1 Introduction

Over the past 2 decades, the statistics of the World Tourism Organization’s annual report indicate that many countries have observed a significant growth in tourism as these countries have adopted tourism promotion policies. In 2018, the number of international tourists increased by 5%, exceeding the 1.4 billion mark. The export revenue of the tourism industry has grown to 1.7 trillion US dollars, which is achieved two years ahead of the prediction of the United Nations World Tourism Organization. Tourism development creates more job opportunities and becomes a catalyst for innovation and entrepreneurship, which in turn promotes global power for economic growth and development (World Tourism Organization (UNWTO), 2019). The emergence of a tourism-led growth hypothesis has found that the inbound tourism development may have a mutual impact on the gross domestic product (GDP) (Balaguer & Cantavella-Jorda, 2002). Inbound tourism plays the role of an important determinant in the long-term GDP growth (Adnan Hye & Ali Khan, 2013; Brida et al., 2016; Dogru & Bulut, 2018; Hatemi et al., 2014, 2018; Katircioglu, 2009; Nowak et al., 2007). For the governments of the major inbound tourism countries, determining the effectiveness of tourism companies is of utmost importance. Optimizing the allocation of resources to promote the development of tourism thereby takes advantage of the multiple policies of tourism.

Although tourism development shows a significant positive impact on the GDP, it may have an adverse impact on the environment. Many tourist activities require large amounts of electricity generated from fossil fuels or coal, natural gas, or oil. Even the logistics business is generated during the travel process, and special logistics services and transportation-related logistics infrastructure cause serious environmental problems due to poor logistics infrastructure (Khan et al., 2019; Khan et al., 2020b; Rehman Khan & Yu, 2020). Therefore, a large amount of CO2 is emitted. For example, in 2016, the global transport and freight on CO2 emissions were estimated to a total of 7.23 billion tons, of which the passenger transport generated 64% of carbon dioxide (World Tourism Organization (UNWTO) & International Transport Forum, 2019). Transportation, accommodation, and other tourism-related activities can have serious adverse effects on the environment (Jones & Munday, 2007). However, with changes in travel policies and methods, CO2 emissions can be significantly reduced with the sustainable development of the environment (Scott, 2011).

After the United Nations Climate Change Conference (UNCCC) in Denmark, the low-carbon economy may become the “fourth industrial revolution.” The new direction of the tourism policy is implemented with a low-carbon economy or a new low-emission technology, which helps reduce CO2 emissions and continues to increase the number of tourists (Tang et al., 2011a). The low-carbon economy or the sustainable development of the environment requires the use of renewable energy in order to replace non-renewable energy because renewable energy can effectively reduce environmental pollution (Khan et al., 2020a; Khan et al., 2020a, b, c; Nathaniel & Khan, 2020). Effective air control and more energy-efficient restrictions on the number of flights or policies can reduce CO2 emissions. Depending on the development of a low-carbon economy, there may be an uncertain relationship between tourism and carbon dioxide.

The relationship between tourism and carbon emissions in the top five countries (France, Spain, United States, China, and Italy) with the most inbound tourism is presented herein. Since the top five countries with the highest number of inbound tourists receive 1/4 of worldwide arrivals as ranked by the World Tourism Organization as of 2018 (UNWTO, 2019), they earn 444 billion US Dollars in the same year. Especially, France, Spain, United States, China, and Italy are also the top countries that emit carbon dioxide from fossil fuels. For example, the USA, China, France, and Spain are the countries that contribute to the highest carbon dioxide fossil fuel emissions. It is more practical to investigate the relationship between tourism development and carbon emissions in these countries. The tourism environmental policies of countries with a large tourist population show a more profound and influential impact on the global tourism environment. Figure 1 shows an increase in the number of inbound tourists from the top five inbound tourism countries. The trend of the inbound population growth in France is relatively flat, and the inbound population growth in the USA, Spain, China, and Italy has grown significantly. However, in countries other than China, inverse U-shaped carbon emissions can be observed, which first increase and then decrease. Among the top five countries, only in China, the long-term trend is the same as that of the inbound growth. It can be observed that the changes in the tourism industry and the sustainable development of the environment may vary between different countries and policies, and it is considered an important issue that requires to be further examined.

Fig. 1
figure 1

CO2 emissions and tourism of top five countries

Full size image

Most of the studies consider multi-country research and generally use the technique of penal data for analysis (Akadiri & Akadiri, 2019; Dogan et al., 2017; Fethi & Senyucel, 2020; Koçak et al., 2020; Lee & Brahmasrene, 2013; Paramati et al., 2017; Saha & Yap, 2014; Zaman et al., 2016). In the research of macroeconomic variables with small sample attributes, the panel analysis is usually used for solving the drawbacks. This research focuses on the country-specific studies with time-series data in order to examine the relationship between tourism and CO2 emissions so that considering the bootstrapping procedure in the ARDL model has a larger size and more power properties with small sample than the ARDL-bound test proposed by Pesaran et al. (2001). Therefore, this research explores the country-specific studies that show the relationship between inbound tourism and the economy and considers the role of CO2 emissions in various countries. The key objectives of this research are the following. (1) To examine the sustainable impact of inbound tourism on different environmental qualities and economic growth because the top five major tourism countries guide global sustainable tourism policies. (2) To find the long-run relationship among CO2 emissions, inbound tourism development, and economic growth by using a specific bootstrap ARDL model developed by McNown et al. (2018), as well as the short-run relationship. This is the first application of the newly proposed econometric research method that investigates the relationship between inbound tourism development, economic growth, and carbon dioxide emissions. (3) To propose the corresponding policy enlightenment according to the difference.

This paper is organized as follows. In Sect. 2, we focus on the critical reviews of the relationship between CO2 emissions from tourism and the economic growth in the literature. In Sect. 3, we introduce data and framework models. In Sect. 4, we discuss the results and findings. Finally, in Sect. 5, we present conclusions and discuss the future course of action.

2 Literature review

So far, given the potential connection, the relationship between socioeconomic factors such as the environment and tourism development has provided useful suggestions for scholars and policymakers. Therefore, in order to understand the role of this study in past studies, the literature discusses the tri-variate relationship among tourism development, CO2 emissions, and economic growth.

2.1 CO2 emissions and economic growth

In the past, most research studies in this area have focused on how the environment impacts economic growth and on the reverse impact. It primarily analyzes the Environmental Kuznets Curve (EKC) hypothesis, in which the relationship between the environment and economic growth demonstrates an inverse U shape. Akadiri et al. (2019) and Grossman and Krueger (1995) first proposed that the initial income of a country’s economy is positively correlated with environmental pollution, but when income grew to a certain threshold, pollution begins to decrease. Many research works showed the existence of the EKC hypothesis (Akbostancı et al., 2009; Bimonte, 2002; Lise, 2006). Selden and Song (1994) have examined the EKC hypothesis by using the penal data analysis that provides consistent results.

The economic growth and energy consumption affect the cointegration and causality of CO2 emissions by using the bootstrap ARDL test for G7 countries. Cai et al. (2018) reexamined and found that the economic growth in France, Italy, and the USA shows no significant long-run and short-run relationships with CO2 emissions and vice versa. The study investigated the role of energy, real income, and globalization in assessing the environment and found that increasing globalization reduced per capita carbon dioxide emissions in Italy (Saint Akadiri et al., 2019a, b, c). The report on China using the ARDL-bound test indicates that the economic growth shows a positive correlation and significant CO2 emissions in the cointegration and error correction method (Alam et al., 2016). The existence of nonlinear relationships shows the inverted U-shaped curve linkage between CO2 emissions and income in Spain (Esteve & Tamarit, 2012). However, in other countries, the report of Turkey using the ARDL-bound test shows a two-way causal relationship between income and CO2 emissions (Halicioglu, 2009). By supporting an inverted U-shaped curve in the EKC hypothesis for Malaysia (Saboori et al., 2012; Jebli et al., 2016) establish long- and short-run relationships between CO2 emissions and economic growth and verify the EKC hypothesis for OECD countries. It is clear that there is a distinct functional relationship between economic factors and environmental changes as compared to the previous literature.

2.2 Tourism development and economic growth

In the past, many studies have shown that the inbound tourism development shows a long-run and stable impact on the region’s economic growth (Chatziantoniou et al., 2013; Crouch, 1994; Ghartey, 2013; Husein & Kara, 2011). The inbound tourism development, which represents the source of income and consumption, is considered a non-standard type of export (Balaguer & Cantavella-Jorda, 2002). Previously, several researchers have studied the correlation between economic growth and tourism development (Cortes-Jimenez et al., 2011; Durbarry, 2004; Oh, 2005). These results show a tourism-led growth hypothesis. Balaguer and Cantavella-Jorda (2002) observed that a cointegration relationship exists in the development of inbound tourism and economic growth in Spain from 1975 to 1997. Income from tourism shows a positive impact on the economic growth of Spain. In contrast to the research of Balaguer and Cantavella-Jordá (2002) and Oh (2005) indicated that there was no long-run relationship between these two variables. He further reported that the one-way cause and effect relationship shows the short-run impact on the economic growth and the development of tourism, but the tourism-led economic growth was not experienced in South Korea. Examining the relationship between tourism and economic growth in the small island developing states, the empirical results show that tourism promotes economic growth (Roudi et al., 2019). Deng et al. (2014) adopted the panel data analysis in 30 provinces of China from 1987 to 2010 and indicated that there had been a slow reverse impact of the tourism-led growth in that period. Their results further showed that tourism activity decreased economic growth by forcing out human capital. In addition to the above economic indicators, the non-economic indicators, such as terrorism, regional situations, and geopolitical risk index, affect tourism and economic growth (Alola et al., 2020; Araña & León, 2008; Saha & Yap, 2014; Saint Akadiri et al., 2020; Uzuner et al., 2020). These past studies have proved that the development of inbound tourism shows different relations with different countries, such as the long-run and short-run economic growth relations.

Tourism development may also show a negative impact on regional development trends. Some studies have observed that more inbound tourism development may remove traditional local industries, such as fisheries, agriculture, and industry, and thus impact the national economic growth, which will turn human resources into tourism (Adams & Parmenter, 1995; McCool & Martin, 1994). Overtourism may have more serious consequences. The destruction of natural landscapes and cultural heritage may develop anti-tourism phenomena, thus resulting in a negative impact on the tourism industry (Kuščer & Mihalič, 2019; Seraphin et al., 2018).

2.3 Tourism development and CO2 emissions

Inbound tourism involves a lot of transportation, accommodation, and activities that depend on fossil fuels. However, fossil fuels have a great dependence on carbon emissions. Solarin (2014) reported a bidirectional impact on CO2 emissions and inbound tourism, and a positive impact was observed on CO2 emissions from the development of inbound tourism in Malaysia. The development of inbound tourism can cause serious pollution problems to the environment quality in less developed and developing countries (León et al., 2014). Similar results were observed in Turkey (Eyuboglu & Uzar, 2019) and Cyprus (Katircioglu et al., 2014). The Organization for Economic Co-operation and Development (OECD) investigated that environmental pollution will become a more serious problem because of tourism development (Dogan et al., 2017) and also reported the same results in EU and candidate countries (Dogan & Aslan, 2017). Bano et al. (2021) supported the tourism-led emission hypothesis for Pakistan. Most of the top 50 tourist countries should pay attention to environmental pollution caused by tourism. Countries such as Turkey, Thailand, Russia, Greece, Saudi Arabia, Macao, Indonesia, Brazil, Dominica, Philippines, Bulgaria, Tunisia, Egypt, Iran, Georgia, Hong Kong, India, and Malaysia must reduce carbon emissions and develop a sustainable environment (Fethi & Senyucel, 2020). Shi et al. (2019) explored different results in tourism development and CO2 emission because of different incomes in the countries. They adopted the panel data analysis for five different income groups that showed causality from inbound tourism to CO2 emission (Haseeb & Azam, 2020). A large number of international tourism countries show an important correlation with the degree of globalization. The globalization–tourism-induced EKC hypothesis has been proposed and proved to be significantly related to the environment. Therefore, the impact of globalization and international tourism contributes to environmental pollutions that are more internal (Akadiri, et al., 2020a, b; Uzuner et al., 2020; Saint Akadiri et al., 2019a, b, c; Saint Akadiri et al., 2019a, b, c). The results show that the impact of inbound tourism on CO2 emissions was positive and significant.

In contrast, this study investigates the relationship between inbound tourism and CO2 emissions from the panel data analysis of the European Union. The results show a cointegration (long-run) relationship between CO2 emissions, tourism, FDI, and economic growth, and the inbound tourism development shows a negative and significant impact on CO2 emissions (Lee & Brahmasrene, 2013). Katircioğlu (2014) examined how the inbound tourism development affects the CO2 emissions in long and short runs from the period 1971 to 2010. The results show that the impact of the development of inbound tourism on CO2 emissions is negative and significant in Singapore. Similar results indicate that in China inbound tourism shows a negative impact on CO2 emissions by using panel causality tests (Zhang & Gao, 2016). Ahmad et al. (2019) investigated the influence of tourism on environmental pollution in the low- and middle-income countries, namely, Indonesia, the Philippines, and Vietnam. The results further confirm that the development of inbound tourism shows a negative impact on environmental pollution in Vietnam; however, tourism development reduces the environmental quality of Indonesia and the Philippines.

The existing literature confirms that the improved environmental quality shows the contribution of a country to the environment and its economic progress. The study primarily showed that the environmental degradation caused by a large number of tourists increases carbon dioxide emissions. However, the current research results do not completely agree with the fact that tourism causes environmental pollution. On the contrary, there is evidence that tourism promotes environmental quality. Tourism development strategies provide different results in different countries.

3 Data and empirical

3.1 Data

Annual data are collected from the period 1995 to 2018 and included five countries. The countries considered were France, the USA, Spain, China, and Italy because these countries were the top five destinations of international tourist arrivals in the world (UNWTO, 2018). CO2 emissions (million tons of carbon dioxide), tourism development (international tourism and number of arrivals), and economic growth (real GDP and current US dollars) were derived from the World Bank Open Data and BP Statistical Review of the World Energy. A natural logarithm form of the three variables was necessary to reduce the heteroscedasticity. The summary statistics of CO2 emissions, economic growth, and tourism development are presented in Table 1.

Table 1 Summary statistics
Full size table

Our empirical model examines the cointegration between three series in the top five tourist arrival countries (Akadiri et al., 2019; Balli et al., 2019; Eyuboglu & Uzar, 2019; Lee & Brahmasrene, 2013), which can be defined as follows:

$$\ln {\text{CO2}}_{t} = f\left( {\ln \,{\text{GDP}}_{t} ,~\ln \,{\text{TOU}}_{t} } \right)$$
(1)

where \({\mathrm{T}\mathrm{O}\mathrm{U}}_{t}\) represents the number of tourists in the country in year \(t\), \({\mathrm{G}\mathrm{D}\mathrm{P}}_{t}\) represents the real GDP, and \({\mathrm{C}\mathrm{O}}_{2t}\) represents carbon emissions in tons in year \(t\). The environmental changes caused by economic development and tourism development require a lot of attention, especially the impact of tourism as the development of the aviation industry has reached a new level in recent years. The examination of the impact of the above function (formula 1) on important tourism arrival countries shows the outcome of a good tourism policy.

3.2 Econometric methodology

Inbound tourism development, carbon dioxide and economic growth are the target variables of this study, and examine the cointegration relationship between them is being applied in our empirical method, which is termed as bootstrap auto regressive distributed lag (bootstrap ARDL) model McNown et al., (2018) for examining dynamics relationship. However, the ARDL model can be specified as follows:

$$\begin{aligned} {\text{lnCO2}}_{t} & = \alpha _{0} + \mathop \sum \limits_{{i = 1}}^{n} \alpha _{{1,i}} {\text{lnCO}}_{{2t - 1}} + \mathop \sum \limits_{{i = 0}}^{n} \alpha _{{2,i}} {\text{lnGDP}}_{{t - 1}} \\ & \quad + \mathop \sum \limits_{{i = 0}}^{n} \alpha _{{3,i}} {\text{lnTOU}}_{{t - 1}} + \mathop \sum \limits_{{j = 0}}^{m} \beta _{j} D_{{t,j}} + e_{t} \\ \end{aligned}$$
(2)

where \(t\) denotes the time period \(t = 1,2, \ldots ,T\), and \(i\) and \(j\) are index of lags \(i = 1,2, \ldots ,n\); \(j = 1,2, \ldots ,m\); \(D_{{t,l}}\) is a dummy variable and \(\beta\) is the coefficient of shot break variable. \(\alpha _{1} ,~\,\alpha _{2} ,~\,\,{\text{and~}}\,\alpha _{3}\) represents parameters for the coefficients of the lag of \({\text{lnCO2}}_{t} ,{\text{~lnGDP}}_{t}\) and \({\text{lnTOU}}_{t}\); \(e_{t}\) represents the normal distributed error term. This starting model was generated in an error correction from as follows:

$$\begin{aligned} \Delta {\text{lnCO2}}_{t} & = \alpha + \delta {\text{lnCO2}}_{{t - 1}} + \theta {\text{lnGDP}}_{{t - 1}} + \gamma {\text{lnTOU}}_{{t - 1}} \\ & \quad + \mathop \sum \limits_{{i = 1}}^{n} \psi _{i} \Delta {\text{lnCO2}}_{{t - i}} + \mathop \sum \limits_{{i = 1}}^{n} \varphi _{i} \Delta {\text{lnGDP}}_{{t - i}} + \mathop \sum \limits_{{i = 1}}^{n} \eta _{i} \Delta {\text{lnTOU}}_{{t - 1}} + \varepsilon _{t} \\ \end{aligned}$$
(3)

where \(\delta = 1 - \sum\nolimits_{{i = 0}}^{n} {\alpha _{{1,i}} }\); \(\theta = \sum\nolimits_{{i = 0}}^{n} {\alpha _{{2,i}} }\) and \(\gamma = \sum\nolimits_{{i = 0}}^{n} {\alpha _{{3,i}} }\). \(\psi _{i} ,~\varphi _{i}\) and \(\eta _{i}\) are function parameters in Eq. (2).

Pesaran et al. (2001) explain that there is a rejection of the null hypothesis, that is, has a cointegration relationship between variables that adopting F test (\(F_{x}\)) \(H_{0} :~\delta = \theta = \gamma = 0\) and t test (\(t\)) \(H_{0} :~\delta = 0\). Pesaran et al. (2001) ARDL-bound test chart critical values for \(F_{x}\) and \(t\) tests. There are the upper bound and lower bound computing in critical value, if the estimated value from the ARDL-bound model is greater than the upper threshold or less than the lower threshold, it indicates that they have cointegration in the model. McNown et al. (2018) suggest to add a new F test (\(F_{y}\)) \(H_{0} :~\theta = \gamma = 0\) to complement the cointegration relationship and estimate critical value for \({F}_{x}\), \(t\) and \({F}_{y}\) test using Monte Carlo simulations. Thus, the bootstrapping procedure has a more size and power properties.

The bootstrap ARDL test can define cointegration, non-cointegration, case 1 and case 2 by the results of \({F}_{x}\), \(t\) and \({F}_{y}\):

  • Cointegration: The all three null hypothesis was rejected in F test (\(F_{x}\)) \(H_{0} :~\delta = \theta = \gamma = 0\), t test (\(t\)) \(H_{0} :~\delta = 0\), and F test (\(F_{y}\)) \(H_{0} :~\theta = \gamma = 0\)

  • Non-cointegration: Do not reject the null hypotheses F test (\({F}_{x}\)), t test (\(t\)) or F test (\({F}_{y}\)).

  • Case 1: Degenerate Case 1 have to reject the null hypotheses F test (\({F}_{x}\)) and F test (\({F}_{y}\)) but no reject the null hypotheses t test (\(t\)).

  • Case 2: Degenerate Case 2 have to reject the null hypotheses F test (\({F}_{x}\)) and t test (\(t\)) but no reject the null hypotheses F test (\({F}_{y}\)).

Non-cointegration, case 1 and case 2 mean that they are no cointegration among the three variables. Pesaran et al. did not show for case 1 in the critical value, because they omitted to test F test (\({F}_{y}\))\(.\) In fact, bootstrap ARDL has been applied in many studies, such as: energy economy (Alhodiry et al., 2021; Ghazouani et al., 2020; Goh et al., 2017a, b; Pata & Aydin, 2020; Tong et al., 2020), environmental economy (Lin et al., 2018; Meirun et al., 2021; Pata, 2019; Wang et al., 2019; Yilanci et al., 2020), and international trade issues (Goh et al., 2017a, b; Nawaz et al., 2019; Saleem & Shabbir, 2020; Wu et al., 2020).

The running of the short-run relationship can be proposed by Granger causality, after directing for the cointegration estimating the bootstrap ARDL. When this test has existed cointegration among the variables, the short-run causality test has impact of GDP and inbound tourism development to CO2 emissions have to include the lagged level of the two independent variable and lagged differences on GDP and tourism development; that is, we estimate whether \(\theta =0\) and \({\varphi }_{i}=0\) or \(\gamma =0\) and \({\eta }_{i}=0\). However, if there is not existing cointegration from GDP and tourism development to CO2 emissions, then the Granger causality test only estimate whether \({\varphi }_{i}=0\) or \({\eta }_{i}=0\).

4 Main results and discussion

4.1 The unit root

The traditional ARDL shows that the all series the inclusion of both I (0) or I (1) time series in cointegration relationship. The bootstrap ARDL test also allows the modeling of variables with the 1 or less order of integration. In other words, at least, all series have to stationary after first differential. We then carefully checked the integration order which contains three variables for each country to understand the properties of each time series. To better understand how to check the unit root properties of three variables that the augmented Dickey–Fuller (Dickey & Fuller, 1981) (ADF) has used to test for with and without structural breaks. Table 2 shows ADF with intercept and trend unit root test results at level and first difference values for CO2 emissions, real GDP, and tourism development. Most, but not all, we indicated non-stationary conclusions in the level column. However, this shows that three variables are I (1) process, because these variables turn stationary in the first difference column. Table 2 shows the ADF unit roots with structural breaks, taking into account the unit breaks in the series of unit root tests. To this end, we introduced the ADF unit root test of Kim and Perron (2009), which contains an unknown structural break of the three series. Thus, it was suitable to report to examine the existence of cointegration among CO2 emissions, inbound tourism development, and economic growth.

Table 2 Univariate unit root tests
Full size table

4.2 Bootstrap ARDL model with structural breaks

After the integration order of the three variables is determined and tested, this paper applies the bootstrap ARDL model in order to examine whether there is a long-run effect on CO2 emissions, tourism development, and economic growth. The development of the bootstrap ARDL test has more advantages than the ARDL-bound test because the bootstrap ARDL testing approach discusses \({F}_{x}\), \(t\), or \({F}_{y}\). Table 2 presents the results of the bootstrap ARDL model that accommodated structural breaks by using dummy variables.

From Table 2, we find that the existence of the cointegration relationship in the model rejected the hypothesis of the \({F}_{y}\) test, \({F}_{x}\) test, and \(t\) test because both economic growth and tourism development are considered independent variables. The main results of these three tests show the presence of the bootstrap ARDL testing of cointegration at a significance level of 10%.

The significant empirical results are obtained for three critical valued tests when the GDP growth and tourism development are used as independent variables in the CO2 emission function in France, the USA, and China. Besides, we only find evidence of cointegration results in the USA when the economic growth is considered as a dependent variable. However, when CO2 is the dependent variable, we have failed to find cointegrated results for Spain and Italy.

The results of long-run coefficients are presented in Table 4. When the economic growth and tourism development increase by 1%, CO2 emissions decrease by 0.06% and 0.355% in the long run in France (Dogan & Aslan, 2017; Shi et al., 2019). In the long run, there is no growth-led emission because more wealth and income make people pay more attention to environmental protection in France. In China, we find that the effect of tourism development is positive and significant to CO2 emissions by 0.416%. It is evident from Zhang and Gao (2016) that support to tourism development would harm the environment in China. However, the empirical results have been confirmed by Eyuboglu and Uzar (2019), Katircioglu et al. (2014), and Shi et al. (2019). Tourism development indicating a statistically significant and positive effect on the environment proves that it has a crucial impact in the long term. This indicates that economic growth positively affects CO2 emissions with a coefficient of 0.212% in the USA, and tourism development negatively affects CO2 emissions (Lee & Brahmasrene, 2013; Shakouri et al., 2017; Shi et al., 2019; Zhang & Gao, 2016). In the USA, when the economic growth is considered as a dependent variable, we find the long-run relationship between CO2 emissions and tourism development as they show a negative and significant impact on the economic growth by 0.423% and 0.122%. These results have also been reported by Lee and Brahmasrene (2013).

Applying the cumulative sums of the recursive residuals (CUSUM test) and squares of recursive residuals (CUSUM of square tests), the robustness of the cointegration tests is studied (Brown et al., 1975), which is based on two different residual tests. The results of the CUSUM and CUSUM of square tests are found stable at a significance level of 5%, as shown in Table 3.

Table 3 Bootstrap ARDL model cointegration test results
Full size table

4.3 Short-run causality test

Furthermore, we also estimated the Granger causality based on ARDL model, as shown in Table 4. Our research shows that the development of inbound tourism stimulates the consumption of the inbound country and the commercial development of tourism. This shows a positive impact on the domestic economic growth and is found statistically significant, which supports the growth hypothesis of tourism dominated by France and China. However, in the USA, inbound tourism development shows a negative and significant impact on economic growth. In China, the impact of CO2 emissions is positively associated with economic growth, but it is negatively and significantly associated with the economic growth in Italy. The economic growth increases while increasing CO2 emissions in the USA, but CO2 emission is significantly reduced in China. We observed that the effect of the inbound tourism development could reduce CO2 emissions in the USA and Italy. In Spain, China, and Italy, the economic growth shows a negative and significant impact on the inbound tourism development. The relationship between CO2 emissions and inbound tourism development is positive and statistically significant in France.

Table 4 Long-run coefficient results
Full size table

4.4 Empirical discussion

This section examines the top five inbound tourism countries through the bootstrap ARDL model with structural breaks, analyzes the cointegration relationship between CO2 emissions, economic growth, and tourism development, and explains the short-run Granger causality. The long-run result shows that the economic growth and inbound tourism development are negative and the most significant contributors to CO2 emissions in France. It indicates that tourism development in France has promoted environmental sustainability (Bella, 2018). A low-carbon economy has developed in the tourism industry in order to reduce CO2 emissions (Lee & Brahmasrene, 2013).

The United Nations Climate Agreement monitors international flights by using carbon offset and the modification of the aircraft design in order to improve the fuel efficiency, route planning, development of biomass fuels, etc., at the COP 21 in Paris. The blueprint of French policymakers for the tourism industry encourages tourist groups to protect and understand the environment quality. The hotel industry in tourism can propose eco-certified services, such as wastewater and waste recycling, in order to encourage tourists to support green transportation and enhance the effectiveness of transportation, such as mass transportation instead of self-driving or car sharing. It also provides a green passport to promote tourists in order to collectively protect ecological diversity (Bella, 2018). The French government implements the United Nations sustainable development goals (SDGs), including 17 goals and 169 targets after 2014. Among them, policies are formulated and implemented in the tourism industry in order to monitor the impact of sustainable development on the creation of employment and the promotion of the sustainable tourism of local culture and products. Its objective is to ensure a sustainable consumption and production model.

When the economic growth of France exceeds the threshold, the economic growth leads to high environmental quality that is associated with the Environmental Kuznets Curve (EKC) hypothesis (Dogan & Aslan, 2017) in the long-run relationship. However, we observed a unidirectional relationship between the inbound tourism development and economic growth and CO2 emissions and inbound tourism development in the short-run relationship in France. The presence of the tourism-led growth (TLG) hypothesis is supported by many research studies, including our study. The TLG hypothesis is validated for France (Dritsakis, 2012).

The USA is the world’s second largest contributor to CO2 emissions, accounting for 14% of the total global CO2 emissions. The main sources of carbon emissions in the USA are transportation sectors (accounting for 38.1% of the total domestic energy consumption), residential sectors (16.2%), industrial sectors (15.8%), and service providers (13.4%) (Muntean et al., 2018). The transportation sector proposes many energy-saving and carbon-reducing policy measures, such as a new generation of air transportation systems, other low-emission aviation, highway, and railway programs, road greenhouse gas assessment tools, renewable fuel standard programs, smart road transportation partnerships, light vehicle fuel economy and environmental labels, the National Clean Diesel Campaign, advanced technology automobile manufacturing loan programs, fuel efficiency and renewable fuel measures, national and alternative fuel supplier fleet programs, and other measures (Salari et al., 2021). Many important tourist locations in the USA have improved the public transportation system, thereby reduced the dependence of tourists on private vehicles. In order to reduce the number of cars, they added new buses, and about half of them use compressed natural gas to operate. In addition to the transportation sector, many hotels used different ways to operate under low carbon. For example, the use of power plants powered by clean burning natural gas can provide more than 50% of the electricity required for nearly 2000 rooms of a hotel and thus can reduce carbon emissions by more than 30%.

In the USA, tourism development is not considered a useful policy for promoting economic growth, and therefore the US government focuses on other sectors such as the export sector (export-led growth) or the manufacturing sector that sets up a healthy fiscal or monetary policy in order to promote the economic growth. In the USA, there is a two-way relationship between economic growth and CO2 emissions in cointegration (in the long run), whereas, in the short term, there is a one-way Granger causality between the economic growth and CO2 emissions. The empirical results of the bootstrap ARDL model in the USA show the growth-led emissions when the impact of the economic growth on the environment quality is statistically significant.

The findings of this research reveal that the inbound tourism development does not affect CO2 emissions in Spain. Instead, it is different from the result of Fethi and Senyucel (2020). However, Spain has also implemented the “National Climate Change Adaptation Plan” in recent years by using assessment tools and testing methods in order to measure the impact of Spain’s ecosystem on climate change and the environmental impact of different socioeconomic sectors (Cantos & Rebollo, 2016). This further increases the participation of all agencies in different sectors and systems so that climate change policies can be implemented into appropriate sectors that are required to be developed. Among them, Spain has also formulated plans and strategies to adapt to climate change. In each autonomous region, in order to improve the sustainable development and environmental protection of tourism activities, each region initiated some actions, such as water and electricity saving incentives, personnel environmental awareness, and customer training in tourist facilities, which provide a lot of information on sustainable tourism. Although Spain has initiated many actions in order to reduce environmental pollution, the impact of international tourism on Spain’s carbon emissions is found to be limited and has no significant effect.

In Italy, however, we do not observe tourism-led growth, but the impact of CO2 emissions on the economic growth and the impact of the inbound tourism development on CO2 emissions are negative and significant. As a member of the European Union, Italy has pledged in its Intended Nationally Determined Contributions (INDCs) to the United Nations Framework Convention on Climate Change (UNFCCC) and the European Union as a whole, by 2030, to reduce at least 40% of the EU’s greenhouse gas emissions as compared to 1990. This plan includes a variety of industries such as energy, industrial processes, agriculture, garbage disposal, and urban land planning, including the transportation sector. Italy has put great efforts in reducing emissions, such as the “Clean Skies” program. As Italy is a member of the European Civil Aviation Organization (ECA), its main objective is to reduce carbon emissions caused by the aviation sector (Saint Akadiri et al., 2019a, b, c). The establishment of the Italian National Energy Strategy (INES) is primarily transformed into a low-carbon economy as a long-term goal (Malinauskaite et al., 2019; Sarrica et al., 2018). The development of a low-carbon economy in the tourism industry can sustain environmental resources, which reduces CO2 emissions and can indirectly improve economic growth. Classical churches, historical museums, and a large number of Roman monuments are still the main tourist attractions of Italy. However, Italy began to tout the expansive natural tourism areas and rural scenery guided tours. Strengthening sustainable tourism promotes the development of new tourist attractions, provides more hotel accommodations in natural attractions, and can also increase the image of the agricultural tourism, which is considered suitable for decentralized tourism in order to create the off-season tourism growth. The Bank of Italy proposed recommendations on the theme of the “green investment” policy in the report to improve the tourism waste recycling system, increase the use of energy efficiency, promote the protection of biodiversity and cultural heritage environment, and develop a sustainable tourism map. In turn, it provides a diversified tourism experience and encourages the development of low-polluting sectors in Italy.

In China, the two most important contributors of CO2 emissions are the tourism-related transportation and accommodation sub-sectors in the tourism industry (Meng et al., 2016). In order to fulfill its emission reduction commitments, the Chinese government has issued tourism-related documents, namely the “Opinions of the State Council on Accelerating the Development of the Tourism Industry” and “Guiding Opinions on Further Promoting the Tourism Industry.” In recent years, China’s tourism industry has introduced low-carbon transportation, low-carbon buildings, and the construction of smart landscape systems. Eliminating old transportation systems with high-carbon emissions and high air pollution and promoting the use of public buses and hybrid vehicles can help reduce carbon dioxide emissions in the tourism industry (Zha et al., 2020). If implemented effectively, it can significantly protect the environment, but the carbon emissions caused by advanced transportation technology can still seriously pollute the environment in China. From the perspective of China’s inbound tourism, it pollutes the environment in the long run (Zhang & Zhang, 2020).

Furthermore, a bidirectional causality is observed between the economic growth and CO2 emissions. The economic growth led to tourism, and tourism led to the economic growth are supported by the existing literature (Dogan et al., 2017). In China, where manufacturing is the main economic driver, the increase in carbon emissions can effectively promote economic development, as shown in Table 5. We observed that the short-run economic growth shows a positive impact on the environment and a negative impact on CO2 emissions, which also exist in the long-run relationship. However, the low-carbon economy reduces CO2 emissions and maintains the sustained economic growth associated with the Environmental Kuznets Curve (EKC) hypothesis (Dogan & Aslan, 2017) in the short-run relationship in China.

Table 5 Short-run Granger causality test based on ARDL model
Full size table

5 Conclusions

This paper examines and analyzes the long-run equilibrium and the short-run Granger causality dynamics of the top five inbound tourists from 1995 to 2018, including CO2 emissions, inbound tourism development, and the economic growth of the bootstrap ARDL test with structural breaks.

This study explores a long-run equilibrium relationship among the three variables, which indicates that tourism development and the levels of statistically significant economic growth impact CO2 emissions in China, the USA, and France. Therefore, in the long run, in order to reduce CO2 emissions, tourism development plays a key role in France and the USA. The main finding for the USA and Italy shows a negative association between the inbound tourism development and environmental quality, which indicates that the tourism industry can improve pollution emissions in the short run. In Italy, the long-run association was never observed, but a short-run effect was found between CO2 emissions and inbound tourism development. In the USA, the coefficient of economic growth is found to be significant and positive for CO2 emissions, considering the EKC hypothesis. Consequently, the energy consumption and the utilization of the domestic industrial sector have increased environmental pollution in the USA. Nevertheless, the economic growth improves environmental quality in China by reducing CO2 emissions with the low-carbon economy. Thus, the economic development does not depend on a large number of polluting industries but on the policy of low-pollution and environmentally friendly sustainable operations in China.

5.1 Policy implications

The findings of our research reveal, first, aspects of the economic growth and inbound tourism development relationships in France, the USA, and China. The positive impact of the inbound tourism development on economic growth has proved the results of past literature in France and China. The development of tourism infrastructure and supporting facilities can further promote the tourism industry. Therefore, the tourism industry should implement the national industrial layout to help find opportunities and plans for long-term tourism goals in the country and assist decision-makers in achieving fast development. However, the inbound tourism development shows a negative and significant impact on the economic growth of the USA. Tourism development has replaced the resource allocation of other efficient industries (Shahbaz et al., 2018); that is, increasing the resource allocation of the tourism industry in the USA does not effectively improve the economy, but instead, it is counterproductive.

Second, in the long run, in order to reduce CO2 emissions, tourism development plays a key role in establishing and developing low-carbon tourism or sustainable tourist attractions in France and the USA. The more efficient energy consumption of air and road transportation and the development of better renewable energy is considered the key factors in reducing carbon emissions caused by tourism in France and the USA. The application of renewable energy in the tourism industry should be implemented more actively.

However, in China, we recommended increasing the use of renewable energy in the tourism industry and not keeping it limited to the use of traditional fossil energy. China should pay more attention to energy consumption and CO2 emissions caused by tourism in order to promote the low-carbon transition of the tourism industry (Zhang & Zhang, 2020). China should also use sustainable tourism, for example, forest resources and wetland park protection, low pollution, and low energy consumption (Tang et al., 2011) to improve environmental quality.

Despite certain limitations, the results of this study emphasize policy implications for countries with large tourist populations. Future research should be extended to different fields and countries, which can provide more plans for the development of sustainable tourism policies in specific regions. Moreover, for example, renewable energy, sustainable tourism investment, and low-carbon economic investment can be used as important variables for future research and analysis. Investment in sustainable tourism can effectively reduce the carbon dioxide emissions caused by transportation used in the tourism industry and can also increase the income generated by tourism. Therefore, the environmental impact model should consider more tourism investment so that the association between the environment and tourism can be further improved.