The adopted methodology is discussed in this section. The proposed approach is advantageous in several aspects. Panel data analysis, particularly through the utilize of panel ordinary least squares (OLS) regression, offers several advantages in econometric research controlling for unobserved heterogeneity, enhanced variability and efficiency of estimation, reduced multicollinearity and panel data enables the identification and measurement of effects that are not detectable in pure cross-sectional or time-series data, such as individual behaviors or specific temporal trconcludes.
Theoretical framework
The theoretical foundation of this study is based on several economic and financial theories that explain the relationship between tourism, globalization, energy consumption, and economic growth. The key theories underpinning this research are as follows.
Tourism-led growth hypothesis
The tourism-led growth hypothesis (TLGH) suggests that an increase in international tourist arrivals positively contributes to economic growth by boosting employment, investment, and foreign exalter earnings (Raifu and Afolabi 2024). This study incorporates tourist arrivals (TA) as a key indepconcludeent variable to test its influence on economic output and growth.
Globalization and economic development theory
The globalization index (KOFGI) measures the extent of economic, political, and social integration with the global economy. According to globalization theories, increased international trade, foreign investments, and cross-border collaborations can stimulate economic growth (Lee 2021). However, some scholars argue that globalization may also introduce inefficiencies, market volatility, and economic inequalities, which justifies investigating its impact on this model.
Energy-growth nexus
The relationship between energy consumption and economic growth is well-established in economic literature. The Fossil Fuel Energy Consumption (FFEC) variable is included to analyze whether energy consumption is a driver of economic performance. The conservation hypothesis suggests that energy consumption follows economic growth, while the growth hypothesis proposes that energy consumption directly influences economic productivity.
Endogenous growth theory
The Endogenous Growth Theory explains how economic growth is primarily driven by internal factors such as human capital, innovation, and policy decisions. The economic output taken as GDP per capita (Y) variables aligns with this theory, emphasizing that sustained investments in economic infrastructure and technological advancements lead to long-term economic development (Leal Filho et al. 2021).
Random vs. resolveed effects in panel data
The study employs both resolveed effects (FE) and random effects (RE) models to assess variable relationships while controlling for unobserved heterogeneity. The Hausman test is applied to determine the most suitable econometric approach, ensuring the robustness of the results.
Dataset
To perform this study, the data is taken from different websites for 1998-2024 of the most visited nations of the world. The ‘globalization’ index is collected from the ‘Swiss Federal Institute of Technology’ while the data of other variables are taken from the ‘WDI World Bank database’. The data is taken for 1998-2024 of the most visited nations of the world. Short definitions and descriptions of variables and their indicators are given in the following subsections.
International Tourist Arrival
The quantity of vacationers who travel to a counattempt other than their typical residency, yet outside their standard environmental elements, is alluded to as worldwide sightseers’ appearance (TA). This study intermediary global traveler appearance for the travel indusattempt (Akadiri et al. 2019; Badulescu and S. 2021) to dispense with the possibility of experiencing multiple collinearity issues, especially when vacationer receipt has been utilized to evaluate the travel indusattempt. World Development Indicators (WDI) insights on unfamiliar guest appearances are utilized and worlddata.info.
Globalization Index
We utilize the recently announced globalization index developed by (Dreher 2006) and modified by Gygli. This index’s composition takes into account a variety of critical metrics, including economic, social, and political globalization.
GDP per capita
This study utilizes proxy GDP per capita (Y) (in current US dollars) for economic growth (Akadiri et al. 2019). GDP per capita is computed by dividing the GDP by midyear population. GDP is determined as the sum of entire resident producers’ contributions to the economy as gross value, minus any subsidies not included in product value plus any product taxes. It is calculated without any assumptions, and the numbers are in current US dollars. GDP is a metric for economic growth. WDI statistics on real GDP per capita are utilized.
Fossil fuel energy consumption
A strong free factor that is intently affected by environmental quality is fuel. Non-renewable energy sources include coal, oil, and petroleum gas items. The information of this variable is taken from WDI as ’Petroleum derivative energy ’utilization (% of aggregate). The fuel ’utilization intermediary utilized in this study is petroleum product energy utilization becautilize of the shortage of this sort of fuel.
Model specification & functional equation
For a selection of scenarios, four models based on carbon emissions functions would have been developed:
$${\rm{Environmental}}\, {\rm{Quality}}=f({\rm{Tourism}},\,{\rm{Globalization}},\,{\rm{Economic}}\, {\rm{Growth}},{\rm{Energy}}\, {\rm{Consumption}}\,,\ldots )$$
(1)
The CO_2 emissions function can be expressed as:
$${\rm{CO}}_{2}=f({\rm{ta}},\,y,\,{\rm{kofgi}},\,{\rm{ffec}}\,)$$
(2)
With a functional form:
$${\text{CO}}_{2}={\alpha }_{it}{\text{ta}}_{it}^{{\beta }_{1}}{y}_{it}^{{\beta }_{2}}{\text{kofgi}}_{it}^{{\beta }_{3}}{\text{ffec}\,}_{it}^{{\beta }_{4}}$$
(3)
where subscript i depicts the panel cross-section (i = 1, 2, …, 10), representing the top 10 visited economies: 10-UK, 9-Germany, 8-Thailand, 7-Mexico, 6-Turkey, 5-Italy, 4-China, 3-USA, 2-Spain, and 1-France. Tourist arrival, GDP, globalization index, and fossil fuel energy consumption are indicated by ta, Y, kofgi, and ffec, respectively.
The subscript t represents the time period 1995–2021.
The log-linearized model is given by:
$${\text{CO}}_{2}={\alpha }_{it}+{\beta }_{1}\ln ({\text{ta}}_{it})+{\beta }_{2}\ln ({y}_{it})+{\beta }_{3}\ln ({\text{kofgi}}_{it})+{\beta }_{4}\ln ({\text{ffec}}_{it})$$
(4)
The model assumes a linear relationship between the indepconcludeent variables and the depconcludeent variable CO2 The indepconcludeent variables should not be perfectly correlated with each other to ensure meaningful coefficient estimates. The variance of the error term should be constant across observations. The residuals should not be correlated over time, which is particularly important in panel data analysis. The indepconcludeent variables should not be correlated with the error term to avoid biased estimates. The error term should be normally distributed for valid hypothesis testing.
Concerning the panel data considerations, if a resolveed effects model is utilized, individual-specific effects then parameters must be correlated with regressors, while in a random effects model, they must be uncorrelated.
Descriptive statistics
The ‘Swiss Federal Institute of Technology’ is utilized for globalization index collection while data on other variables has been taken from WDI. Eviews 9 is utilized to analyze panel data. Descriptive statistical analysis is represented in Table 3, the variables are available as a spread of applied statistical units. The international Tourism variable is measured in the number of arrivals, GDP per capita is measured in current international $, environmental quality is measured CO2 in metric tons/capita, whereas Globalization is measured in globalization index Kofgi. The variables measure mass into composite indicators, they required to be standardized or normalized (Badulescu and S. 2021). In this study, a natural logarithm measure is utilized with that tourism and economic growth variables are reborn to a typical scale that assumes a “normal” distribution (i.e. a median of zero and a typical deviation of 1). Table 3 displays the basic statistics of the five variables. Specifically, it contains the standard deviation, the mean, and minimum and maximum values of the variables utilized in this study.
Panel unit root test
To start analysis on this panel data, it is necessary to find the unit root of all the series of this panel data. We direct unit pull tests for the four factors to check regardless of whether these four picked factors are resolveed. As per new exploration, unit pull tests for board sets have more noteworthy strength than unit pull tests for individual time series. E-Views might figure one of the five board unit root tests (LLC) curtailed as (Breitung 2001; Im & Pesaran 2003; Levin et al. 2002), Fishers’ tests in light of the ADF upgraded Dicky Fuller test, and PP (Phillips and Perron) tests. Hypothetical assist for these tests is given by (Ateljevic 2001; Maddala and Wu 1999; Munday 2007).
“Panel Unit Root” is the normal name for panel information investigation. In principle, these tests are only numerous relapse conditions with unit attaches applied to the designs of board information series. At the point when individual time series are inspected utilizing increased unit root, a solitary condition with a few factors that are different slacked terms arise. At the point when a board information series is attempted for stationarity or unit root, the presence of cross-sectional sources creates the test yield various series instead of a solitary condition. Table 4 depicts the test results.
Since testing methods depict, at the level, these 4 variables are stationary. Consequently, in the next analysis, this study utilizes the level of stationarity while conducting the process of regression. Therefore, the regression results imply that tourism and the globalization index are indepconcludeent factors influencing the emission of carbon. All variables, including the depconcludeent variable CO2 emissions and indepconcludeent variables TA, GDP, FFEC, and KOFGI, have been stationary at the level.
The Levin, Lin & Chu t* method assumes:
Null Hypothesis (H0): Unit root (assumes a common unit root process).
By considering the following AR(1) process for panel data:
$${y}_{i,t}={\rho }_{i}{y}_{i,t-1}+{\epsilon }_{i,t}+{X}_{i,t}\beta$$
(5)
The ID code for each counattempt is represented by i = 1,2,…,N in the equation above as 10-UK, 9-Germany, 8-Thailand, 7-Mexico, 6-Turkey, 5-Italy, 4-China, 3-USA, 2-Spain, and 1-France, depicts cross-sectional series or units, and t = 1998, …, 2024 (demonstrates time period). An exogenous variable is denoted by Xi in equation (5), and resolveed effects and individual trconcludes are also included in the exogenous variables. To demonstrate autoregressive coefficients, i has been added to the equation, and there is an i,t term for mistakes that are taken into consideration.
The above panel unit root results give a clear indication for the methodology, as all series are stationary at level, so panel least squares might be applicable.
