LNTSO as the dependent variable, though have the predicted EKC signs, are however, not statistically significant. The coefficients are neither statistically significant nor have the predicted EKC signs when LNCO2PK and LNTCO2PK are the dependent variables. The coefficient on TRADE is only statistically significant with LNGHGPK, LNTGHG and LNTCO2 as the dependent variable. In the cubic specification, the coefficients on GDPPK and GDPPKSQ are statistically significant with CO2PK, TCO2 and TNO as the dependent variables; the signs are however, opposite to that predicted by the EKC analysis. Also, the sign on the GDPPKCB is negative and statistically significant for these variables; this signifies the existence of more than one turning point in the curve. This pattern implies that environmental degradation will first decrease as GDP per capita rises, then increase The coefficients are neither statistically coefficient nor with the predicted signs in the regressions with other indicators as the dependent variables. The coefficient on TRADE is statistically significant with GHGPK, CO2PK, TGHG, TCO2 and TNO as the dependent variables; the sign is positive, showing that an increase in trade volume increases the emissions of greenhouse gases, nitrogen oxides and carbon dioxide at the per capita level. The logarithmic form was dropped in the cubic specification due to multicollinearity. 11

12 FIGURE 1 GHGPK NOPK GDPPK GDPPK (a) SOPK CO2PK GDPPK GDPPK (b) 12

13 TGHG (c) GDPPK TNO GDPPK (d) TSO TCO GDPPK GDPPK 13

14 3 4 LNGHGPK LNNOPK (e) LNGDPPK LNGDPPK LNSOPK LNCO2PK LNGDPPK LNGDPPK (f) 14

15 LNTGHG LNTNO LNGDPPK LNGDPPK (g) 13 LNTSO LNTCO (h) LNGDPPK LNGDPPK 15

16 Table2. Results of EKC Models (1) (2) (3) (4) GHGPK NOPK SOPK CO2PK GDPPK (0.45) (0.94) (0.27) (-0.44) GDPPKSQ -1.47e e-08 (-0.47) (-1.15) (-0.45) (0.44) TRADE * (2.54) (1.19) (0.29) (1.75) _cons (0.08) (-0.48) (0.08) (0.88) N t statistics in parentheses * p < 0.05, ** p < 0.01, *** p < (a) (1) (2) (3) (4) TGHG TNO TSO TCO2 GDPPK (0.54) (1.24) (0.25) (-0.81) GDPPKSQ (-0.27) (-1.38) (-0.41) (0.98) TRADE ** * (3.52) (1.56) (0.57) (2.17) _cons (0.04) (-0.78) (0.06) (1.23) N t statistics in parentheses * p < 0.05, ** p < 0.01, *** p < (b) 16

17 (1) (2) (3) (4) GHGPK NOPK SOPK CO2PK GDPPK ** (-1.89) (-1.97) (-1.50) (-2.89) GDPPKSQ * (1.93) (2.02) (1.51) (2.87) GDPPKCB -1.12e e e e-11 * (-1.95) (-2.09) (-1.54) (-2.85) TRADE ** ** (3.23) (1.95) (0.82) (2.93) _cons ** (1.95) (1.99) (1.54) (2.97) N t statistics in parentheses * p < 0.05, ** p < 0.01, *** p < (c) (1) (2) (3) (4) TGHG TNO TSO TCO2 GDPPK * * (-1.47) (-2.21) (-1.63) (-2.77) GDPPKSQ * * (1.52) (2.28) (1.64) (2.75) GDPPKCB e-09 * -3.58e * (-1.54) (-2.36) (-1.67) (-2.69) TRADE *** * ** (3.95) (2.47) (1.15) (3.30) _cons * * (1.53) (2.21) (1.66) (2.87) N t statistics in parentheses * p < 0.05, ** p < 0.01, *** p < (d) 17

18 (1) (2) (3) (4) LNGHGPK LNNOPK LNSOPK LNCO2PK LNGDPPK (0.37) (1.06) (0.34) (-0.63) LNGDPPK SQ (-0.37) (-1.08) (-0.35) (0.62) LNTRADE * (2.52) (1.24) (0.93) (1.86) _cons (-0.36) (-1.03) (-0.32) (0.64) N t statistics in parentheses * p < 0.05, ** p < 0.01, *** p < (e) (1) (2) (3) (4) LNTGHG LNTNO LNTSO LNTCO2 LNGDPPK (-0.02) (1.20) (0.29) (-1.34) LNGDPPK SQ (0.05) (-1.22) (-0.31) (1.36) LNTRADE *** * (4.07) (1.52) (0.99) (2.42) _cons (0.13) (-1.17) (-0.27) (1.40) N t statistics in parentheses * p < 0.05, ** p < 0.01, *** p < (f) 18

19 V.3. Test results for co-integration An F deletion test was applied to equations (2) - (4) for each form of the environmental degradation indicators (at per capita and total concentrations), per capital GDP and trade in order to test the existence of a long-run relationship. The results of bounds testing show that in the level form, there is a long-run relationship between the variables when per capita GDP is the dependent variable and GHGPK, TGHG and TCO2 are the explanatory variables; when TRADE is the dependent variable and GHGPK, TGHG, CO2PK and TCO2 are the explanatory variables, because its F-statistic exceeds the upper bound critical value at a 5% level of significance. There is also a co-integration when each of NOPK, SOPK, TGHG, TNO, TSO and TCO2 is the dependent variable. The null hypothesis of no co-integration however, cannot be rejected when each of GHGPK and CO2PK, is used as the dependent variable because F-statistics is below the lower bound critical value at a 5% level of significance. In the logarithmic form, there is a cointegration relationship when each of LNGDPPK and LNTRADE is the dependent variable, and LNGHGPK, LNSOPK and LNTSO is the explanatory variable; and also when each of LNNOPK, LNSOPK, LNTGHG, LNTNO, LNTSO and LNTCO2 is the dependent variable. Also, there is no co-integration when each of LNGHGPK and LNCO2PK is the dependent variable. The results of bounds testing are presented in table 3. VARIABLE F- STATISTICS VARIABLE F- STATISTICS GDPPK/GHGPK LNGDPPK/LNGHGPK 6.29 GHGPK/GDPPK 3.27 LNGHGPK/LNGDPPK 1.56 TRADE/GHGPK LNTRADE/LNGHGPK 7.63 GHGPK/TRADE 3.27 LNGHGPK/LNTRADE 1.56 GDPPK/NOPK 0.51 LNGDPPK/LNNOPK 1.55 NOPK/GDPPK 4.59 LNNOPK/LNGDPPK 6.86 TRADE/NOPK 0.42 LNTRADE/LNNOPK 0.94 NOPK/TRADE 4.59 LNNOPK/LNTRADE 6.86 GDPPK/SOPK 2.68 LNGDPPK/LNSOPK 6.64 SOPK/GDPPK 52.8 LNSOPK/LNGDPPK TRADE/SOPK 1.48 LNTRADE/LNSOPK 5.67 SOPK/TRADE 52.8 LNSOPK/LNTRADE GDPPK/CO2PK 4.28 LNGDPPK/LNCO2PK 2.36 CO2PK/GDPPK 4.13 LNCO2PK/LNGDPPK 3.2 TRADE/CO2PK LNTRADE/LNCO2PK

20 CO2PK/TRADE 4.13 LNCO2PK/LNTRADE 3.2 GDPPK/TGHG 4.73 LNGDPPK/LNTGHG 0.78 TGHG/GDPPK 8.73 LNTGHG/LNGDPPK 2.05 TRADE/TGHG 7.04 LNTRADE/LNTGHG 1.52 TGHG/TRADE 8.73 LNTGHG/LNTRADE 2.05 GDPPK/TNO 1.36 LNGDPPK/LNTNO 3.85 TNO/GDPPK 5.82 LNTNO/LNGDPPK 7.37 TRADE/TNO 0.56 LNTRADE/LNTNO 1.92 TNO/TRADE 5.82 LNTNO/LNTRADE 7.37 GDPPK/TSO 2.56 LNGDPPK/LNTSO 6.08 TSO/GDPPK LNTSO/LNGDPPK TRADE/TSO 1.45 LNTRADE/LNTSO 4.69 TSO/TRADE LNTSO/LNTRADE GDPPK/TCO LNGDPPK/LNTCO TCO2/GDPPK 6.76 LNTCO2/LNGDPPK 4.33 TRADE/TCO LNTRADE/LNTCO TCO2/TRADE 6.76 LNTCO2/LNTRADE 4.33 *The critical value ranges of F-statistics are and at 5% and 10% level of significance respectively [Paresh Kumar Narayan (2005)]. IV.4. Test results for Vector Error Correction and Granger causality The optimal lag length for the Vector Error Correction model was determined using the AIC and the SBIC. It was found to be 2. This is shown in Table4 (a- d) below. Table4. Lag length determination for VEC GDPPK GHGPK NOPK SOPK CO2PK TRADE Lag LL LR df p FPE AIC HQIC SBIC e * * * * * (a) 20

21 GDPPK TGHG TNO TSO TCO2 TRADE Lag LL LR df p FPE AIC HQIC SBIC e e e+26* * * * * (b) LNGDPPK LNGHGPK LNNOPK LNSOPK LNCO2PK TRADE Lag LL LR df p FPE AIC HQIC SBIC e e e-21* * * * * (c) LNGDPPK LNTGHG LNTNO LNTSO LNTCO2 TRADE Lag LL LR df p FPE AIC HQIC SBIC e e e-20* * * * * (d) The results of short- and long-run Granger causality are determined within the VECM framework. The short-run causal effects are demonstrated through the chi square-statistics of the explanatory variables and long run causality is tested with the help of statistical significance and sign of the error correction term. The short run granger causality results are present in Table5 (ad) below. The results show that there is a granger- causality from per capita GDP and trade to each of the environmental degradation indicators(at both the per capita and total concentrations), at the 5% significance level, both in the level and logarithmic forms, except the logarithmic form of per capita and total greenhouse gas emissions. There is also a granger-causality from each of the indicators to per capita GDP; also from each of the indicators to trade, except for per capita nitrogen oxides and total nitrogen oxides emission in the level form, at 5% significance level. 21

22 Table5. Short run Granger-causality test Equation Parms RMSE R-sq chi2 P>chi2 D_GHGPK D_GDPPK D_TRADE D_NOPK D_GDPPK D_TRADE D_SOPK D_GDPPK D_TRADE D_CO2PK D_GDPPK D_TRADE (a) Equation Parms RMSE R-sq chi2 P>chi2 D_TGHG D_GDPPK D_TRADE D_TNO D_GDPPK D_TRADE D_TSO D_GDPPK D_TRADE D_TCO D_GDPPK D_TRADE (b) 22

23 Equation Parms RMSE R-sq chi2 P>chi2 D_LNGHGPK D_LNGDPPK D_LNTRADE D_LNNOPK D_LNGDPPK D_LNTRADE D_LNSOPK D_LNGDPPK D_LNTRADE D_LNCO2PK D_LNGDPPK D_LNTRADE (c) Equation Parms RMSE R-sq chi2 P>chi2 D_LNTGHG D_LNGDPPK D_LNTRADE D_LNTNO D_LNGDPPK D_LNTRADE D_LNTSO D_LNGDPPK D_LNTRADE D_LNTCO D_LNGDPPK D_LNTRADE (d) VI. CONCLUSION AND POLICY IMPLICATION The results in this paper derived from four indicators of air pollution in Canada show that the reduced form models of the EKC do not provide an adequate representation of the growthenvironment relationship in Canada. It however, reveals a positive relationship between trade and the emissions of greenhouse gases in general, carbon dioxide and nitrogen oxides. This implies that an increase in the volume of the Canadian trade leads to the generation of more of these pollutants. The results also show that there is a long run relationship between each of these pollution indicators and economic growth and trade; and also a bi-directional granger-causality between each of the indicators and economic growth and trade. Canada s per capita GHG 23

24 decreased by nearly 5 percent between 1990 and 2010, while its total GHG emissions grew 18 percent (Environment Canada, 2014). The largest contributor to Canada s GHG emissions is the energy sector, which include power generation (heat and electricity), transportation, and fugitive sources (Conference Board of Canada). The energy sector was responsible for 81 percent of Canada s total GHG emissions in 2010, out of which energy combustion accounts for 45 percent (Conference Board of Canada). Carbon dioxide accounts for the largest proportion of the GHG emissions; it contributed 79 percent of Canada s total emissions in 2012 (Environment Canada, 2014). Canada is one of the largest emitters of GHG in the world, being one of the world s largest energy exporters. The main reason for the increase in Canada s GHG has been growth in exports of petroleum, natural gas and forest products. Air quality is affected by sulphur oxides emitted from smelters, electricity generators, petroleum refineries, iron and steel mills, and pulp and paper mills. Between 1990 and 2009, Canada decreased its per capita sulphur oxides emissions by 34 percent. While the reduction was good, Canada s progress was weaker than the progress made by 14 of its 17 OECD s peer countries (Conference Board of Canada). In 2012, sulphur oxides emissions decreased by 0.3 percent from 2011 levels, and were 59 percent lower than in 1990 (Environment Canada,2014). Nitrogen oxides contribute to smog and acid rain and are hazardous to human health and the environment. Nitrogen oxides are released during the combustion of fossil fuels, mainly by vehicles, electricity generation, and manufacturing process. In 2012, nitrogen oxides emissions in Canada decrease by 5 percent from 2011 levels, and were 27 percent lower than in 1990 (Environment Canada, 2014). Canada needs to do more to reduce emissions from the transportation, electricity, and industrial sectors (Conference Board of Canada). These results suggest that, Canada the main challenge for Canada is to further reduce urban and regional air pollutants through more pollution control, technological progress, energy savings, and sustainable transportations. There are already some moves in this direction. The Canadian federal government recently set a new target of reducing total GHG emissions by 17 percent from 2005 levels by To achieve this it has introduced three major initiatives: passenger automobile and light truck GHG emissions regulation; heavy duties vehicles emissions regulations; regulations on coal fire electricity regulations (Government of Canada, 2012). Federal and provincial and United States agreements on capping sulphur oxides; and introduction of cleaner technology and fuels for vehicles. However, Canada still needs to do more, in order to maintain a cleaner environment. REFERENCE Antweiler, W., B.R. Copeland, and M.S. Taylor (2001) Is free trade good for the environment? American Economic Review 91, Arrow, K., B. Bolin, R. Constanza, P. Dasgupta, C. Folke, C.S. Holling, B.O. Jansson, S. Levin, K.G. Maler, C. Perrings, D. Pimentel (1995) Economic growth, Carrying capacity, and the Environment. Science 268,

25 Beckerman, W. (1992) Economic Growth and the Environment: Whose Growth? Whose Environment? World Development 20, Cavlovic, T., K.H. Baker, R.P. Berrens, and K. Gawande (2000) A Meta-Analysis of Environmental Kuznets Curve Studies. Agricultural and Resource Economics Review 29, Cole M.A. (2004) Trade, the Pollution Haven Hypothesis and the Environmental Kuznets curve: Examining the linkages. Ecological Economics 48, Cole, M.A., A. J. Rayner, J.M. Bates (1997) The environmental Kuznets curve: an empirical analysis. Environment and Development Economics 2, Dasgupta, S., B. Laplante, H. Wang, and D. Wheeler (2002) Confronting the Environmental Kuznets curve. Journal of Economic Perspectives 16, Day K., and R.Q. Grafton (2001) Economic Growth and the Environment: A Canadian perspective. Working Paper 0101E. Ottawa: University of Ottawa, Department of Economics Devlin R.A., and R.Q. Grafton (1998) Trends in Canada s Natural Capital. Paper prepared for the conference on the State of Living Standards and the Quality of Life in Canada, Ottawa, October 30-31, 1998 Environment Canada (2014) National Inventory Report : Greenhouse gas sources and sinks in Canada. Gale, L.R., and J.A. Mendez (1998) The Empirical Relationship between Trade, Growth and the Environment. International Review of Economics and Finance 7, Government of Canada 2012 Grossman, G,M., and A,B. Krueger (1995) Economic Growth and the Environment. Quarterly Journal of Economics 110, Grossman, G,M., and A,B. Krueger (1993) Environmental Impacts of the North American Free Trade Agreement. In: The U.S.-Mexico Free Trade Agreement, P. Garber, Massachusetts Institute of Technology. The MIT Press, Cambridge, MA., Grossman, G,M., and A,B. Krueger (1991) Environmental Impacts of the North American Free Trade Agreement. National Bureau of Economic Research Working Paper 3914, NBER, Cambridge MA 25