Tourism,the environment,and energy policies

Abstract

This article examines the impact of tourism development on carbon dioxide (CO₂) emissions for Organization for Economic Co-operation and Development (OECD) countries by particularly exploring the role of energy markets in the environment–tourism relation. We find that tourism growth raises more CO₂ emissions in the future, and that greater CO₂ emissions return a lagged and negative impact on tourism development. Our empirical results suggest that an improvement in energy efficiency simultaneously benefits the sustainability of both tourism development and the environment.

Keywords

energy environment sustainability tourism

Introduction

Tourism development is an important issue to nearly all economies around the world, because it generates many benefits such as more foreign exchange earnings and export revenues from international tourists. Unlike other industries, tourism involves a variety of business sectors, such as restaurants, airlines, and casinos, and it relates to several factors (e.g. market size emphasized by Brau et al., 2007). Tourism growth also helps stimulate economic growth, especially for developing countries due to more investment in human capital and physical capital (e.g. Fayissa et al., 2008).

Economic activities inevitably lead to environmental degradation, and the issue of cutting greenhouse gases for mitigating climate change is very much stressed nowadays. A few studies (e.g. Yabuuchi, 2013) simultaneously examine the impact of tourism development on environmental quality, while others look at its effect on the economy (e.g. Tkalec and Vizek, 2016). Tourism development has a strong influence on economic growth, but it may also damage the environment of tourism destinations at the same time. Environmental degradation could even initiate a bad feedback effect on tourism development or economic development (e.g. Giannoni and Maupertuis, 2007). Thus, some studies (e.g. Hernandez and Leon, 2013) suggest limiting investment into the tourism sector. Interestingly, the environmental Kuznets curve (EKC) presents an inverted U-shaped relationship between environmental degradation and economic development. In other words, environmental quality will improve when per capita income hits a certain value, which drives the present research to search for the causes.

In addition to an examination of the EKC model, this article highlights the potential impact of carbon dioxide (CO₂) emissions on tourism development, and we particularly explore which energy policies can simultaneously maintain tourism development and environmental quality. Climate changes have caused many natural disasters, and countries have begun to execute policies to help prevent them from happening, to which cutting CO₂ emissions is the primary goal (e.g. Kyoto Protocol). In fact, there could be bidirectional causality between industry development and CO₂ emissions. While many studies (e.g. Lee and Brahmasrene, 2016; Shakouri et al., 2017) support the effect that tourism development has on CO₂ emissions, only a few (e.g. Cheung and Law, 2001; Deng et al., 2017) confirm the impact that air pollution (haze pollution) has on tourism development.¹ Our article thus looks to fill the gap in this related literature.

Economic or tourism development intuitively points to more consumption of natural resources or energy, and environmental quality seemingly and inevitably diminishes. However, a few studies (Lee and Brahmasrene, 2013) find that tourism growth can lower CO₂ emissions, to which Scott et al. (2010) offer a possible cause from the angle of technological energy efficiency (EE) gains. So far, there are several theories (please see the following section) for interpreting the positive influence of human activities on improving environmental quality, but the related analysis is limited to the environment–tourism nexus. We thereby intend to fill the gap in the existing literature to explore the effect of energy market conditions on the CO₂ emissions–tourism relation.

Through panel data analysis on OECD countries over the period 1995–2014, we have several main findings. First, the EKC hypothesis only holds in the model with present CO₂ emissions and lagged per capita income. Second, although current high tourism growth leads to low present CO₂ emissions, it has a lagged and positive influence on future CO₂ emissions. Third, higher current CO₂ emissions will lower future tourist arrivals—that is, CO₂ emissions raise a negative feedback effect on tourism. Fourth and finally, the improvement of energy use efficiency is a relatively effective policy to help tourism development reduce CO₂ emissions, and it thus benefits the sustainabilities of tourism development and environmental quality simultaneously.

The rest of the article is organized as follows. The second section is the literature review. The third section describes the variables, model specifications, and the data. The fourth section describes the empirical process and methodologies. The fifth section is our empirical analysis and suggestions. The final section presents conclusions.

Literature review

Although economic development is crucial for governments, it leads to a depletion of natural resources and more energy consumption, which inevitably causes environmental degradation. Interestingly, the EKC hypothesis suggests that environmental quality will improve when per capita income is higher than a certain threshold, but the empirical findings of the EKC hypothesis are mixed, to which several factors (e.g. different methodologies, time periods, groupings, or measures) are used to account for them (e.g. Bartz and Kelly, 2008; Bengochea-Morancho et al., 2001; Cherniwchan, 2012; Dijkgraaf and Vollebergh, 2005; Galeotti et al., 2009; Musolesi and Mazzanti, 2014; Wang, 2013). For example, Bernard et al. (2015) are interested in estimating the tipping point on the EKC. They analyze several groupings and employ various parametric and nonparametric methods due to the consideration of endogeneity, weak identification of the tipping point, estimation uncertainty, and so on. Their analyses only support an inverted U-shaped EKC in OECD countries but not elsewhere, and the tipping point is still difficult to estimate precisely. Similarly, Halkos and Managi (2017) employ a conditional directional distance function estimator that is better at measuring environmental efficiency and incorporates dynamic effects based on nonparametric frontier analysis. Their findings present that the income–environmental efficiency relation exhibits an inverted U-shaped relationship for developed countries but an N-shaped form for developing countries.

As to the inverted-U environmental degradation–income relation, there exist some interesting theories interpreting it. McConnell (1997) argues that the willingness to pay (WTP) for pollution reduction varies with income, while the improvement of environmental quality could be regarded as a luxury good (Ghalwash, 2008) or not (e.g. Jacobsen and Hanley, 2009). Barbier et al. (2017) therefore theoretically and empirically examine the income elasticity of the marginal WTP for reducing pollution. They find that income elasticity’s range is 0.1–0.2 for low-income people and 0.6–0.7 for high-income people—that is, the income elasticity of the marginal WTP for pollution reduction varies with income. Therefore, environmental degradation could improve along with income growth.

Bosi and Desmarchelier (2018) refer to two effects: the compensation effect and the distaste effect (see, e.g. Heal, 1982; Finkelstein et al., 2013). People consume more to compensate for the utility loss from pollution (compensation effect). However, higher pollution can damage people’s health, and they then typically lower their consumption (distaste effect). Linking up with the EKC hypothesis, we can imagine that the damage from environmental degradation to people’s health is generally more prominent when a country’s economic development (e.g. industrialization) has met a certain level—that is, people’s activities cause pollution, and pollution affects them in turn.

Marsiglio et al. (2016) emphasize the role of structural changes to the EKC pattern by employing decomposition analysis for the EKC reduced form. The growth in household income is accompanied by structural changes, such as changes in the final demand from goods to services, changes to inter-industry labor mobility, or inter-industry productivity differentials (Schettkat and Yocarini, 2006), which then alter the system of economic production from agriculture (low pollution) to industry (relatively high pollution) and then again to services with low pollution (Panayotou, 1993). Marsiglio et al. (2016) indicate that the EKC hypothesis holds based on the standard balanced growth path analysis. However, the negative pollution–income relation only persists transitorily, and it shifts to a positive relation in the long run.

Lisciandra and Migliardo (2017) target the role of corruption upon the environment and find that corruption indeed deteriorates environmental quality. Corruption impacts the environment through two channels: (i) environmental policy and regulation and (ii) pollution levels. Generally speaking, the control of corruption is not stringent in developing countries in comparison to the developed ones. Therefore, when a country’s economic situation becomes much mature, environmental quality could improve after enhancing the control of corruption. Similarly, Tarverdi (2018) explores the effect of several governance dimensions on pollution from the role of socioeconomic institutions on economic performances or environmental welfare (e.g. North, 1990). Among several governance dimensions, he finds that control of corruption raises a negatively nonlinear effect on CO₂ emissions by employing parametric and nonparametric methods.

It is noted that the EKC is a simple reduced form model, making it difficult to capture how the effect of economic development or income per capita transfers to environmental degradation, and thus relevant policies cannot be implemented efficiently. However, He (2009) indicates that the EKC reduced form model based on country datasets could be “spurious” and may not be “optimal” for an individual country from a strict econometrical perspective. For example, when interpreting the EKC hypothesis, the position based on shifts of production technologies or abatement technologies is widely known and explains it directly (see, e.g. Andreoni and Levinson, 2001; Egli and Steger, 2007; Smulders et al., 2011), but there may be further doubt as to what drives these shifts. In analyzing the structural EKC model, He (2009) finds that industrial composition, scale, technical progress, and China’s international trade can all affect the density of industrial sulfur dioxide emissions. Similarly, stressing the spatial correlation between countries, Aklin (2016) finds that trade could affect pollution in industrialized countries; particularly, the EKC pattern disappears once spatial correlation is considered. Conversely, Zugravu-Soilita (2017) notes that foreign direct investment (FDI) can impact industrial pollution depending on the stringency of environmental regulations in FDI host countries. Whether it is international trade or FDI, they both involve the diffusion of new pollution abatement technologies or a more efficient use of resources. Of course, a country’s technological innovations on energy markets (e.g. nuclear power and better energy-saving devices) can directly reduce emissions (Lindmark, 2002), and this issue is our primary concern for the CO₂ emissions–tourism relation.

Tourism is an important industry for most countries. Exports of tourism services (export-led growth hypothesis) are regarded as one theoretical channel for enhancing economic product (e.g. Balaguer and Cantavella-Jorda, 2002). Nowak et al. (2007) propose another theoretical viewpoint that tourism development attracts the financing imports of foreign capital goods and thereby raises domestic capital formation (e.g. research and development) and economic product. Hence, tourism development is also associated to environmental quality just like economic development does.

Environmental quality can be a determinant of tourism development, relating to foreign tourists’ satisfaction. In other words, the feedback of pollution on tourism development is likely present. Cerina (2007) theoretically demonstrates that tourism growth, better environmental quality, and economic growth can happen simultaneously under certain conditions, but this situation is only maintained temporarily when pollution abatement expenditures are no longer available. Thus, tourism development can reduce environmental resources or quality, and additional policies are necessary. Marsiglio’s (2015) extended theoretical model offers similar suggestions, and the theoretical model of Giannoni and Maupertuis (2007) also proposes that policymakers should disinvest in tourism in order to tackle the degradation of the environment resulting from tourism development, and implementing a tax on tourism revenue might be beneficial. Accordingly, excessive tourism development might result in a negative influence backward on itself if environmental quality cannot be maintained. However, Forsyth et al. (1995) theoretically analyze many economic instruments (e.g. user pays, quantitative restrictions, etc.) and indicate that these instruments could be ineffective under some practical situations. Moreover, setting prices or quantitative restrictions may even result in worse results under the balance between the environment and tourism development due to some uncertainties, such as excluding visitors and transactions costs. Therefore, an empirical analysis is necessary.

There are mixed empirical findings on the CO₂ emissions–tourism relation. Some studies (e.g. Shakouri et al., 2017) support that tourism growth causes more CO₂ emissions, but a few studies (Lee and Brahmasrene, 2013) present the opposite results. We note that some investigations show that bad air conditions can damage tourism. Huybers and Bennett (2000) indicate that air conditions can affect UK tourists’ holiday destination choices. Zhang et al. (2015) also find that haze pollution significantly impacts tourist arrivals in Beijing. There are studies in the literature focusing on the reduction of emissions. Alfsen et al. (1995) and Smeral (1996) suggest that a carbon or energy tax is useful for cutting emissions. Rasmussen (2001) note that a learning-by-doing policy in renewable energy (RE) can lower CO₂ emissions, while Mazzanti and Musolesi (2017) offer the use of environmental policies (e.g. Kyoto Protocol agreement) to do the same thing. Ward et al. (2017) propose that an improvement in efficient technologies can effectively reduce global energy consumption and CO₂ emissions.

From the discussion above, tourism industry development accompanied by the relevant buildup of infrastructures can increase CO₂ emissions. To tackle the possible negative feedback effect of CO₂ emissions on tourism development, one has to search for effectual policies to prevent it. This article looks at three energy market conditions. The first is a RE market that curtails the use of fossil fuels and thus exhausts less CO₂ emissions. Second, we consider the condition of energy concentration or its converse energy diversification. More and more researchers note the advantages of an energy diversification policy, such as ensuring continued energy security, but the effect of energy diversification on CO₂ emissions is sparsely investigated. Energy diversification could promote the development of RE (Papiez et al., 2017) and likely reduce CO₂ emissions; however, it could also raise CO₂ emissions through decreasing energy prices and a greater consumption of fossil fuels. Our final consideration is EE. As suggested in previous studies (e.g. Park et al., 2017), improving energy use efficiency can effectively reduce energy consumption and possibly curtail CO₂ emissions. We focus on the interactions between tourism development and these three energy conditions on CO₂ emissions.

Model specifications, variables, and data

Model specifications

Our empirical models are extended from the EKC hypothesis, where environmental quality is measured by CO₂ emissions, and economic development is gauged through per capita income. Moreover, we incorporate tourism development as a factor for determining CO₂ emissions because it relates to economic performance. Furthermore, as theoretical and empirical analyses have shown, the emissions–tourism development relation could change due to energy market conditions. We hence incorporate three kinds of energy market conditions respectively into the model. Finally, energy consumption is regarded as a crucial determinant of CO₂ emissions (e.g. Tsai and Pao, 2010).

We accordingly express CO₂ emissions as

{CO}_{2} = f (E, y, y^{2}, T, EM)

where E, y, T, and EM indicate energy consumption, per capita income, tourism development, and energy market conditions, respectively. Generally, E and T are expected to positively affect CO₂ emissions. The variables y, y², and CO₂ emissions relate to the EKC hypothesis, and the impact of EM on CO₂ emissions depends on the types of energy market condition. Moreover, we follow Paramati et al. (2017) to transform all variables, except for RE and energy concentration, into natural logarithms in order to mitigate the problem of different distributional properties of the data series. Thus, we can represent it as

ln {CO}_{2 i t} = α_{0} + α_{1} ln E_{i t} + α_{2} ln y_{i t} + α_{3} ln y_{i t}^{2} + α_{4} ln T_{i t} + α_{5} ln {EM}_{i t} + e_{i t}

where “ln” means taking a natural logarithm of the variables defined in equation (1), subscripts i and t indicate countries and periods, respectively, and e is the error term. However, the variables of equation (2) are not stationary. All of them are integrated of order one (i.e. I(1)) from our examination. Therefore, we take first differences of all variables. This is expressed as

dln {CO}_{2 i t} = β_{0} + β_{1} dln E_{i t} + β_{2} dln y_{i t} + β_{3} dln y_{i t}^{2} + β_{4} dln T_{i t} + β_{5} dln {EM}_{i t} + e_{i t}

where the sign “d” indicates first difference.

In our empirical analysis, we substitute energy market conditions, denoted as ln EM, with RE, HHI, and ln EE, respectively. Furthermore, according to our prior literature review, the conditions of energy markets can affect the CO₂ emissions–tourism relation, and so we modify equation (3) as follows

dln {CO}_{2 i t} = γ_{0} + γ_{1} dln E_{i t} + γ_{2} dln y_{i t} + γ_{3} dln y_{i t}^{2} + γ_{4} dln T_{i t} + γ_{5} dln {EM}_{i t} + γ_{6} dln T_{i t} \cdot dln {EM}_{i t} + e_{i t}

where the interaction term dln T·dln EM presents which energy market conditions are available to reduce CO₂ emissions when tourism sustainability is considerable. As to this equation, we expect that the parameters γ ₁, γ ₂, and γ ₄ are positive, and γ ₃ is negative. Importantly, if γ ₅ and γ ₆ are both negative, then we could advocate that this energy market condition helps not only to reduce CO₂ emissions but also simultaneously achieves the sustainabilities of tourism development and environmental quality.

We have so far discussed that CO₂ emissions could affect tourism development, which can lead to the endogeneity problem and subsequently a biased estimation. We thus employ the lags of the independent variables to overcome this problem and also consider the possible lagged interaction between tourism development and CO₂ emissions (e.g. Katircioglu, 2014). Thus, we further examine the following equation

dln {CO}_{2 i t} = θ_{0} + θ_{1} dln E_{i t - 1} + θ_{2} dln y_{i t - 1} + θ_{3} dln y_{i t - 1}^{2} + \sum_{m = 1}^{M} σ_{m} dln T_{i t - m} + \sum_{m = 1}^{M} μ_{m} dln {EM}_{i t - m} + \sum_{m = 1}^{M} π_{m} dln T_{i t - m} \cdot dln {EM}_{i t - m} + e_{i t}

where M is the maximal lag. We shall discuss it in the empirical analysis.

We further examine the impact of CO₂ emissions on tourism development. Our empirical models are based on the theoretical models of Kort et al. (2002) and Giannoni and Maupertuis (2007), in that the number of visitors is determined by infrastructure stock and pollution stock, and the number of tourists is measured by gross fixed capital formation, CO₂ emissions, and one-period lagged tourist arrivals, respectively. Note that the variable for one-period lagged tourist arrivals also has the effect of mitigating the biased estimation resulting from missed variables. The empirical model is expressed as

dln T_{i t} = ρ_{0} + ρ_{1} dln T_{i t - 1} + ρ_{2} dln {GE}_{i t - 1} + \sum_{m = 1}^{M} τ_{m} dln {CO}_{2 i t - m} + e_{i t}

where ln GE is the gross fixed capital formation in logarithmic terms. The parameter ρ ₁ is expected to be positive, implying a link to previous performance, while ρ ₂ is also expected to be positive due to the relevant tourism investments. In addition, CO₂ emissions might raise a negative impact on tourism development. If so, then it presents a feedback effect between tourism development and CO₂ emissions.

Variables

The main variables are measured as follows.

CO₂ (CO₂ emissions): CO₂ emissions (metric tons per capita);

E (energy consumption): Energy use (kg of oil equivalent per capita);

y (per capita income): Gross domestic product (GDP) per capita, purchasing power parity (PPP) (current international US$);

T (tourism development): International tourism, number of arrivals (thousands);

GE (infrastructure stock): Gross fixed capital formation (constant 2010 US$);

RE: Renewable energy consumption (% of total final energy consumption);

HHI (energy concentration): Herfindahl-Hirschman Index (HHI) is an inverse index of energy diversification calculated by $HHI = \sum_{i} {ES}_{i}^{2}$ , where ES _i is the percentage of total electricity production from source i; here, it contains hydroelectric, gas, nuclear, oil, coal, and renewable sources²;

EE: Energy intensity level of primary energy (MJ/$2011 PPP GDP), which is the ratio between energy supply and GDP measured at PPP.

Data

Our analysis aims at 35 OECD countries with annual data over the period 1995–2014, as some variables may have too many missing observations outside this period.³ The data are gathered from the World Bank database. Table 1 shows the descriptive statistics. The maximal amount of CO₂ emissions occurs in Luxembourg, and the minimum is in Latvia. France has the maximal international tourist arrivals. The maximal amount of energy concentration (HHI) occurs in Norway, whose main electricity production source is hydroelectricity. Iceland’s EE is the highest. Figure 1 illustrates a seemingly negative relationship between CO₂ emissions and tourist arrivals. Figure 2 represents the three energy market conditions. RE consumption is increasing since 2001, energy concentration is going down since 2008, and the average EE for all OECD members shows a decreasing trend.

Table 1.

Descriptive statistics.

	CO₂	E	y	T	RE	HHI	EE
Mean	9.0	4235.5	29014.4	13954.3	16.6	44.0	5.5
Min.	2.7	1052.7	5504.7	190.0	0.4	19.0	2.3
Max.	24.8	18178.1	101831.8	83701.0	77.4	99.0	19.2
SD	4.2	2360.6	13656.3	18036.1	15.6	19.1	2.4

Note: CO₂, E, y, T, RE, HHI, and EE are carbon dioxide emissions, energy consumption, per capita income, international tourist arrivals, renewable energy consumption (%), energy concentration (%; inverse of energy diversification), and energy efficiency, respectively. SD represents standard deviation.

Figure 1.

Tourist arrivals (T) and CO₂ emissions for OECD members. CO₂: carbon dioxide.

Figure 2.

RE consumption, HHI, and EE for OECD members. RE: renewable energy; HHI: energy concentration; EE: energy efficiency.

Empirical design and methodologies

In the following, we shall describe our empirical design and the corresponding methodologies. First, we test stationarity of the variables via the IPS (Im-Pesaran-Shin; Im et al., 2003), Fisher-ADF (augmented Dickey-Fuller; Maddala and Wu, 1999), and Fisher-PP (Phillips-Perron; Choi, 2001) tests. Second, we examine the EKC hypothesis and at the same time check if the panel data exhibit a fixed or a random effect by the Hausman test (Hausman, 1978). Third, we analyze the effects of tourism development and its interaction with energy markets on CO₂ emissions as in equation (4), which considers no feedback effect. We also diagnose the problems of omitted/redundant variables and cross-sectional dependence of residuals via the method of Pesaran (2004). Ignoring cross-sectional dependence can lead to invalid statistics and efficiency loss in testing. Fourth, we employ the test of De Wachter and Tzavalis (2012) to detect if there are structural breaks in panel data models. Fifth and finally, we consider the existence of the feedback effect and a lagged influence as in equations (5) and (6), respectively.

Empirical analysis and suggestions

According to our tests, all the variables are of I(1), but we do not report them in order to save space. Therefore, we take the first difference of all variables.

We now turn to the main analysis and first test the EKC hypothesis. Table 2 presents the results.⁴ For both specification 1 and specification 2, the EKC hypothesis does not hold. The only significant evidence presents that per capita income exhibits a negative effect on CO₂ emissions over the long run. We illustrate this in Figure 3 for the OECD group (i.e. average values of the 35 members for each year). The shape resembles the right side of an inverted-U. One reason for the results in Table 2 is that OECD members have been developed countries for a long time, and a very short period may not be able to capture their inverted-U shape. More interestingly, a theoretical parabola function could be an inappropriate setting for the empirical EKC hypothesis analysis. Figure 3 shows that a switch in the environmental degradation–per capita income relation might occur only when per capita income surpasses a wide range, like the nearly flat part in Figure 3 ranging from 9.8 to 10.4. By intuition, households or firms need to pay a lot of money to replace energy-saving devices or to adjust to energy reforms. This might be the cause of the mixed findings in previous studies, but a further detailed examination is needed in the future.

Table 2.

EKC analysis.

	Specification 1		Specification 2
	Y = dln CO₂		Y = ln CO₂
Variable	dln y	dln y²	ln y	dln y²
Coefficient	0.852	−0.023	−0.451*	0.060

Note: EKC: environmental Kuznets curve; CO₂: carbon dioxide. In specification 1, the model is specified as a panel data model with the random effect with the period fixed via the Hausman test. Here, we do not present the constant term. Specification 2 presents the results of any existing long-run cointegrating relation.

* Statistical significance at the 5% level.

Figure 3.

The relationship between CO₂ emissions and per capita income for the OECD group. CO₂: carbon dioxide.

We next examine equation (4). Table 3 presents the results of different model specifications based on the assumption of no feedback effect.⁵ Similarly, the EKC hypothesis does not hold. Instead, energy consumption raises a strong and positive influence on CO₂ emissions. An additional 1% increase in energy consumption leads to about a 1.2% increase of CO₂ emissions. The findings are similar to previous studies (e.g. Tsai and Pao, 2010). Of significance is that the majority of OECD countries using fossil fuels contributes to this result, and governments should conduct some energy reforms to improve it if sustainable environment is the primary purpose.

Table 3.

Consideration of interactions between tourism and energy policies.

	Specification 1	Specification 2	Specification 3	Specification 4
c	−0.009** (0.002)	−0.006** (0.002)	−0.009** (0.002)	−0.006** (0.002)
dln y_it	−0.063 (0.564)	−0.118 (0.510)	0.199 (0.521)	−0.223 (0.552)
dln y ² _it	0.003 (0.027)	0.007 (0.025)	−0.008 (0.026)	0.011 (0.026)
dln E_it	1.292** (0.073)	1.171** (0.030)	1.208** (0.030)	1.235** (0.072)
dln T_it	−0.033** (0.015)	−0.011 (0.015)	−0.018 (0.013)	−0.029* (0.016)
dln EE _it	−0.062 (0.073)	—	—	−0.041 (0.072)
dln T _it·dln EE _it	−0.631** (0.273)	—	—	−0.630** (0.279)
d RE _it	—	−0.063** (0.013)	—	−0.068** (0.013)
dln T_it · d RE _it	—	−0.039 (0.110)	—	−0.030 (0.113)
d HHI _it	—	—	−0.018 (0.019)	−0.044** (0.019)
dln T _it · d HHI _it	—	—	−0.019 (0.196)	0.113 (0.205)
R ²	0.762	0.771	0.760	0.775

Note: The dependent variable is dln CO_2it. CO₂, E, y, T, RE, HHI, and EE are carbon dioxide emissions, energy consumption, per capita income, international tourist arrivals, renewable energy consumption, energy concentration, and energy efficiency, respectively. The value in the parenthesis is standard error.

** Statistical significance at the 5% levels.

* Statistical significance at the 10% levels.

We further observe the effects of tourism development, energy market conditions, and their interactions on CO₂ emissions. In specifications 1 and 4, when considering energy use efficiency, tourism development significantly and negatively affects CO₂ emissions, and the interaction between tourism and EE also raises a negatively significant effect on CO₂ emissions. In other words, the improvement of energy use efficiency itself not only lowers CO₂ emissions but also helps tourism development to reduce CO₂ emissions. This suggests that countries focusing on tourism development should be devoted to implementing reforms of energy use efficiency (e.g. subsidies of energy-saving devices) that benefit the sustainabilities of tourism development and the environment simultaneously.

In specifications 2 and 4, an increase in RE consumption has a significant influence on reducing CO₂ emissions as suggested in Rasmussen (2001), but it does not help tourism to improve CO₂ emissions. This phenomenon can be ascribed to many factors, such as higher prices of RE or a lower share of RE consumption among total energy resources, so that the use of fossil fuels is difficult to be replaced. As Rasmussen (2001) suggests, the policy of learning-by-doing in RE may be available. In specifications 3 and 4 of Table 3, the effects of energy concentration on CO₂ emissions are not consistent. In addition, the interaction of energy concentration and tourism development does not affect CO₂ emissions. Therefore, we tend to interpret that the condition of energy concentration does not affect CO₂ emissions or tourism development. However, we still support policies of energy diversification because our later analysis exhibits its merits.

The values of R ² are similar across model specifications 1–3. In addition, the null hypothesis that dln EE, dln T·dln EE, d RE, dln T·d RE, d HHI, and dln T·d HHI are redundant variables is rejected at the 5% significance level. Thus, the results of specification 4 are also apparent, but the existence of collinearity in statistics may be another problem. Summarily, our findings support that high tourism development can reduce current CO₂ emissions for OECD countries when we do not consider the feedback or lagged effect. In addition, reforms of energy use efficiency are strongly recommended to governments.

We finally check the existence of structural breaks employing the methodology of De Wachter and Tzavalis (2012). The tested model is specification 4 of Table 3. There are two ways to detect a structural break for the methodology. One is to specify a known break and test it, and the other is to assume there are unknown breaks and try to detect them. For the test of known breaks, we start from 1998 to 2011 (i.e. trimming 30% of the whole period). Table 4 reports the results. By means of specifying a known break, there are no significant breaks that occur. On the other hand, under the assumption of unknown breaks, although we find a possible existing structural break in 2009, the Sargan test statistic shows no statistical significance. In other words, whether under known or unknown breaks, we do not find any structural breaks in the model.

Table 4.

Detection of structural breaks.

Known break	Sargan test statistic	Unknown break	Sargan test statistic
1998	0.729	2009	0.269
1999	0.319
2000	1.021
2001	0.622
2002	0.406
2003	0.377
2004	0.222
2005	0.976
2006	1.821
2007	0.699
2008	3.022
2009	0.097
2010	0.114
2011	5.007

Note: The null hypothesis for the Sargan test is that there is no break. The tested model is specification 4 of Table 3 after removing all interaction terms.

We consider both the feedback and lagged effects of tourism development on CO₂ emissions. We first employ equation (5), excluding energy market conditions, to help capture appropriate lagged periods. Table 5 presents that the EKC hypothesis holds based on the use of the lags, which might relate to the responses of energy consumption to income changes as found in previous studies (e.g. Campbell, 1991), whereby lagged incomes can be used to predict consumption. Nevertherless, governments can conduct some energy markets reforms, such as cuts in fossil fuel use, that benefit the reduction of CO₂ emissions.

Table 5.

Considerations of feedback and lagged effects without energy conditions.

	Specification 1	Specification 2	Specification 3	Specification 4
c	−0.010* (0.004)	−0.011* (0.004)	−0.013* (0.004)	−0.013* (0.004)
dln y_it ₋₁	3.309* (1.028)	2.821* (1.077)	3.997* (1.162)	4.829* (1.194)
dln y ² _it ₋₁	−0.164* (0.051)	−0.141* (0.053)	−0.200* (0.057)	−0.240* (0.059)
dln E_it ₋₁	−0.241* (0.06)	−0.225* (0.063)	−0.244* (0.064)	−0.303* (0.066)
dln T_it ₋₁	0.029 (0.026)	0.031 (0.027)	0.040 (0.028)	0.029 (0.029)
dln T_it ₋₂	—	0.040 (0.026)	0.040 (0.026)	0.040 (0.027)
dln T_it ₋₃	—	—	0.057* (0.027)	0.066* (0.027)
dln T_it ₋₄	—	—	—	0.001 (0.028)
R ²	0.089	0.079	0.114	0.141

Note: The dependent variable is dln CO_2it. CO₂, E, y, and T are Carbon dioxide emissions, energy consumption, per capita income, and international tourist arrivals, respectively. The value in the parenthesis is standard error.

* Statistical significance at the 5% level.

Energy consumption has a negative and lagged effect on CO₂ emissions, on the other hand, but it does not counteract the positive influence of current energy consumption on CO₂ emissions as in Tables 3.⁶ The results of specifications 1–4 indicate that the influence of tourism development on CO₂ emissions is deferred by up to three periods/years. Therefore, we consider three periods as the maximal lag length in equation (5). The following estimation is used to interpret the result

dln E_{i t} = - {0.05}^{^{+}} - 0.09 \cdot dln E_{i t - 1} + 0.025 dln T_{i t} + 0.029 dln T_{i t - 1} + 0.025 dln T_{i t - 2} + {0.047}^{^{+}} dln T_{i t - 3}

The symbol ‘⁺’ of the coefficients in the above equation means statistical significance at the 5% level. Lagged tourism development causes a significantly positive influence on energy consumption, which explains the lagged effect of tourism development on CO₂ emissions. The lagged effect might be ascribed to the reason that relevant tourism investments only depend on past tourism performance.

We now consider the complete equation (5). Table 6 reports that the effects of income, energy consumption, and tourism development on CO₂ emissions are similar to the results of Table 3. We turn to the impacts of energy conditions and their interactions with tourism development on CO₂ emissions. In specification 1, the improvement of EE not only reduces CO₂ emissions but also helps tourism development curtail CO₂ emissions. In particular, the influence of EE on CO₂ emissions and the tourism–CO₂ emissions nexus only defers one period/year. The efforts from EE reforms are much quicker.

Table 6.

Considerations of feedback and lagged effects with energy conditions.

	Specification 1	Specification 2	Specification 3
c	−0.010** (0.004)	−0.005 (0.005)	−0.013** (0.004)
dln y_it− ₁	3.089** (1.239)	3.421** (1.177)	3.760** (1.173)
dln y ² _it− ₁	−0.165** (0.060)	−0.172** (0.058)	−0.188** (0.058)
dln E_it− ₁	0.061 (0.148)	−0.297** (0.066)	−0.255** (0.064)
dln T_it− ₁	0.001 (0.031)	0.015 (0.032)	0.039 (0.028)
dln T_it− ₂	0.027 (0.029)	0.017 (0.030)	0.040 (0.027)
dln T_it− ₃	0.054* (0.029)	0.059* (0.030)	0.061** (0.027)
dln EE _it− ₁	−0.295** (0.149)	—
dln EE _it− ₂	−0.103 (0.067)	—
dln EE _it− ₃	0.009 (0.065)	—
dln T_it− ₁·dln EE _it− ₁	−1.249** (0.547)	—
dln T_it− ₂·dln EE _it− ₂	0.052 (0.538)	—
dln T_it− ₃·dln EE _it− ₃	0.162 (0.538)	—
d RE _it− ₁	—	−0.061** (0.027)	—
d RE _it− ₂	—	−0.048* (0.026)
d RE _it− ₃	—	−0.042* (0.025)
dln T_it− ₁·dRE _it− ₁	—	0.458* (0.248)	—
dln T_it _{− 2} ·dRE _it− ₂	—	0.332 (0.248)
dln T_it− ₃·d RE _it− ₃	—	−0.073 (0.227)
d HHI _it− ₁	—	—	0.031 (0.039)
d HHI _it− ₂	—	—	0.017 (0.039)
d HHI _it− ₃	—	—	0.103** (0.039)
dln T_it− ₁·d HHI _it− ₁	—	—	−0.467 (0.415)
dln T_it− ₂·d HHI _it− ₂	—	—	−0.348 (0.407)
dln T_it− ₃·d HHI _it− ₃	—	—	−0.334 (0.428)
R ²	0.137	0.139	0.128

** Statistical significance at the 5% levels.

* Statistical significance at the 10% levels.

In specification 2, the development of RE markets can decrease CO₂ emissions, but its interaction with tourism development will increase CO₂ emissions. This may relate to the consideration of tourism managers toward higher prices of RE than to fossil fuels or other factors. A more detailed analysis is needed in the future. In specification 3, high energy concentration leads to more CO₂ emissions, but it does not affect the CO₂ emissions–tourism development relation. As Papiez et al. (2017) find, high energy concentration does not benefit the development of RE. Thus, high energy concentration leads to more CO₂ emissions. Summarily, policies of energy use efficiency reforms are better than other energy markets for tourism managers.

We finally examine whether CO₂ emissions affect tourism development with equation (6). According to Table 7, CO₂ emissions have a negatively lagged influence on tourism development for OECD countries. In other words, more current CO₂ emissions will reduce future international tourist arrivals. Combined with previous findings that tourism growth can have a lagged effect to increase CO₂ emissions, we realize that the feedback effect does exist. Therefore, it is important to reduce CO₂ emissions for tourism sustainability. Our empirical analyses suggest that the improvement of energy use efficiency deserves to be considered more thoroughly.

Table 7.

Effects of CO₂ emissions on tourism.

	Specification 1	Specification 2	Specification 3	Specification 4
c	0.036* (0.004)	0.035* (0.004)	0.034* (0.004)	0.036* (0.004)
dln T_it ₋ ₁	0.065 (0.042)	0.062 (0.043)	0.065 (0.045)	0.044 (0.046)
dln GE _it ₋ ₁	0.005 (0.048)	0.022 (0.05)	0.015 (0.051)	0.008 (0.055)
dln CO_2it ₋ ₁	0.034 (0.069)	0.039 (0.073)	−0.002 (0.076)	−0.002 (0.079)
dln CO_2it ₋ ₂	—	0.000 (0.072)	−0.014 (0.076)	−0.006 (0.079)
dln CO_2it ₋ ₃	—	—	−0.181* (0.075)	−0.171* (0.079)
dln CO_2it ₋ ₄	—	—	—	−0.037 (0.079)
R ²	0.222	0.226	0.244	0.262

Note: The dependent variable is dln T_it . GE and CO₂ are gross fixed capital and carbon dioxide emissions, respectively. The value in the parenthesis is standard error.

* Statistical significance at the 5% level.

Conclusions

This article analyzes the impact of tourism development on CO₂ emissions for the panel data of OECD countries covering the period 1995–2014. In addition to examining the EKC hypothesis, we explore whether there exists a feedback effect between tourism development and CO₂ emissions and which energy policies are worthwhile for the sustainabilities of tourism development and the environment.

Our empirical results present several findings and policy implications. First, the EKC hypothesis only holds under conditions of a long-term relation and the presence of lagged per capital income, which suggest the use of lagged variables for a relative empirical analysis or long-term analysis. Second, although current tourism growth has a weakly negatively effect on present CO₂ emissions, it does raise a lagged and larger positive impact on CO₂ emissions. More importantly, the third finding indicates that more CO₂ emissions will hurt future tourism development (i.e. the feedback effect), thus implying that governments should implement relevant policies to maintain environmental quality and tourism development simultaneously. Fourth and finally, our empirical results indicate that improving EE is better for the sustainability of tourism development and the environment than other energy policies that are similar to the suggestions or findings in previous studies (e.g. Lindmark, 2002; Scott et al., 2010). Policymakers can offer relevant subsidies for energy-saving devices to enhance energy use efficiency (e.g. Kelly, 2012). If energy use efficiency does not improve in the short-term, then policymakers may consider strengthening tourism regulation, controlling tourist inflows, limiting tourism investment, or implementing a carbon or energy tax for sustainability as recommended by previous studies (e.g. Alfsen et al., 1995; Cerina, 2007; Hernandez and Leon, 2013; Smeral, 1996; Marsiglio, 2017). Moreover, as international trade or FDI relates to technological diffusion or innovations (e.g. Aklin, 2016; He, 2009; Zugravu-Soilita, 2017), policymakers can promote policies to enhance cleaner production technologies or pollution abatement technologies. Countries focusing on developing their tourism industry can target this type of reform.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Aklin

(2016) Re-exploring the trade and environment nexus through the diffusion of pollution. Environmental and Resource Economics 64: 663–682.

Alfsen

Birkelund

Aaserud

(1995) Impacts of an EC carbon/energy tax and deregulating thermal power supply on CO₂, SO₂ and NO_x emissions. Environmental and Resource Economics 5: 165–189.

Andreoni

Levinson

(2001) The simple analytics of the environmental Kuznets curve. Journal of Public Economics 80: 269–286.

Balaguer

Cantavella-Jorda

(2002) Tourism as a long-run economic growth factor: The Spanish case. Applied Economics 34: 877–884.

Barbier

Czajkowski

Hanley

(2017) Is the income elasticity of the willingness to pay for pollution control constant? Environmental and Resource Economics 68: 663–682.

Bartz

Kelly

(2008) Economic growth and the environment: theory and facts. Resource and Energy Economics 30: 115–149.

Bengochea-Morancho

Higon-Tamarit

Martinez-Zarzoso

(2001) Economic growth and CO₂ emissions in the European Union. Environmental and Resource Economics 19: 165–172.

Bernard

Gavin

Khalaf

. (2015) Environmental Kuznets curve: tipping points, uncertainty and weak identification. Environmental and Resource Economics 60: 285–315.

Bosi

Desmarchelier

(2018) Limit cycles under a negative effect of pollution on consumption demand: the role of an environmental Kuznets curve. Environmental and Resource Economics 69: 343–363.

10.

Brau

Lanza

Pigliaru

(2007) How fast are small tourism countries growing? Evidence from the data for 1980–2003. Tourism Economics 13: 603–613.

11.

Campbell

(1991) The response of consumption to income: a cross-country investigation. European Economic Review 35: 723–767.

12.

Cerina

(2007) Tourism specialization and environmental sustainability in a dynamic economy. Tourism Economics 13: 553–582.

13.

Cherniwchan

(2012) Economic growth, industrialization, and the environment. Resource and Energy Economics 34: 442–467.

14.

Cheung

Law

(2001) The impact of air quality on tourism: the case of Hong Kong. Pacific Tourism Review 5: 69–74.

15.

Choi

(2001) Unit root tests for panel data. Journal of International Money and Finance 20: 249–272.

16.

De Wachter

Tzavalis

(2012) Detection of structural breaks in linear dynamic panel data models. Computational Statistics and Data Analysis 56: 3020–3024.

17.

Deng

(2017) Evaluating impact of air pollution on China’s inbound tourism industry: a spatial econometric approach. Asia Pacific Journal of Tourism Research 22: 771–780.

18.

Dijkgraaf

Vollebergh

HRJ

(2005) A test for parameter homogeneity in CO₂ panel EKC estimations. Environmental and Resource Economics 32: 229–239.

19.

Egli

Steger

(2007) A dynamic model of the environmental Kuznets curve: turning point and public policy. Environmental and Resource Economics 36: 15–34.

20.

Fayissa

Nsiah

Tadasse

(2008) Impact of tourism on economic growth and development in Africa. Tourism Economics 14: 807–818.

21.

Finkelstein

Luttmer

Notowidigdo

(2013) What good is wealth without health? The effect of health on the marginal utility of consumption. Journal of the European Economic Association 11: 221–258.

22.

Forsyth

Dwyer

Clarke

(1995) Problems in use of economic instruments to reduce adverse environmental impacts of tourism. Tourism Economics 1: 265–282.

23.

Galeotti

Manera

Lanza

(2009) On the robustness of robustness checks of the environmental Kuznets curve hypothesis. Environmental and Resource Economics 42: 551–574.

24.

Ghalwash

(2008) Demand for environmental quality: an empirical analysis of consumer behavior in Sweden. Environmental and Resource Economics 41: 71–87.

25.

Giannoni

Maupertuis

(2007) Environmental quality and optimal investment in tourism infrastructures: a small island perspective. Tourism Economics 13: 499–513.

26.

Halkos

Managi

(2017) Measuring the effect of economic growth on countries’ environmental efficiency: a conditional directional distance function approach. Environmental and Resource Economics 68: 753–775.

27.

Hausman

(1978) Specification tests in econometrics. Econometrica 46: 1251–1272.

28.

(2009) China’s industrial SO₂ emissions and its economic determinants: EKC’s reduced vs. structural model and the role of international trade. Environment and Development Economics 14: 227–262.

29.

Heal

(1982) The use of common property resources. In: Smith

Krutilla

(eds), Explorations in Natural Resource Economics. Baltimore: The Johns Hopkins University Press for Resources for the Future, 72–106.

30.

Hernandez

Leon

(2013) Welfare and environmental degradation in a tourism-based economy. Tourism Economics 19: 5–35.

31.

Huybers

Bennett

(2000) Impact of the environment on holiday destination choices of prospective UK tourists: implications for tropical north Queensland. Tourism Economics 6: 21–46.

32.

Pesaran

Shin

(2003) Testing for unit roots in heterogeneous panels. Journal of Econometrics 115: 53–74.

33.

Jacobsen

Hanley

(2009) Are there income effects on global willingness to pay for biodiversity conservation? Environmental and Resource Economics 43: 137–160.

34.

Katircioglu

(2014) International tourism, energy consumption, and environmental pollution: the case of Turkey. Renewable and Sustainable Energy Reviews 36: 180–187.

35.

Kelly

(2012) Sustainability at home: policy measures for energy-efficient appliances. Renewable and Sustainable Energy Reviews 19: 6851–6860.

36.

Kort

Greiner

Feichtinger

. (2002) Environmental effects of tourism industry investment: an inter-temporal trade-off. Optimal Control Applications and Methods 23: 1–19.

37.

Lee

Brahmasrene

(2013) Investigating the influence of tourism on economic growth and carbon emissions: evidence from panel analysis of the European Union. Tourism Management 38: 69–76.

38.

Lee

Brahmasrene

(2016) Tourism effects on the environment and economic sustainability of sub-Saharan Africa. International Journal of Sustainable Development & World Ecology 23: 221–232.

39.

Lindmark

(2002) An EKC-pattern in historical perspective: carbon dioxide emissions, technology, fuel prices and growth in Sweden, 1870–1997. Ecological Economics 42: 333–347.

40.

Lisciandra

Migliardo

(2017) An empirical study of the impact of corruption on environmental performance: evidence from panel data. Environmental and Resource Economics 68: 297–318.

41.

Maddala

(1999) A comparative study of unit root tests with panel data and a new simple test. Oxford Bulletin of Economics and Statistics 61: 631–652.

42.

Marsiglio

(2015) Economic growth and environment: tourism as a trigger for green growth. Tourism Economics 21: 183–204.

43.

Marsiglio

Ansuategi

Gallastegui

(2016) The environmental Kuznets curve and the structural change hypothesis. Environmental and Resource Economics 63: 265–288.

44.

Marsiglio

(2017) On the carrying capacity and the optimal number of visitors in tourism destinations. Tourism Economics 23: 632–646.

45.

Mazzanti

Musolesi

(2017) The effect of Rio convention and other structural breaks on long-run economic development-CO₂ relationships. Economia Politica 34: 389–405.

46.

McConnell

(1997) Income and the demand for environmental quality. Environment and Development Economics 2: 383–399.

47.

Musolesi

Mazzanti

(2014) Nonlinearity, heterogeneity and unobserved effects in the carbon dioxide emissions-economic development relation for advanced countries. Studies in Nonlinear Dynamics and Econometrics 18: 521–541.

48.

North

(1990) Institutions, Institutional Change and Economic Performance. Cambridge: Cambridge University Press.

49.

Nowak

Sahli

Cortes-Jimenez

(2007) Tourism, capital good imports and economic growth: theory and evidence for Spain. Tourism Economics 13: 515–536.

50.

Panayotou

(1993) Empirical tests and policy analysis of environmental degradation at different stages of economic development. Working Paper WP238, Technology and Employment Programme (Geneva: International Labor Office, 1993).

51.

Papiez

Smiech

Frodyma

(2017) Determinants of the renewable energy development in the EU countries. A 20-year perspective. In: The 11th Professor Aleksander Zelias International Conference on Modelling and Forecasting of Socio-Economic Phenomena. Zakopane, Poland, 9–12 May 2017, pp. 270–279. Zakopane, Poland: Foundation of the Cracow University of Economics.

52.

Paramati

Alam

Chen

(2017) The effects of tourism on economic growth and CO₂ emissions: a comparison between developed and developing economies. Journal of Travel Research 56: 712–724.

53.

Park

Xing

Hanaoka

. (2017) Impact of energy efficient technologies on residential CO₂ emissions: a comparison of Korea and China. Energy Procedia 111: 689–698.

54.

Pesaran

(2004) General diagnostic tests for cross section dependence in panels. Working Papers in Economics No. 435, University of Cambridge, and CESifo Working Paper Series No. 1229.

55.

Rasmussen

(2001) CO₂ abatement policy with learning-by-doing in renewable energy. Resource and Energy Economics 23: 297–325.

56.

Schettkat

Yocarini

(2006) The shift to services employment: a review of the literature. Structural Change and Economic Dynamics 17: 127–147.

57.

Scott

Peeters

Gossling

(2010) Can tourism deliver its “aspirational” greenhouse gas emission reduction targets? Journal of Sustainable Tourism 18: 393–408.

58.

Shakouri

Yazdi

Ghorchebigi

(2017) Does tourism development promote CO₂ emissions? Anatolia 28: 444–452.

59.

Smeral

(1996) Economic policy measures for reducing the environmental impact of tourism. Tourism Economics 2: 173–184.

60.

Smulders

Bretschger

Egli

(2011) Economic growth and the diffusion of clean technologies: explaining environmental Kuznets curves. Environmental and Resource Economics 49: 79–99.

61.

Tarverdi

(2018) Aspects of governance and CO₂ emissions: a non-linear panel data analysis. Environmental and Resource Economics 69: 167–194.

62.

Tkalec

Vizek

(2016) The price tag of tourism: Does tourism activity increase the prices of goods and services? Tourism Economics 22: 93–109.

63.

Tsai

Pao

(2010) CO₂ emissions, energy consumption and economic growth in BRIC countries. Energy Policy 38: 7850–7860.

64.

Wang

(2013) Functional sensitivity of testing the environmental Kuznets curve hypothesis. Resource and Energy Economics 35: 451–466.

65.

Ward

Radebach

Vierhaus

Fugenschuh

Steckel

(2017) Reducing global CO₂ emissions with the technologies we have. Resource and Energy Economics 49: 201–217.

66.

Yabuuchi

(2013) Tourism, the environment, and welfare in a dual economy. Asia-Pacific Journal of Accounting & Economics 20: 172–182.

67.

Zhang

Zhong

. (2015) Tourists’ perception of haze pollution and the potential impacts on travel: Reshaping the features of tourism seasonality in Beijing, China. Sustainability 7: 2397–2414.

68.

Zugravu-Soilita

(2017) How does foreign direct investment affect pollution? Toward a better understanding of the direct and conditional effects. Environmental and Resource Economics 66: 293–338.