Revisiting the Environmental Kuznets Curve Hypothesis in Pakistan

Several studies have already determined an inverted U-shaped Environment Kuznets Curve (EKC) in Pakistan. The existing literature has not considered structural breaks (SBs) in EKC-related studies in Pakistan. This study aims to understand whether SBs explain the EKC hypothesis in Pakistan from 1980-2016. The variables used include total energy consumption (TEC), real GDP per capita, foreign direct investment (FDI), and trade openness (TO). The current study has used conventional time series econometric methods to analyze the issue. A structural break (SB) can significantly impact the forecasting performance of a model. Therefore, we have used the Zivot-Andrews unit root test (ZAURT) with one structural break (SB) and the Gregory-Hansen cointegration test approach for empirical analysis. The Gregory-Hansen cointegration test also suggests that the long-run equilibrium relationship is affected by structural breaks (SBs). Historical data suggests that Pakistan has gone through some structural changes during the period 2000-2004, which includes implementing the structural adjustment program of IMF and liberalization of trade and investment policies to attract foreign investors. The 9/11 tragedy also played an important role as Pakistan remained on the front lines in the war against terrorism. Thus, the study concludes that structural breaks (SBs) have important implications for the EKC hypothesis in Pakistan.


Introduction
Foreign direct investment (FDI) is essential for capital inflows and economic development. FDI is equally important for both under-developed and developed countries (Solarin & Al-Mulali, 2018). Developing countries generally have a shortage of capital; therefore, their development process depends on capital inflows. On the 1 decline in FDI and fluctuations in GDP growth, CO2 emissions reached approximately 166,300 kt by 2014-2015 (Sengupta & Puri, 2020). Many researchers think that an increase in economic development (ED) deteriorates the environmental condition (EC) of a country (Siping, et al., 2019;Aung, Saboori & Rasoulinezhad, 2017). Past studies also suggest that economic development (ED) initially promotes environment decay (ED). It reaches the highest possible level in the next few years, and subsequently, it declines as the economy develops further (Selden & Song, 1994;Grossman & Krueger, 1991;Rothman & de Bruyn, 1998).
The study aims to revalidate the EKC hypothesis for Pakistan. The EKC suggests that economic growth (EG) and development initially contribute towards environmental decay (ED), but in the long run, it reduces environmental decay (ED) (Rothman & de Bruyn, 1998). Past literature suggests that energy consumption (EC) in a country and environment quality (EQ) are highly associated. Therefore, this study has taken "Co2 emission and energy consumption (EC) in the model" (Ali, Ashraf, Bashir & Cui, 2017). Many studies have examined EKC in Pakistan, but they have not investigated the significance of structural breaks (SBs) for Pakistan's environment-growth nexus (EGN) (Zhang, Wang & Wang, 2017;Gokmenoglu & Taspinar, 2018). The current research contributes to the existing literature on the EKC hypothesis as we have incorporated one endogenous structural break (SB) in the model for the period 1980-2016. We have also included "FDI and trade openness (TO)" in the model to determine their relevance with Pakistan's environmental degradation (ED).

Literature Review
Researchers argue that when a host country receives more investment, its environmental protection policies and implementation become strict (Perman & Stern, 2003). Many studies in developing countries have documented that sulfur emissions (SE) and economic growth (EG) are highly associated. (Kim & Baek, 2011;Abdo, Li, Zhang, Lu & Rasheed, 2020). Similarly, Liddle & Messinis (2018) also validated the same results. Stern (2004) found "empirical evidence in support of the EKC. " Boyd & Smith (1992) also found an association between "environmental degradation (ED) and economic growth (EG). " Demena & Afesorgbor (2020) also found that environmental degradation (ED) declines "after a certain level of economic growth (EG). " A similar study using a data set of countries belonging to different income groups found a negative but insignificant growth-environment nexus in high-income countries (Porter & Van-der-Linde, 1995). Other studies also found support for the EKC in high-income countries (Ulucak & Bilgili, 2018). Dogan & Inglesi-Lotz (2020) found evidence of EKC in "middle and lowincome countries. " At the same time, Choi & Han (2018) also found that it promotes environmental degradation as the income level increases. Similarly, Chen, Fan & Guo (2020) suggest that a country's economic prosperity promotes environmental decay and pollution. However, this relationship is not linear. It varies from developed countries (DC) to developing countries (Norbutas & Corten, 2018). Boyd & Smith (1992) suggest that FDI is not effective in countries with liberalization and deregulation policies. The literature also suggests that growth and development in a country promote environmental degradation and adversely affect human well-being (Kim & Baek, 2011). Similary, Nováková, Šujanová & Nováková (2019) suggest that the "association between economic prosperity (EP) and environmental decay (ED) do not increase at the same rate. " It depends on the GDP of a country. For example, countries whose GDP is low may adversely suffer due to economic growth. However, countries with a higher GDP contribute less to environmental degradation (Grossman & Kureger 1995;Ghebrihiwet & Motchenkova, 2017). At the same time, FDI helps in technology transfer to the host country, contributing to its overall growth (Romer, 1993). Alfaro, et. al., (2004) and Šušteršič & Kejžar (2020) found that FDI does not linearly affect all sectors of an economy. It significantly promotes the manufacturing sector and has an insignificant effect on the service sector (Herlitah, Fawaiq & Herlindah, 2020). Similarly, Herzer & Klasen (2008) based on the data set of twenty-eight developing countries, also found similar results. A few researchers investigating the growth-environment nexus found that a country at the initial rapid growth phase promotes environmental decay (Pandey, Dogan & Taskin, 2020). However, the environmental conditions improve after growth, and per capita income reached a certain level (Liddle & Messinis, 2018;Grimes & Kentor, 2003). Many researchers have also found that foreign investors prefer investing in economies with moderate environmental protection policies (Copeland & Taylor, 2005;Kurniawan, Sugiawan & Managi, 2021) or where the governments don't focus on environmental quality to attract foreign investment (Wei & Smarzynska, 1999).
Beak & Koo (2011) examined the EKC hypothesis in India and China. The study found that in India, FDI contribution towards energy emission is insignificant in the short run and significant in the long run. Comparatively, in China, "FDI has significantly increased energy consumption (EC), economic growth (EG), and CO2 emissions" (Salim, Yao, Chen & Zhang, 2017). Kim & Beak (2011), using an ARDL bounds approach, found that in advanced countries, economic growth (EG) increases energy emission (EE). Similarly, other studies also concluded that energy demand promotes energy emission while FDI insignificantly affects environmental decay (Khan, Hussain, Bano & Chenggang, 2020;Rafindadi, Muye & Kaita, 2018). also found that the association between FDI and CO2 "emission level depends on the countries' income levels. " FDI contribution towards CO2 emission is high in middleincome countries and insignificant in high-income countries (Muhammad & Khan, 2019;Pazienza, 2019). Based on empirical evidence, Ugur & Gultekin (2018) concluded that FDI in a country promotes CO2-related pollution, although its intensity may be on the lower side. Pao & Tsai (2011) also found support for the EKC hypothesis in BRIC countries. The study also found that FDI inflow causes environmental degradation (ED). On the contrary, many researchers believe that the conventional econometric methods lack the power to validate the EKC hypothesis (Pata, 2019;Stern, 2004;Herzer & Klasen, 2008;Xing & Kolstad, 2002). Given this constraint, researchers have focused "on structural breaks (SBs) while validating the EKC hypothesis. " Tiwari (2012) examined the association between GDP, energy consumption (EC), and pollution in India. The study used the static and dynamic frameworks and found the "structural breaks (SBs) in the model. " At the same time, Jaunky (2011) did not find support for the "EKC hypothesis in the presence of structural breaks (SBs). The results were based on the panel data of 36 countries. Mahmood & Chaudhary (2012) based on the Zivot-Andrews test (Zivot & Andrews, (2002) found "structural breaks (SBs)" in the model. The study also found that "FDI, CO2, and population density (PD) are associated in the long-term only. In the short term, these variables have no association. Yousaf et al. (2016), in a study of Pakistan over the period 1972-2013, found that foreign loans and aid promote CO2 emissions. Ahmed & Long (2010) examined the validity of the EKC hypothesis in Pakistan over 1971-2008 by using the ARDL approach. The study found "evidence of both long run and short run EKC in Pakistan. " Ur-Rehman et al. (2019) used the nonlinear ARDL method to confirm the population haven hypothesis in Pakistan. The study also found evidence for the EKC hypothesis with the nonlinear specification in Pakistan. Cetin et al. (2018), based on data from 1960-2014, found the "presence of EKC with one structural break (SB). " Pata (2019) adopted the "bootstrapped autoregressive distributed lag (ARDL) model to validate the EKC hypothesis and the presence of structural breaks (SBs)" in Turkey from 1969-2017. The results suggest a long-term association between "trade openness (TO), per capita income, per capita real income, and CO2 emissions, and the presence of one structural break. " Salahuddin et al. (2019), in a study in South Africa, used the Zivot-Andrews unit root test and found a strong association between "CO2 emissions, globalization, and urbanization. " Ugur & Gultekin (2018) reinvestigated the association between "environmental degradation (ED) and economic growth (EG) in Turkey for the period 1960-2011. The study used the Zivot-Andrews unit root test (ZAURT) and Gregory-Hansen cointegration (GHC) method. The study also found evidence of the EKC hypothesis in Turkey with one structural break in 1992. Alvarado & Toledo (2017), based on empirical evidence, concluded that it is possible to reduce environmental degradation, which is also a sign of developed economies. Hundie & Daksa (2019) found that there exists an "inverted U-Shaped Curve for Environment-growth nexus. " Felix-Fofana (2018) suggests that the industrialization and environment quality relationship is nonlinear. At the initial stage of industrialization, a country's environmental quality is adversely affected. But at the later stage of industrialization, environmental decay decreases. Thus, countries need to align development, growth, and energy consumption (Ozcan, Tzeremes & Tzeremes, 2020). Perman & Stern (2003) analyzed the EKC hypothesis for 23 OECD countries using carbon emission data and GDP per capita. The study used a model that incorporated multiple endogenous structural breaks (SBs). The study found support for the EKC hypothesis in only 4 out of 23 countries. For another 15 countries, the authors found insignificant effects of income on CO2 emissions due to positive but declining energy emission elasticity. The study concluded that the presence of the EKC hypothesis is country-specific and time-varying.

Methodology
This paper aims to analyze the EKC hypothesis in Pakistan. The data for the period 1980 to 2016 was obtained from secondary sources. The variables used in the model are inclusive of "FDI, real-GDP per capita, CO2 emission and trade openness (TO). " The validity of EKC hypothesis with structural breaks (SBs) was tested by extending the work of Mahmood & Chaudhary (2012)  The EKC hypothesis suggests that economic growth (EG) increases energy consumption (EC) proportionally in the short run (SR). However, in the "long run (LR), economic growth (EG) increases energy consumption (EC) at a slower rate. " Thus, we expect β 1 to have a positive sign. β 3 , in the long run, may have a negative sign showing a declining trend of energy consumption (EC). Per capita, energy consumption (EC) may contribute towards pollution. Thus, the expected sign of β 5 will also be positive. We have also added two other important variables in the model, i.e., trade openness (TOP) and FDI. We expect that trade openness will negatively affect energy emission, and FDI will increase environmental degradation. FDI influences the production capacity, and higher production "increases energy consumption (EC) and carbon emissions. " Before performing time series analysis (TSA), we checked the order of integration of the variables. Subsequently, we examined their long-term relationship.

Results and Discussion
The study aims to identify if structural breaks (SBs) significantly affect "FDI, CO2 emissions and Pakistan's economy (PE)" for the period 1980-2016. The study has ascertained the "order of integration based on unit-roots. " We have used "both the conventional unit root tests, i.e., Augmented Dickey-Fuller and the Phillips-Perron unit root test. " Table 1 "suggests that all the variables are non-stationary at level, " suggesting that a unit root is present among all data series. However, all the 'variables become stationary when tested at the first difference. " The results confirm that all individual data series were of order I(1). The Zivot-Andrews unit root test with one structural break also confirms the results of conventional unit root tests in Table 2.  After determining the "non-stationary variables and the order of integration, we determined the long-run equilibrium relationship in the model using the Johansen cointegration test. " The optimal lag length was determined using the Akaike Information Criteria (AIC). The Johansen cointegration test results in Table 3 "confirms the presence of a long-run association between the variables in the model" as the trace statistics show that 4 co-integrating equations and Eigenvalue statistics indicate 3 co-integrating equations, thus confirms the presence of cointegration in the model. Table 4 showed a "positive relationship between FDI and GDP with CO2 emission. " GDP 2 was found to have a negative sign as expected. Interestingly TOP "also seems to have a positive relationship with CO2 emissions, " which indicates that trade openness also hampers the environmental condition in Pakistan.
To confirm the conventional cointegration test results and "determine the possible significance of structural breaks (SBs) in the model, we used the Gregory-Hansen cointegration test. In the model, we also incorporated one endogenous structural break (SB). The Gregory-Hansen cointegration test in Table 5 further confirms the cointegrating relationship "in the model in the presence of one structural break" at the 1% level of significance. The coefficient of GDP 2 reported in Table 4 shows an expected negative sign thus, "confirming the presence of EKC in Pakistan. " The study found support for the long-term relationship in the model both with and without a structural break. Subsequently, the study determined the error-correction terms in the model. The error correction model results in Table 6 show the error correction term for energy consumption (EC), CO2 emission, and GDP 2 . However, our results suggest that FDI, GDP, and TOP have no short-run impact on the model. The study "also found a positive sign for GDP and a negative sign" for GDP 2 in the error correction model.   We used the Wald test of causality within the error correction framework to conclude the "direction of the causal relationship between the variables. " The results of Granger causality in Table 7 indicate that unidirectional causality exists "between FDI and CO2 emissions, and the direction of causality runs from FDI to CO2 emissions. " A "bidirectional causal relationship exists between GDP growth and CO2 emissions. " The results also "show a bidirectional causal relationship between energy consumption (EC) and CO2 emission. " The results suggest that as the "foreign direct investment inflow increases in the economy, environmental degradation also increases. " The results also indicate bidirectional causality between TOP and CO2 emissions, suggesting an increase in trade volume due to trade openness would increase air pollution. The increased production will also affect the environment. FDI and GDP also have a bidirectional causal relationship. The results also indicate bidirectional causality between TOP and FDI, suggesting trade liberalization and FDI are interrelated and

Conclusion
The study determines the presence of the EKC hypothesis in Pakistan for the period 1980-2016. The study documents some important findings. The results support the EKC hypothesis in Pakistan. The coefficients of trade openness and FDI also have important policy implications as it is evident that FDI and trade openness positively affect CO2 emissions in Pakistan. Many researchers believe that Pakistan has not properly implemented environment protection policies in manufacturing sectors such as textile and chemicals. The Gregory-Hansen cointegration test also suggests that the long-run equilibrium relationship is affected by structural breaks (SBs). Historical data suggests that Pakistan has gone through some structural changes during the period 2000-2004, which includes implementing the structural adjustment program of IMF and liberalization of trade and investment policies to attract foreign investors. The 9/11 tragedy also played an important role as Pakistan remained on the front lines in the war against terrorism. Thus, the study concludes that structural breaks (SBs) have important implications for the EKC hypothesis in Pakistan.