THE IMPACTS OF URBAN SPRAWL ON ENVIRONMENTAL POLLUTION, AGRICULTURE, AND ENERGY CONSUMPTION: EVIDENCE FROM AMMAN CITY

Purpose: Urban sprawl can cause an increase in economic development, and as a result, it is followed by increased consumption of energy, waste generation, urbanization, and reduced green land. In this sense, the current study aims to investigate the impacts of urban sprawl on environmental pollution, in terms of CO2 emissions, while considering other control variables. Further, this study explores the relationship between population growth and agricultural land and energy consumption, in the period between 2008 to 2020, in Amman city. Method/design/approach: This study employs the STIRPAT model and the panel unit root test, and the panel cointegration test (ADF), in order to investigate the effect of urban sprawl on the emissions of CO2. Theoretical framework: Urban expansion is a low-density, dispersed, and poorly planned form of spatial growth that has a number of detrimental effects on the quality of the environment. Results and conclusion: The results describe that waste generation, economic development, and urbanization had negative impacts on CO2 emissions. Research implications: The findings indicated the influence of urban growth on the consumption of energy and agricultural land in Amman. Originality/value: This work can provide possible solutions to reduce the impacts of urban sprawl on the environment through implementing regulations and policies and re-adjustment of urban land use in an efficient manner.


INTRODUCTION
Urbanization is an intricate process by which the rural way of lifestyles is transformed into urban ones. It is commonly defined as the shifts that take place in a region's territorial and socioeconomic development, including the general change from non-developed to developed land use categories (Makhamreha & Almanasyeha, 2011). Globally, urban areas have been growing gradually and rapidly, often at the expense of natural and semi-natural territory (Kourtit et al., 2014;Martellozzo et al., 2018). The global human population will further increase by 30-70% in this century, which will lead to a population shift from rural to urban areas and to significant land-uptake for urban expansion (Gerland et al., 2014;Hennig et al., 2015a). In the recent years, the world has seen a significant trend toward urbanization. Nearly 50% of the world's population currently resides in urban regions, and by 2030, 60% of all people will live in urban areas, with 100 megacities predicted to exist (Avelar et al., 2009). By 2050, as the accelerating rate of urbanization, it is anticipated that two thirds of the worlds' inhabitants will be residing in metropolitan regions (Abu Hatab et al., 2019).
The idea of urban sprawl has been expressed by different researchers using key terms such as economic sprawl (Cobbinah & Amoako, 2014), transportation sprawl (Cobbinah & Amoako, 2014;Madre et al., 2002), and geographical sprawl (Cobbinah & Amoako, 2014;Pirotte & Madre, 2011), has emerged in 1937 by Earle Draper, who created the concept "sprawl" in the United States of America, and ever since, this key-term has been used by city planners to describe the inefficient style of urban expansion and growth (Cobbinah & Amoako, 2014;Howard, 1965). Urban sprawl is a form of uncontrolled development pattern over the boundary of a metropolis and is a progressively popular feature of the built environment particularly in the developed regions. The phenomenon of urban sprawl diminishes the organization of physical development that is responsible for generating economically efficient land use and managing the borders of rapidly urbanizing metropolises (Cobbinah & Amoako, 2014). Urban policy has a significant impact on urban sprawl, which in turn effect changes in the emissions of GHGs from urban regions. According to studies, in recent years, urban regions accounted for 60 to 80 percent of the world's energy consumption, and more than 70 percent of the world's GHG emissions are generated in urban cities (Fragkias et al., 2013;Fujii et al., 2017).
The increased prevalence of urban sprawl is resulting in many severe environmental, economic, and social concerns. Some of the examples include the transformation of agricultural and other lands into built-up areas, which have led to augmented consumption and demand in energy and mobility, higher air pollution and landscape fragmentation, increased spread of invasive species, deterioration and/or loss of most ecological soil functions and declined resilience in the ecosystem (Hennig et al., 2015b;Travisi et al., 2010;Wilson & Chakraborty, 2013). For example, to improve the production of goods and meet the demands of expanding populations, industrialization is inevitable in developing economies and is considered as essential to the advancement of the economy of these countries (Rodríguez Martín et al., 2015). Increased pollution is a conceptual environmental issue linked to urban sprawl in both space and time. The recent and continuous migration of people from rural to urban areas has resulted in rapid urbanization and industrialization, which has increased demand for transportation and energy, accompanied with an increase in anthropogenic activities (e.g., vehicle traffic or industrial activities), and this demand is directly related with the atmospheric anthropogenic emissions (Valdes et al, 2022;Grigoratos et al., 2014). Atmospheric pollution is considered one of the most significant issues in metropolitan settings, and it is greatly influenced by dry or wet deposition (Fantozzi et al., 2013). The release of heavy metals (such as Pb, Zn, Cr, Ni, or Cd) in the urban atmosphere occurs in a variety of particle sizes and is mostly related to trafficrelated emissions brought on by incomplete combustion of fossil fuels by diesel-powered cars or industrial operations (Gutiérrez et al., 2022). The significance of this phenomenon must be considered in the perspective of perpetual urban growth, which is prevalent in the majority of the world (Lv et al., 2013;Rodríguez Martín et al., 2015).
As a developing country, Jordan is a nation with little natural resources, and like other developing nations, agriculture is the main industry for ensuring food security and the main source of income for many households. Due to the increase in population growth, urban expansion grew significantly in the later decades of the 20th century, and this population growth, led to raising the demand of lands for housing and human services. As a result of urban sprawl and poor planning, agricultural land began to disappear progressively, and the country of Jordan experienced a rapid cultural shift and population increase that changed the country's traditional relationship between people and the environment (Madallah & Tarawneh, 2014a). Despite the severe scarcity of natural resources, particularly water and electricity, that Jordan experiences, during the last decades, Jordan has experienced fast urbanization due to the influx of refugees and large-scale immigration from Syria, Palestine, and Iraq. The degradation of green spaces and water supplies is a byproduct of this largely unplanned and thus uncontrolled urbanization (Alnsour, 2016;Rawashdeh & Saleh, 2006).
According to the World Bank and the Jordanian Department of Statistics, the estimated population of Jordan in the year 2021 was 10.27 million inhabitants, with a population growth of more than 2 percent, and urbanization rate increasing from 51% in 1960 to 92% in 2021 (Department of Statistics, 2022; The World Bank, 2021). Due to this growth, the number of metropolitan centers has increased in the country, while the population of rural areas has decreased. In addition, it resulted in a significant growth in the rate of inflation in urban areas, owing to the lack of ability of current services to adapt the surge in population and housing. The continuing urbanization and refugee influx have led to a number of problems, including informal settlements, overcrowding, the deterioration of agricultural land, poor quality urban services, traffic congestion, air pollution, and a lack of green space (Alnsour, 2016). The continuous urbanization, changes, and challenges in Jordan justify the need to investigate the impacts of this "urban sprawl" on different aspects of the environment. This investigation can help the authorities to manage the ongoing urban growth, hence corrective regulations, policies, and practices can be proposed and put into act in order to deal with the inevitable issue of urban sprawl rapidly and successfully. Evaluation of the negative impacts of urban growth can help people in management to improve the capability of urban management to better react to the burdens and effects of rapid urban growth. Given that, the main research objective of this work is to evaluate the impacts of urban sprawl in the capital city of Jordan, Amman, on air pollution, through measuring the development of carbon dioxide (CO2) in the ambient air. The second objective is to assess the relationship between urban sprawl in Amman with the consumption pattern of energy and agricultural lands. These indicators can help us to understand the current environmental impacts of urban expansion on such a small, yet dense city as Amman. The contribution of this research highlights the issues associated with urban sprawl in Amman and it is expected to help managing authorities to further implement policies to deal with this problem.

REVIEW OF LITERATURE
Urban growth is a spatial phenomenon caused at a particular stage of social and economic development. Research on urban growth has gained a lot of attention since the 1960s, and it has steadily evolved into the theoretical foundation for "regionalism," "new urbanism," and "smart growth" (Zhao et al., 2016). Urban expansion was described by researchers as a "low-density, private transportation dependent, land development model that takes place on the borders of cities". Some scholars described urban sprawl as the extreme growth in the economy and space of the urban city after the failure of the "invisible hand" (Rimal et al., 2019). In addition, it is argued that urban expansion is primarily influenced by several factors, which include policy factors, market factors, cultural factors, and social factors, with robust "spatial flexibility" and policy guidance (Zhang et al., 2022). In general, urban expansion is a lowdensity, dispersed, and poorly planned form of spatial growth that has a number of detrimental effects on the quality of the environment. Urban sprawl is a highly complicated and multifaceted process that is influenced by the economy, population, and land. At the moment, urban growth is primarily determined using single and multiple indicators. The most common indicators include growth rate (i.e., land use, economic, and population), density (i.e., employment density, residential density, and population density), landscape pattern (i.e., spatial connection, fractal dimension, and aesthetic degree), and spatial form (proximity, degree of fragmentation, accessibility) (Debbage et al., 2016;Zhang et al., 2022). For instance, to assess the growth of urban areas, Feng et al. created population density and growth (Feng et al., 2019). While Li et al. utilized urban built-up areas in order to describe the level of urban sprawl . Other researchers applied density, discontinuity, and accessibility to assess the expansion in urban cities (Yue et al., 2016).
Several studies have investigated the impacts of urban sprawl on air quality and reported the relationship between urban expansion with different environmental aspects, such as energy consumption, water consumption, and land use. For example, a comprehensive study by Chen et al. investigated the casual relationship between urbanization. energy intensity, energy consumption, economic growth, and concentrations of PM2.5 for 141 countries during the period 1998-2014. The studied countries were classified into four categories according to their income. The findings showed that all panels had cointegration connections between the examined variables and PM2.5 values. The panel Granger causality test using a Vector Error-Correction Model revealed that urbanization, economic growth, energy consumption, and energy intensity all contributed to rising PM2.5 concentrations over time. The economic growth factor was the primary variable that had an effect on the concentrations of PM2.5 in the global panel, the high-income panel, and the upper-middle income panel. The results argued that in all nations, with the exception of those with low incomes, PM2.5 concentrations can be reduced in the short term by increasing energy intensity. Contrarily, lowering the level of urbanization in the short term is ineffective for reducing PM2.5 concentrations. In addition, the findings showed that in lower-middle-income and low-income countries, the energy consumption structure was the biggest contributor on PM2.5 concentrations (Chen et al., 2018).
In the case of Jordan, several studies have investigated the impacts on urban sprawl on different aspects of the environment. For example, the disintegration of rainfed lands caused by urban sprawl was studied (Hammad, 2017), and the results showed that due to urban expansion, the area of agricultural land significantly decreased from 204.6 km 2 in 1987 to 105.7 km 2 in 2005. Riad et al. also investigated the landscape transformation processes in Amman, Jordan, and indicated a large reduction in the green peri-urban land of Amman by 122.4 km 2 (Riad et al., 2020). Nevertheless, Amman's urban core has changed significantly over the past years with high rate of population growth. For instance, the urban sprawl in Amman increased the total settlement areas from 36 to more than 250 km 2 , at the expense of agricultural land (Makhamreha & Almanasyeha, 2011). Other studies on the impact of urban sprawl on land-use and transformation in Jordan have been also investigated in different studies (al shawabkeh et al., 2019;Madallah & Tarawneh, 2014b;Rawashdeh & Saleh, 2006). However, to the best of our knowledge, the impacts of urban sprawl on the environmental pollution in terms of air pollution for example, have never been investigated in Jordan.
The environmental air pollution is greatly ascribed to the generation and consumption of fossil fuels and agricultural development. Nonetheless, efficiency of energy consumption, energy intensity, emissions per capita, agricultural economic growth vary amongst countries. As a result, it can be seen that the levels of air pollution characterized by PM2.5 and CO2 fluctuate among nations. For example, Qiao et al. studied, in a panel of 19 countries of the G20 nations from 1990 to 2014, the impacts of economical agricultural development and renewable energy advancement and growth on the emissions of carbon dioxide (CO2) within the structure of the environmental Kuznets curve (EKC) (Qiao et al., 2019). Their research analysis was conducted using different analytical tests, including panel data unit root tests, cointegration tests, and the panel fully modified ordinary least squares (FMOLS) estimator. The main results of their research confirmed a long-run relationship between the chosen variables based on the performed panel data unit root test and cointegration. Further, air pollution characterized by CO2 was significant in the developing economics of the G20 due to agricultural development, whereas in the developed economic of the G20, the implementing of renewable energy techniques and strategies abridged the emissions of CO2. Nevertheless, the findings confirmed the existence of the EKC in the developed economies of the G20, while in developing countries of the G20, economic growth showed a positive impact on the emissions of CO2 emissions.
Urban sprawl is strongly linked to increased CO2 emissions in highly urbanized areas, especially in developing countries. This is due to the increase in the urban transport sector. The continuous and growing reliance on personal automobiles, increased travel distance and frequency, and elevated fuel ineffectiveness are all significant reasons that can rapidly raise carbon emissions. These factors interact with one another and co-determine one another, proving that they are not independent forces. A study by Andong et al. investigated the key aspects of the transportation industry, especially its public mass transportation component, in a city in a developing country that makes it the primary source of carbon emissions, as well as the spatial features in a city of a developing country, and how the developed public transportation system and patterns of transportation can contribute to high emissions of CO2 (Andong & Sajor, 2015). The findings from their study claimed that the upsurge in CO2 emissions is primarily caused by the convergence of several interrelated factors, including urban sprawl, the affordability of housing close to places of employment, commuters' high reliance on public transportation, their longer commute distances, and the public utility vehicles' poor fuel efficiency.
Urban sprawl rate is controlled by a number of factors, and these same factors also determine in what direction the urban sprawl growth will continue. According to researchers, the following factors can be identified to affect and drive forces of the urban growth, including population growth, developed transportation system, leap-frog development, industrialization, 6 industrial and commercial activities, and unrestricted/uncontrolled external expansion (Gutiérrez, 2022;Cobbinah & Amoako, 2014). Environmental pollution in urban areas primarily caused by urban sprawl is due to effects such as, increase in the use of urban land and the increased population density. The devastation of the environment due to growth in economy is largely caused by extreme energy consumption. It is worth noting that the growth of economies relies on energy, and the development of economies is complemented by heightened energy consumption, including both clean and polluting energy sources, with the utilization of polluting energy being dominating than clean energy. Economic growth and energy use are strongly related; greater economic growth necessitates more energy consumption, while greater energy efficiency necessitates greater economic growth. The direction of causality may therefore not be known in advance (Liang & Yang, 2019).

H1.
There is a direct effect of urban expansion and population growth on the environmental pollution. H2. Urban sprawl impacts the area of agricultural land. H3. There is a direct effect of urban expansion or population growth on energy/ electricity consumption (i.e., higher levels of sprawl will show higher per-capita residential energy consumption).

VARIABLES SELECTION AND METHODOLOGICAL FRAMEWORK
This research study intends to investigate the impact of urban sprawl on environmental pollution characterized by CO2 emissions in Amman, Jordan from 2008 to 2021. Hence, this work includes one dependent variable and five control variables. The dependent variable represents carbon dioxide emissions (ECO2), while the control variables include population density (DPOP), economic development (GDP), which is characterized by the gross domestic product, urbanization (URB), energy consumption (ELEC), development of agrarian practices (AGR), and waste generation (WSG). For this, Table 1 presents variable definitions, variable codes, and information on data source used for each variable. Table 2 reports the descriptive statistical analysis of the selected variables. Moreover, subsection 3.1 explains the construction and specific definition of each selected variable.

1)
For this study, the impact on environmental pollution was characterized by CO2 emissions (ECO2) (kt) and CO2 emissions per capita (ECO2C) and was selected as the dependent variable in this study. The emissions of CO2 and other greenhouse gases (GHGs) into the atmosphere have instigated climate change and global warming, causing serious threats to the survival human beings, and living things. Meanwhile, its use as a proxy for atmospheric pollution could be justified by the fact that CO2 is the main (in quantity) greenhouse gas responsible for climate change and global warming. It is therefore the mutual responsibility of all nations in the globe to mitigate the emissions of GHGs.

2)
Population density (DPOP) is an important variable in environmental pollution. Indeed, a demographic surge in Jordan leads to an increase in food demand, medical and housing needs, and transportation, leading to overexploitation of the environment and therefore to a reduction in the stock of available resources and an increase in polluting emissions. It is therefore envisaged a positive sign between population density and CO2 emissions. 3) The economic development in this study was characterized by the gross domestic product (GDP) and gross domestic product per capita (GDPC) variables, which represents the average consumption of each person in the population. As each person's consumption increases, so does the environmental impact. A common indicator for measuring consumption is GDP per capita. Although GDP per capita measures output, it is often assumed that consumption increases as output increases. GDP per capita has increased steadily in recent years in the countries under review and is said to have increased the human impact on the environment.

4)
The concentration of the population is becoming increasingly strong in the districts of the capital Amman. This leads to a development of transport networks, an increase in household waste, a factor responsible for GHGs. The Middle East is urbanizing according to a singular process that weighs heavily on the natural environment of cities and destroys their ecological heritage. Urbanization (URB) could therefore be a factor affecting the environment.

5)
Energy consumption plays a significant role in the emissions of GHGs into the ambient atmosphere. For this study, the consumption of energy was measured by electricity consumption (ELEC) for each area.

6)
Population growth or pressure is a stimulus, or even a necessary prerequisite for agricultural progress. The increase in rural densities, the gradual scarcity of land in relation to the population can lead to more intensive use of land, requiring more anthropological activities, leading to increases in productivity and a general change in production structures.

7)
The development of agrarian practice (AGR) in donum refers to the amount of land that is being cultivated or used for agriculture in a given area. This leads to the destruction of green lands, therefore affecting environmental pollution. The expected sign of the coefficient related to the latter is positive. Waste include solid biomass, liquid biomass, biogas, industrial waste, and household waste and are measured as a percentage of total energy use. The production of household waste is increasing day by day, and either buried or burned, it generates direct GHG emissions. The same applies to industrial waste. It is therefore considered a positive sign between waste generation (WSG) and CO2 emissions.

Methodological Framework
The model used for this research is derived mainly from the STIRPAT model by (Dietz & Rosa, 1994). However, the use of models to describe and predict environmental impact based on socio-economic variables is not a recent undertaking. In fact, the STIRPAT model is a reformulation of an IPAT ecological model from the early 1970s (Krebs, 1982). And even before that, Duncan formulated a similar ecological model known as POET (1959). It is important to understand the basis of these models as well as their differences and similarities. The following section will delineate these models in chronological order leading to the current reformulated model proposed in this work.

STIRPAT
Most of the studies done on the analysis of the impact of population growth on the rate of carbon dioxide emissions use the STIRPAT model. As part of this study, this same model will be used with some modifications that will be justified later. The STIRPAT model is written in the following form: (1) where i is the index transversality representing the countries on which the study will be conducted. a,b,c and d are the parameters to be estimated and e represents the term of the error. I presents the total CO2 emission measured in tons. P denotes the entire population. A and T are affluence and technology respectively. The model retains the multiplicative form of accounting identity originally developed by (Dietz & Rosa, 1994).
In addition, Eq.
(1) includes a constant (a) to scale the model, exponents for the three factors (b, c, and d), indices (i) for I, P, A, T to indicate that these quantities vary from one unit of observation to another, and an error term (e ), the residual, to indicate the variation between observation units (York et al., 2003). T is generally included with e because there is no widely accepted operational definition or corresponding indicator of T. Therefore, in the STIRPAT model, the variable e incorporates T as residual, what remains between what is predicted and what is observed.
Based on the IPAT model, STIRPAT becomes an interdisciplinary model that links the natural sciences (an ecological accounting equation) to the social sciences (social science theory and methods) (Dietz & Rosa, 1994). In addition to allowing assumptions and allowing variation in the effects of factors on impact or impacts, STIRPAT can be expanded to include any other relevant variables such as political, social, and cultural factors (Dietz & Rosa, 1994). This can be done by disaggregating T, as T represents all factors other than P and A, as shown in the previous work of York et al. (York et al., 2003).

STIRPAT model specification
Several authors estimate the log-linear form of the model of Eq. (1) using panel data. In 2011, researchers used a non-cylinder panel data model to estimate a modified version of the STIRPAT model developed by Dietz and Rosa (Dietz & Rosa, 1994;Martínez-Zarzoso et al., 2011). Following these authors, in order to test whether the explanatory variables contained in the STIRPAT model influence the level of carbon dioxide emissions over time and countries, the specification of our model takes the following form : (2) where νi and ηt are the country-specific fixed effect and the temporal fixed effect, respectively. Indices i and t refer to different countries and numbers of years.
is the term of the error. Since the variables in Eq. (2) are logarithm, the coefficients in the model can be directly interpreted in terms of elasticity. The temporal effect ηt represents all variables common to all countries but which vary over time. νi represents the country-specific effect that is constant over time. The above model, although relatively simple, cannot be used because it has been shown that the errors resulting from the estimation of said model are self-correlated. This means that the effect of some independent variables is not instantaneous. Statistically, the autocovariance matrix of ordinary least squares estimation does not have its usual form. Several solutions to this problem exist. Researchers hypothesizes that model errors follow an autoregressive process of order 1 and uses generalized least squares to estimate it. However, this estimator requires strong assumptions such as the exogeneity of regressors. An alternative to this problem is to model Eq. (2) by adding the logarithm of the delayed explained variable of order 1 as regressor. This results in an ADL (Autoregressive Distributed Lag) model written in the following form: (4) Eq.
(3) has the advantage of being less restrictive than Eq. (2). The coefficient ζ measures only the present effect of the population's impact on CO2 emissions. β is the elasticity of the population in relation to carbon dioxide emissions. γ and δ are also elasticities. They are respectively the elasticity of gross domestic product per capita in relation to CO2 emissions and the elasticity of the share of manufacturing output in GDP per capita in relation to CO2 emissions. Finally, σ is the elasticity of the share of services in GDP per capita in relation to CO2 emissions.
Since Eq. (3) is dynamic, then the overall effect is given by the long-term elasticity given by β/(1 -ζ) obtained in the steady state. This precision is very important because we will see later that the estimate β of the Eq. (3) tends to be smaller than that of the model in Eq. (2). This is because β of model (3)

The Model
Firstly, the augmented Dickey-Fuller (ADF) panel unit root test was used in order to test for the order of integration in the energy consumption structure, energy intensity, economic growth, urbanization, and PM2.5 concentrations. If the variables were stationary at the same order difference, the Pedroni cointegration test would be employed to examine whether there was long-term relationship among these variables. Thirdly, if the variables were cointegrated, we would use panel Granger causality tests to explore casual links between the variables. The overall study framework is illustrated in Figure 1.

Panel Unit Root Test
Panel unit root tests are widely utilized and are considered more powerful than unit root tests based on individual time series (Chen et al., 2018;Han, 2020). In the current work, the ADF test, which belongs to Fisher-type tests and contains several individual unit root tests that are used to obtain a panel-specific result, was applied in order to test the order of integration of the variables.
The proposed following formula for the calculation of the entire panel was adopted from the previous studies (Choi, 2001;Maddala & Wu, 1999).
(4) P is the result of the combination of the p-value from the unit root test for each country/area/province. Note that − 2lnPi are a distribution χ 2 with 2 degrees of freedom. P follows a distribution of χ 2 with 2N degrees of freedom for Ti⟶∞ and finite N. Like the IPS test, the null hypothesis is the existence of unitary root of all series and the alternative is the stationarity of some variables. The Fisher test has many advantages over the IPS test. Its application does not necessarily require a cylinder panel. The test that will be employed in this study for the stationarity of our variables is the fishers because we have a non-cylinder panel.

Panel Cointegration Test
Two variables are co-integrated if there is at least one linear combination of them such that their combination gives a stationary variable, i.e., integrated of order 0. For example, if {xt} and {yt} are I(1) and co-integrated, then {Δxt}, {Δyt} and {xt + αyt} for a given α are stationary (I(0)).
Using this definition and the tests developed by several researchers, we will test a possible cointegration of our variables. In general, co-integration tests are performed on time series. However, authors such as Pedroni, Kao, and Westerlund have proposed co-integration tests that apply to longitudinal data. All test set "no cointegration" as the null hypothesis (Kao, 1999;Pedroni, 1999;Westerlund & Edgerton, 2007). The use of co-integration techniques in panel data makes it possible to test the presence of long-term relationships between integrated variables. One of the advantages of co-integration tests on panel data is the increase in terms of gaining the power of the test (Chen et al., 2018). In this work, both Pedroni and Kao's cointegration test were adopted (Kao, 1999;Pedroni, 1999).

Multicollinearity Test
Multicollinearity is an intrinsic characteristic of data. Its presence in the data can make it difficult to interpret regression coefficients, if not treated correctly. In this study, we use the correlation matrix to study a possible phenomenon of multicollinearity. The correlation matrix consists of a two-to-two analysis of correlations between explanatory variables. It can be considered that obtaining correlation coefficients greater than 0.5 is indicative of a problem of multicollinearity between the variables concerned (Bourmont, 2012).

FMOLS and DOLS Estimation Methods
Since the variables are co-integrated, to identify the tests on the cointegration vectors, it is essential to apply an efficient estimation method. At this point, several methods stand out: the Fully Modified Ordinary Least Squares (FMOLS) method used by Pedroni, the Dynamic Ordinary Least Squares (DOLS) method, and the Generalized Method of Moments (GMM). For instance, Kao and Chiang investigated the finite sample properties of MCO, FMOLS and DOLS estimators, point out that the MCO estimator suffers from a significant bias problem and that the FMOLS estimator does not substantially improve the MCO estimator. They then conclude in terms of the superiority of the DOLS estimator in estimating cointegration relationships on panel data. In addition, they showed in the case of panel data, that the FMOLS method and the DOLS method led to asymptotically distributed estimators according to a reduced centered normal distribution (Kao & Chiang, 2000).
The use of ordinary least squares on co-integrated panel data will result in biased estimators because the estimators will depend on the nuisance parameters associated with the dynamic nature of the model (Pedroni, 1999). Hence, according to Pedroni, we cannot use the MCO provided that we have exogenous regressors and a homogeneous dynamic between the individuals of the panel. Since these conditions are not satisfied with our data, we will then use the FMOLS method which not only considers a possible existence of a serial correlation but solves the problem of endogeneity of regressors. Finally, we use the DOLS method to also conduct the analyses. Comments on these different results will be discussed in the results and discussion section.

Impact of Urban Sprawl on the Agricultural Land And Energy Consumption in Amman
The capital city of Jordan, Amman, has witnessed significant growth in its population and economy over the past decades, due to the huge influx of immigrants and improved way of life, rise in job opportunities, etc. (Alnsour, 2016). As a consequence of this expansion and growth in the economy and population, coupled with poor urban planning and management, it has affected the city of Amman in a negative way. Herein, we will discuss the direct effect of this phenomenon on the consumption of energy and change in agricultural land in Amman. For this, data was obtained from the Department of Statistics for the period between 2008 to 2020, to show how urban sprawl has influenced these two parameters.

RESULTS AND DISCUSSION
For the analysis of the statistical properties of the variables, first, we will test the process of estimating the models by studying the different series. After that, a stationarity test will be performed, then an examination of the cointegration of the series will be implemented. Finally, a multicollinearity analysis test will be performed.

Multicollinearity Test
The multicollinearity problem is almost certain for correlation coefficients greater than 0.8 (Gujarati, 2003). Examination of the correlation coefficients between the different explanatory variables shows that they are generally low except for the variables URB and DPOP, GDP and ELEC, GDPC and GDP, for which the correlation coefficient is between 0.60 and 0.76. This result confirms that there is no risk of multicollinearity between the explanatory variables.

Panel unit root test results
The results in Table 3 show that at the 5% threshold, the null hypothesis that confirms the presence of unit root cannot be rejected even if we include the temporal trend and consider a single delay because all p-values are greater than 0.05. It is therefore concluded that the variables studied are stationary for Amman, which constitutes the database. The inclusion of a trend does not fundamentally change the results with the exception of CO2 emissions per capita, which becomes stationary. In addition, it is also noted that the GDP variable, which was 13 previously observed before the trend was introduced. It is then necessary to proceed to the first difference of this variables and redo the test. The results from the second test are presented in Table 4. As can be seen, the variables are stationary after taking the first difference. Whether it is the addition of a trend or not, the variables always remain stationary except for the GDP variable. They are therefore integrated of order 1. A regression of one of these variables over the others could lead to "fallacious" results. To overcome this disadvantage, it is necessary to do a cointegration test to check if the variables are co-integrated. If they are, then an Error Correction Model (ECM) will be used for the estimates, because when variables are cointegrated, the ECM model is the most suitable.

Pedroni Cointegration test Results
The Pedroni cointegration test was carried out for all the variables and the results are presented in Table 6. The two statistics considered, in the case where the dependent variable chosen is the per capita CO2 emission rate, are in favor of the existence of a long-term relationship between carbon dioxide and the other variables. On the other hand, when we look at the CO2 emission rate as a variable of interest, only one of the statistics works in favor of cointegration between all the variables. Better still, the other statistic weakly rejects the absence of co-integration. It will therefore be concluded that the variables are cointegrated and an errorcorrected model will be used to estimate this long-term relationship.

Kao's cointegration test results
Kao's cointegration test is more edifying about the existence of a long-term relationship between the different variables in the study. Indeed, all statistics reject the null hypothesis of non-cointegration of the variables considered in the test. It can therefore be concluded that there is co-integration between the variables for the cases of the two selected dependent variables. In the light of the two established tests, it is clear that our variables are co-integrated. As a result, there is a long-term relationship between CO2 emissions and each of the explanatory variables involved.

Results of the FMOLS and DOLS Estimation
The purpose of this section is to study and evaluate the long-term relationship between CO2 emissions and selected explanatory variables for their relevance in the city of Amman. In terms of CO2 emissions per capita, the determinants are waste generation (LnWSG) and economic development, in terms of gross-domestic product (LnGDP). Indeed, these two independent/control variables are significant at the 5% threshold. The other variables are not significant at the same threshold of significance. In addition, since the variables are in logarithms, then we have elasticities. On the other hand, the results of Table 8 show that 100% of the logarithm of CO2 emissions is explained by logarithms of GDP and WSG. When it comes to CO2 emissions per capita, waste generation explain the entire variability of the phenomenon. Population density is not significant for DOLS regardless of the dependent variable studied. These findings were in agreement with other studies (Chen et al., 2018).The same applies to population, electricity consumption, agricultural production, and GDP per capita. On the other hand, GDP significantly and positively influences CO2 emissions. However, waste generation influence on CO2 emissions is negative when considering the DOLS method. It is worth mentioning that some studies indeed failed to confirm that waste generation can directly increase CO2 emissions (Magazzino et al., 2021).  The 1% increase in GDP leads to an increase in the levels of CO2 emissions of more than 0.4%. This can be explained by the fact that an increase in economic development will require an increase in energy consumption, such that for the transport of people and goods. Indeed, in Jordan and the countries of the Middle East, the urban public transport of people is more developed and the more the income increases, the more people tend to resort to urban transport with buses or and private vehicles. The economic development will lead to an increase in demand for such services. In order to meet the additional demand, environmental resources will be overexploited and thus causing degradation of the environment.
Another explanation for this could be an intensification of the level of industrialization of Middle Eastern countries mainly large cities (in this case Amman in Jordan), and therefore a strong release of CO2 into the atmosphere. On the other hand, the richer households are, they acquire more household appliances, thus increasing the rate of energy consumption. This result of carbon Dioxide-Gross Domestic Product elasticity was found by other researchers such as Dietz and Rosa (1994) who used different approaches (Dietz & Rosa, 1994).
The negative sign of the parameter associated with the logarithm of waste generation (LnWSG) attracts our attention singularly. In all countries of the Middle East, when the generation of waste increased, the emission of CO2 increased. By controlling for all other variables, in the long term, a 1% increase in waste generation leads to a 0.11% increase in the CO2 emission rate. One of the explanations is the implementation of resolutions related to the preservation of the environment. These include the polluter pays principle. This is a principle that stems from the ethics of responsibility, which consists in having each economic actor consider the negative externalities of its activity. Encouraging recycling is also an environmental policy that reduces CO2 emissions in these countries. It should be noted that although the method used is different from that of Dietz and Rosa (Dietz & Rosa, 1994), we find the same nature of the cause-and-effect relationship. Other studies, on the other hand, have found a positive sign in the case of OECD countries. It should also be pointed out that countries whose industrial sector in renewable energy supply or which pursue a policy of preserving the ecosystem have a high rate of carbon dioxide emissions compared to those with the largest share of fossil fuel industries. The findings of this research are similar with those obtained from other studies (Hanif, 2018a;Magazzino et al., 2021).
Due to the rapidly expanding economies in the region under study, including Jordan, there is a major increase in the consumption of fossil fuels to meet the rising demand for energy, which results in the production of more waste in the form of CO2 and other GHGs. According to the findings, developing economies like Jordan could quickly run out of natural resources if they continue to use non-renewable energy to meet their energy needs. As a result, it is important that these economies move to alternate energy sources. The main causes of the increased consumption of fossil fuels in these emerging nations, which has a negative impact on environmental quality, are their low cost (in comparison to renewable energy sources) and the use of oil-based technology to generate goods (Valdes et al, 2022;Hanif, 2018b). Although urbanization raises living standards, which benefits the general welfare, but it also presents severe environmental difficulties, including deteriorating of public health, urban pollution, and issues with solid waste management. The findings of this study confirm that poor urban sprawl management can exacerbate pervasive environmental and economic problems.

Impact of Urban Sprawl on the Agricultural Land and Energy Consumption in Amman
The impact of urban expansion and growth and their relationship on the change in the area of agricultural land and on the consumption of energy was investigated in this study as well. As shown in Figure 2. it is noticed that there was a notable increase in the agricultural area land from the period 2008 until 2020 (< 40%). However, this increase was not constant throughout this period, and there was some fluctuation notable during that period. It is noteworthy to mention that the population growth during that period did indeed increase the demand for agricultural land, due to the increased population, and therefore, increased demand for food (Alnsour, 2016). A peak was noticed in 2015, it is worth to mention that during that period, this the peak in agricultural land could be a result of the influx of immigrants from the Syrian war, which demanded further necessity for agricultural land (Alnsour, 2016;Hammad, 2017). Nevertheless, Figure 3. shows the effect of urban sprawl and population growth in Amman on the consumption of energy. As depicted in the Figure, it can be noted that there was a constant and gradual increase in the consumption of energy from the year 2008 until it reached a peak in 2014. This gradual increase in energy consumption per capita is also related to the influx of immigrants and growth of the population each year, which was at an average of 2.2 % annually. However, from 2014 until 2020, there was a notable decrease in energy consumption in Amman city. This decrease might be related to the increased use of renewable energy among the population, and the increased prices of energy, which in turn, forces the consumers to reduce they energy consumption to save on the utilities, as well as improvements in the environmental regulations (Han, 2020). Regardless of that, although there was a decrease in energy consumption, but still, this is a serious issue that can further affect the quality of air, emitting GHGs into the atmosphere, and if there is no future plans for reducing the consumption of energy, this can increase the environmental pollution associated with urban sprawl and energy consumption.

CONCLUSIONS
This research study employs panel data from 2008 to 2020 obtained from the Jordanian Department of Statistics and the World Bank, from the perspective of urban sprawl in order to investigate the impact of the relationship between economic development, urbanization, development of agrarian practices, energy consumption, and waste generation on the environmental degradation in terms of CO2 emission. The current study employed the STIRPAT model, and the panel unit root test, as well as the panel cointegration test to investigate the data. A multicollinearity test was also utilized. For the estimation methods, the FMOLS and DOLS procedures were employed in this study. Meanwhile, the major limitation of this study was the acquiring of data, as data belonging to before the year 2008 for all variables was difficult to obtain.
Despite the fact that prior studies have demonstrated relationships between the structure of energy consumption, energy intensity, urbanization, and the rates of economic growth in a given country and the CO2 concentrations it experiences, the majority of these studies were carried out within national boundaries, ignoring the significant effects of a country's income level on estimating CO2 concentrations. This article sought to fill this research gap by addressing the casual relationships between energy consumption, urbanization, economic development, waste generation, development of agricultural practices, and CO2 concentrations were in the capital city of a developing country, Amman, Jordan.
The obtained results from the conducted analysis showed that the increase in economic development (GDP) did in fact increased CO2 emissions. This can be explained by the fact that an increase the GDP is often followed by an increase in energy consumption, transportation, and increase demand for additional products (i.e., food). Eventually, in order to meet these requirements, the environmental resources can be consumed and deteriorated, resulting in negative effects on the environment. Nevertheless, the increase in economic development, especially in developing countries, would usually require households to buy and consume more electrical appliances, thus, increasing energy consumption, leading to increased CO2 emissions. Therefore, we conclude that there is a long-term relationship between economic development and CO2 emissions. Meanwhile, generation of waste showed to negatively affect the emissions of CO2 into the environment, but not significantly. On the other hand, the results obtained indicated that the control variable, population density did not have significant effect on the CO2 emissions.

LIMITATIONS AND FUTURE RESEARCH RECOMMENDATIONS
This study has some limitations that can help define future research scope in this area of study. One of the study's limitations is the difficulty in obtaining accurate data on the variables used. The data was obtained from the Jordanian Department of Statistics and the World Bank, but only from 2008 to 2020. Data prior to this point were unavailable, limiting the scope of the analysis. Furthermore, the study used GDP and CO2 emissions data at the country level to represent Amman, which may have impacted the accuracy of the findings. In order to have a more complete understanding of the factors that contribute to environmental degradation in Amman, future studies should use more precise and up-to-date city-level data.
Its is recommended to conduct more research to better understand the distinct factors, such as regional policies, patterns of urbanization, and infrastructure development, that affect environmental pollution, agriculture, and energy consumption in Amman City. Consider incorporating other relevant data sources, such as satellite imagery or land use and land cover maps, to provide a more comprehensive understanding of the impacts of urban sprawl in the city.