Nonlinear Impact of Air Quality on Tourist Arrivals: New Proposal and Evidence

Abstract

This study makes a novel contribution to the tourism literature by proposing an inverted U–shaped impact of air quality on tourist arrivals and examining this relationship using 58 major tourist cities in China from 2004 to 2015. The estimation of linear and nonlinear dynamic panel regression tests is based on the system generalized method of moments. The linear test results show that air quality measured by the concentration of fine particulate matter (PM_2.5) has a significantly negative impact on both inbound and domestic tourist arrivals. The nonlinear test results validate an inverted U–shaped link between air quality and both inbound and domestic tourist arrivals.

Keywords

Air quality Tourist arrivals Dynamic panel regression System generalized method of moments approach

Introduction

China has seen miraculous economic growth since its reform and opening-up policy in 1978. In 1978, China’s economy was only US$149.541 billion, less than 1/15th of that of the United States. In 2010, China surpassed Japan as the world’s second-largest economy, with a gross domestic product (GDP) of US$6.100 trillion, more than three-fifths that of the United States. With an annual average real GDP growth rate of 9.507 percent from 1978 to 2017, China’s share of global GDP rose from 1.752 percent in 1978 to 15.157 percent in 2017 (World Bank 2018).

However, this unprecedented economic boom in so populous a country has come at a heavy environmental cost. Deteriorating air quality or air pollution as a consequence of poor air quality is now one of the most serious challenges facing China. Newspapers and websites regularly show photographs of smoggy skies in large cities like Beijing, Shanghai, and Xi’an. Note that air pollution could exacerbate global warming (Ramanathan and Feng 2009), decreases precipitation, degrades visibility (Clarke and Kapustin 2010; Ramanathan et al. 2005), and drives extreme weather events (Booth et al. 2012).

Moreover, there is growing awareness of the detrimental effects of air pollution on human health (Brunekreef and Holgate 2002; Pope et al. 1995). Air pollution has long been linked to respiratory disease, cardiovascular diseases, stroke, and lung cancer (WHO 2016). According to the WHO (2018), 4.20 million deaths worldwide were attributable to outdoor air pollution in 2016. Severe air pollution in China has therefore aroused much concern over the past decade.

Like its economy, China’s tourism industry is undergoing tremendous growth (Figure 1). The National Tourism Administration of the People’s Republic of China (CNTA 2018) claims that China’s inbound tourist arrivals has increased from 3.50 million in 1980 to 41.76 million in 2004 with an annual average growth rate of 10.882%. China’s ranking in terms of international tourist arrivals in the world tourism market rose from 18th in 1980 to fourth in 2004. While China’s inbound tourism market in terms of total international tourist arrivals has soared, the growth of China’s inbound tourism market has even declined since 2005. The annual average growth rate of China’s international tourist arrivals was only 2.195 percent from 2005 to 2017, about one-fifth of that from 1980 to 2004.

Figure 1.

Inbound tourist arrivals (ITA) and its growth in China (1980-2017).

There are several explanations for the decreased growth of China’s inbound tourism market. One possible reason is the country’s poor air quality (China Tourism Academy 2015). Previous studies have identified the adverse effects of poor air quality on inbound tourism (Becken et al. 2017; Deng, Li, and Ma 2017; Li et al. 2015; Zhang, Gao, and Ding 2017).

Tourism is an environment-dependent activity, and tourism is usually considered to be a highly environment-sensitive economic sector (Butler 1991; Goodall 1995). Environmental quality is both an important element of tourism competitiveness (Dwyer and Kim 2003; Hu and Wall 2005; Mihalič 2000; Ritchie and Crouch 2010) and a crucial determinant of tourism demand (Law and Cheung 2007).

Given this increased concern with environmental degradation over the past two decades, there is now an expanding body of literature on the effects of environmental change on tourism. One research stream has sought to illuminate the consequences of climate change to the global and national tourism industry (Agnew and Palutikof 2006; Amelung and Nicholls 2014; Hamilton, Maddison, and Tol 2005; Scott, McBoyle, and Schwartzentruber 2004), destinations (Atzori, Fyall, and Miller 2018; McKercher et al. 2015), particular tourism sectors (Ciari et al. 2017; Pongkijvorasin and Chotiyaputta 2013), and tourists (Gössling et al. 2006; Maddison 2001).

Several studies found that poor air quality or air pollution has negative impacts on tourism expansion. Li et al. (2015) examined whether smog in Beijing had affected tourists’ risk perception and satisfaction with their travels, and concluded that it increased foreign travelers’ perception of risk and reduced their satisfaction with their trip. Zhang et al. (2015) revealed that haze pollution had a considerable potential impact on international travel to China and identified distinct differences among travel elements and tourism market segments. From the perspective of destination image and risk perception, Becken et al. (2017) confirmed that potential travelers from the United States and Australia expressed negative views about China’s air quality, and these feelings appeared to erode destination image and diminish their intention to visit China.

These studies centered on the effects of air pollution on destination image (Becken et al. 2017), tourism experience (Li et al. 2015) and intention to visit (Zhang et al. 2015) by administering questionnaires to tourists or potential tourists. However, the quantitative effects of air quality on the tourism industry remain unclear. A few recent studies have investigated the quantitative effects of air pollution on tourism market.

Chen, Lin, and Hsu (2017) explored whether air pollution affected the business cycle of tourism demand at the Sun Moon Lake in Taiwan. They showed that the effects of air pollution on tourism demand depended on the phases of business cycle. Using panel data of 31 provinces in China, Deng, Li, and Ma (2017) confirmed that air pollution had significant adverse effects on international tourists visiting China. Wang, Fang, and Law (2018) used transaction data from Ctrip (China’s largest online travel agency) to explore the impact of air quality on demand for China’s outbound tourism (packaged tour). They concluded that air quality in the place of origin exerted a push effect as local outbound tourism demand increased when air quality deteriorated.

This study makes a novel contribution to the tourism research literature by exploring the relationship between air quality and tourist arrivals using dynamic panel data of 58 major tourist cities in China from 2004 to 2015. In addition to inbound tourist arrivals, the study considers domestic tourist arrivals and determines whether there is a difference between the consequences of deteriorating air quality on inbound and domestic tourist arrivals.

Domestic tourism plays a major role in China’s tourism market. China’s domestic tourism market has experienced continuous and rapid growth in the past two decades, with the exception of 2003 (Figure 2). The number of China’s domestic tourist arrivals rose from 524 million in 1994 to five billion in 2017, an average annual growth rate of 10.304 percent (CNTA 2018). China’s domestic tourism market was 82 times larger than its inbound tourism market in terms of the number of total tourist arrivals in 2017.

Figure 2.

Domestic tourist arrivals (DTA) and its growth in China (1994–2017).

While China has the world’s largest domestic tourism market, few studies have investigated the relationship between air quality and domestic tourism market. Peng and Xiao (2018) argued that residents of mainland China had serious concerns about the potential travel risk caused by Beijing’s smog. However, the effects of air quality on China’s domestic tourism demand still remain unknown. Hence, greater effort should be devoted to reveal the relationship between air quality and domestic tourist arrivals in China.

This study takes an unprecedented step in proposing a nonlinear impact of air quality on tourist arrivals. It makes a threefold contribution to the literature. First, this study helps to fill the research gap in previous studies by estimating the effects of air quality on tourist arrivals while considering the nonlinear nexus between air quality and tourist arrivals. Note that air pollution poses risks to human health, the natural environment, and traffic safety, all of which are harmful to the tourism industry (Kozak, Crotts, and Law 2007). The potential risks of air pollution depend on the concentration of pollutants. Organisms and ecosystems tend to tolerate pollutants up to a critical threshold (Fenn et al. 2011), below which there are no detectable adverse effects.

The environmental Kuznets curve (Grossman and Krueger 1995) demonstrated that in poor countries, air pollutants appear to increase with economic growth. China is a less developed country, and its continued economic growth gradually harms air quality (Liu and Diamond 2005). While economic growth benefits tourism in terms of increasing the attractiveness and competitiveness of a destination to tourists, the deterioration of air quality as a result of economic growth is a deterrent to tourism development by creating travel risks resulted from exposure to pollution.

This study proposes that tourist arrivals would continue until air quality in tourist cities reaches a threshold value (Phase I). When air quality exceeds this threshold value, the impact of air quality on tourist arrivals could become negative (Phase II) because the risk associated with air pollution would endanger tourists’ safety. In other words, this study hypothesizes that as the economy grows, tourist arrivals are accompanied by air quality, and once the level of air quality reaches a threshold, it damages tourism arrivals. Thus there might be a nonlinear (inverted U–shaped) relationship between air quality and tourist arrivals as shown in Figure 3.

Figure 3.

Theoretical connection between air quality and tourist arrivals.

In addition, when there is a possible potential endogeneity problem, the OLS, fixed effects, and random effects estimators of panel data regression could be biased and inconsistent (Arellano and Bond 1991; Blundell and Bond 1998). This study uses the system generalized method of moments (SYS-GMM) for dynamic panel data model to estimate the effects of air quality on tourist arrivals. As De Vita (2014) noted, this technique both accounts for the underlying data dynamics and corrects for serial correlation, measurement error, and endogeneity.

Finally, unlike the study by Deng, Li, and Ma (2017), which used provincial data to evaluate the impact of air pollution on China’s inbound tourism industry, this study uses the data from China’s largest tourist cities to analyze the effects of air quality on both inbound and domestic tourist arrivals. Given that the size of an average Chinese province is 0.31 million square kilometers, larger than the United Kingdom or Italy, the air quality of cities within the same province can vary considerably. The aggregated provincial-level data may not accurately reflect these differences.

The rest of this article is organized as follows. The next section defines the variables and describes the original data sources. The third section presents the dynamic panel data models and SYS-GMM estimator. The results from the SYS-GMM estimator are reported in the fourth section. The fifth section concludes this article with a discussion of major findings.

Data and Variables

This study selects 58 of China’s major tourist cities. All of these cities maintained data on tourist arrivals and air pollution from 2004 to 2015 (Figure 4). Among these 58 cities, 30 are municipalities and provincial capitals, such as Beijing, Shanghai, Hangzhou, and Xi’an, and 28 are popular tourist destinations, such as Guilin and Suzhou (see Table 1). Annual data of inbound (ITA) and domestic tourist arrivals (DTA) for each city are taken from the Yearbook of China Tourism Statistics.

Figure 4.

The distribution of 58 tourist cities and their concentrations of PM_2.5 in 2015.

Table 1.

Tourist Cities in Four Different Levels of PM_2.5 Concentrations (µg/m³) in 2015.

Air Quality Category^a	PM_2.5 Concentrations^a	PM_2.5 Health Effects^a	Precautionary Actions^a	City
Good	0–12.0	Little or no risk.	None	Sanya, Xining
Moderate	12.1–35.4	Unusually sensitive individuals may experience respiratory symptoms.	Unusually sensitive people should consider reducing prolonged or heavy exertion.	Beihai, Chengde, Chongqing, Datong, Fuzhou, Guangzhou, Guilin, Guiyang, Haikou, Hohhot, Kunming, Lanzhou, Luoyang, Nanning, Ningbo, Quanzhou, Shantou, Shenzhen, Taiyuan, Urumchi, Wenzhou, Xi’an, Xiamen, Yinchuan, Zhangzhou, Zhanjiang, Zhongshan, Zhuhai
Unhealthy for Sensitive Groups	35.5–55.4	Increasing likelihood of respiratory symptoms in sensitive individuals, aggravation of heart or lung disease, and premature mortality in persons with cardiopulmonary disease and the elderly	People with respiratory or heart disease, the elderly, and children should limit prolonged exertion	Beijing, Changsha, Chengdu, Dalian, Hangzhou, Huangshan, Jilin, Jiujiang, Nanchang, Qingdao, Qinhuangdao, Shijiazhuang, Weihai, Yantai, Zhengzhou.
Unhealthy	55.5–75.0	Significant aggravation of heart or lung disease and premature mortality in persons with cardiopulmonary disease and the elderly; significant increase in respiratory effects in the general population	People with respiratory or heart disease, the elderly, and children should avoid prolonged exertion; others should limit prolonged exertion	Changchun, Harbin, Hefei, Jinan, Lianyungang, Nanjing, Nantong, Shanghai, Shenyang, Suzhou, Tianjin, Wuhan, Wuxi

Source: US Environmental Protection Agency.

Air pollutants of the greatest public health and environmental concern are particulates, ozone, nitrogen dioxide, and sulphur dioxide (World Health Organization 2006). Fine particulate matter (PM_2.5) with an aerodynamic diameter of 2.5 micrometers (µm) or less poses more danger to human health and environment than other air pollutants (Environmental Protection Agency, 2013). PM_2.5 can penetrate deep into the lungs and cardiovascular system, inducing stroke, heart disease, lung cancer, chronic obstructive pulmonary diseases and respiratory infections such as pneumonia (WHO 2016).

PM_2.5 can also cause impaired visibility, environmental and materials damage (Environmental Protection Agency, 2013). Consequently, PM_2.5 is the key indicator of air pollution. PM_2.5 concentration is constantly monitored by governments around the world. According to the US Environmental Protection Agency (2013), air quality is considered good if an annual mean concentration of PM_2.5 is less than 12 µg/m³, and moderate with a mean concentration of PM_2.5 between 12.1 and 35.4 (see Table 1). If the mean concentration of PM_2.5 is between 35.5 and 55.4, the air quality is considered unsafe for sensitive populations. The health risks will increase with rising concentrations of PM_2.5 (see Table 1).

This study therefore uses the annual average concentrations (micrograms per cubic meter) of ground-level PM_2.5 (PM) to measure air quality in all 58 sample tourist cities. The data on PM_2.5 concentrations are taken from the data set “Global Annual PM_2.5 Grids from MODIS, MISR and SeaWiFS Aerosol Optical Depth (AOD) with GWR, v1 (1998 – 2016).” This data set is provided by the Socioeconomic Data and Applications Center (SEDAC) in NASA’s Earth Observing System Data and Information System (EOSDIS), hosted by the Center for International Earth Science Information Network (CIESIN) at Columbia University.

Following previous literature (De Vita 2014; Wang, Fang, and Law 2018), GDP per capita, accommodation facilities, and transportation accessibility used as control variables are included in estimated models. The variable GDPP denotes GDP per capita (in Chinese Yuan) and is a city’s economic output per person measured in the base year (2001) prices, which reflects one of economic resources a city can exploit to develop its tourism economy. GDPP is supposed to be positively related to tourist arrivals (Kim, Chen, and Jang 2006; Kim, Lee, and Mjelde 2018). The data on GDP per capita and population is obtained from China City Statistical Yearbook.

The variable SH is the number of star-rated hotels and used to control for the impact of accommodation facilities on tourist arrivals. The data on star-rated hotels are taken from the CEInet Statistics Database. Accommodation facilities are important structural elements of tourism destination, defined as a component of destination attractiveness (Ritchie and Crouch 2010). As major accommodation facilities in China, star-rated hotels can reflect tourism service infrastructure and facilities in tourism destination. There are five ranks, based on “Classification and Accreditation for Star-Rated Tourist Hotels” issued by General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China. Many of these hotels are rated as three-star or above. Most of these facilities are comfortable and convenient.

The dummy variable HSR equals 1 if a city has a high-speed railway station and 0 if it does not. The variable TAP, the total number of air passengers carried in the city, is used to control the effect of civil aviation infrastructure on tourist arrivals. Both variables capture the effects of transportation accessibility on tourist arrivals. Transportation accessibility is vital to tourists. An improvement in the transportation accessibility at a tourist destination is expected to promote the revitalization of urban and business tourism by reducing transportation costs (Albalate, Campos, and Jiménez 2017). The data on TAP is obtained from China City Statistical Yearbook.

Table 2 reports the descriptive statistics of variables. As shown in Table 2, ITA ranged from 0.553 to 1206.445 with a mean of 95.348 and a standard deviation of 171.055 from 2004 to 2015, while DTA ranged from 137.000 to 27569.420 with a mean of 3838.183 and a standard deviation of 4483.705 (see Table 2). The mean of DTA is about 40 times that of ITA, indicating that the size of the domestic tourism market is much larger than that of the inbound tourism market in China.

Table 2.

Descriptive statistics of variables.

Variable	Observation (City-Year)	Mean	Standard Deviation	Minimum	Maximum
Panel A	Inbound tourism
ITA	476	95.348	171.055	0.553	1,206.445
PM	476	37.777	14.777	10.539	81.929
GDPP	476	33,332.970	19,095.380	7,045.615	105,147.900
SH	476	106.013	104.833	15	836
HSR	476	0.223	0.416	0	1
TAP	476	579.9902	1,117.637	0	7,721

Panel B	Domestic tourism
DTA	468	3,838.183	4,483.705	137.000	27,569.420
PM	468	38.738	14.744	10.539	79.292
GDPP	468	38,341.030	23,424.350	7,880.396	143,327.000
SH	468	110.810	109.020	17	836
HSR	468	0.318	0.466	0	1
TAP	468	747.086	1,404.214	0	8497

Note: The number of inbound tourist arrivals (ITA), domestic tourist arrivals (DTA), and TAP is in 10,000s. SH is the number of star-rated hotels. HSR is a dummy variable and equal to 1 if a city has a high-speed railway station and 0 if it does not. TAP is the total number of air passengers carried in the city.

The mean value of PM is about 38 µg/m³ (37.777 µg/m³ in panel A and 38.738 µg/m³ in panel B) from 2004 to 2015, which is more than three times the level of good air quality for PM_2.5 concentrations (12 µg/m³) according to the Environmental Protection Agency (2013). The minimum of PM (10.539 µg/m³) is slightly lower than the safe levels for PM_2.5 concentrations, while the maximum of PM (around 80 µg/m³) at about seven times the safe level is unhealthy.

Methodology

Linear Model

To estimate the linear effects of air quality on tourist arrivals, the following basic panel data regression models based on a tourism demand function is used:

I T A_{i, t} = α_{0} + {x^{'}}_{i, t} α + η_{i} + ν_{i, t},

(1a)

and

D T A_{i, t} = β_{0} + {x^{'}}_{i, t} β + ω_{i} + u_{i, t},

(1b)

where the subscript i and t represent city and year separately; x is a vector of explanatory variables (PM) and control variables (GDPP, SH, HSR, and TAP); α and β are vectors of estimated coefficients for independent variables; α₀ and β₀ are separately intercept; η_i and ω_i reflect unobservable city-specific effects, ν_i,t and u_i,t are random error terms, with E(ν_i,t) = 0 and E(u_i,t) = 0 for all i and t.

The current observation of tourist arrivals could depend on its own lagged values (Garín-Muñoz and Montero-Martín 2007). To reduce the bias in model specification, this study includes lagged dependent variable as regressor and expands the static models of model (1a)-(1b) to dynamic model (2a)-(2b):

I T A_{i, t} = α_{0} + \sum_{k = 1}^{q} ρ_{k} I T A_{i, t - k} + {x^{'}}_{i, t} α + η_{i} + ν_{i, t},

(2a)

and

D T A_{i, t} = β_{0} + \sum_{k = 1}^{q} λ_{k} D T A_{i, t - k} + {x^{'}}_{i, t} β + ω_{i} + u_{i, t},

(2b)

where q denotes the maximum lag length. To estimate the parameters in a dynamic panel data model with unobserved individual specific heterogeneity, the first-differenced GMM estimators (Arellano and Bond 1991) is most commonly used. Sequential moment conditions are then used where one or more lagged variables are instruments for the endogenous differences and the parameters estimated by GMM (Arellano and Bond 1991). However, lagged levels of the series provide only weak instruments for the differenced equations; the first-differenced GMM estimators are likely to perform poorly when the time series is persistent and the number of time periods is small (Blundell and Bond 1998).

The SYS-GMM estimator (Blundell and Bond 1998) for dynamic panel data model combines moment conditions for the model in first differences with moment conditions for the model in level, and can improve on the GMM estimator in differenced model in terms of bias and root mean squared error (Bun and Windmeijer 2010). Therefore, this study presents estimates of equations (2a)-(2b) using the two-step SYS-GMM estimator which circumvents the finite sample bias.

In respond to skewness toward large values, this study takes the natural logarithms of variables except for dummy variable HSR. The baseline econometric models are specified as

\begin{array}{l} L I T A_{i, t} = α_{0} + \sum_{k = 1}^{q} ρ_{k} L I T A_{i, t - k} + α_{1} L P M_{i, t} + α_{2} L G D P P_{i, t} + \\ α_{3} L S H_{i, t} + α_{4} H S R_{i, t} + α_{5} L T A P_{i, t} + η_{i} + ν_{i, t}, \end{array}

(3a)

and

\begin{array}{l} L D T A_{i, t} = β_{0} + \sum_{k = 1}^{q} λ_{k} L D T A_{i, t - k} + β_{1} L P M_{i, t} + β_{2} L G D P P_{i, t} + \\ β_{3} L S H_{i, t} + β_{4} H S R_{i, t} + β_{5} L T A P_{i, t} + ω_{i} + u_{i, t} . \end{array}

(3b)

There is a linear relationship between air quality on tourist arrivals only if α₁ or β₁ is significantly different from zero, meaning that α₁ or β₁ = 0 can be rejected. For example, there is a linear relationship between air quality on tourist arrivals if the coefficient of air quality on tourist arrivals is significantly different from zero (α₁ or β₁≠0). In other words, air quality has a significant impact on tourist arrivals.

Nonlinear Model

We propose that there is a nonlinear effect of air quality on tourist arrivals. Thus, this study introduces the quadratic term of LPM to the baseline models (3a)–(3b) to test the possible presence of a nonlinear impact of air quality on both inbound and domestic tourist arrivals, and specifically whether there is an inverted U–shaped relationship between LPM and LITA (LDTA):

\begin{array}{l} L I T A_{i, t} = α_{0} + \sum_{k = 1}^{q} ρ_{k} L I T A_{i, t - k} + α_{1} L P M_{i, t} + α_{2} L P M_{i, t}^{2} + \\ α_{3} L G D P P_{i, t} + α_{4} L S H_{i, t} + α_{5} H S R_{i, t} + \\ α_{6} L T A P_{i, t} + η_{i} + ν_{i, t}, \end{array}

(4a)

and

\begin{array}{l} L D T A_{i, t} = β_{0} + \sum_{k = 1}^{q} λ_{k} L D T A_{i, t - k} + β_{1} L P M_{i, t} + β_{2} L P M_{i, t}^{2} + \\ β_{3} L G D P P_{i, t} + β_{4} L S H_{i, t} + β_{5} H S R_{i, t} + \\ β_{6} L T A P_{i, t} + ω_{i} + u_{i, t} . \end{array}

(4b)

Accordingly, the nonlinear link between air quality and tourist arrivals exists if α₂ (or β₂ ) is significantly different from zero. If α₁ = 0 and α₂ = 0 are all rejected, and if α₁ > 0 (α₂ < 0) and α₁ < 0 (α₂ > 0), it suggests an inverted (upright) U–shaped relationship between LPM and LITA. However, if α₁ = 0 is rejected and α₂ = 0 cannot be rejected, the link between air quality and inbound tourist arrivals is linear.

Estimation Results

Linear Relationship

The estimated coefficients of baseline econometric models (3a) and (3b) based on the two-step SYS-GMM estimators are presented in Table 3. According to Table 3, the regression coefficient of LPM on LITA is −0.2856, which is statistically significant at the 1% level. The regression coefficient of LPM on LDTA is −0.0362, also statistically significant at the 1% level. These results indicate that air quality has significant adverse effects on China’s inbound and domestic tourist arrivals.

Table 3.

Test results of linear impact of air quality on tourist arrivals.

	LITA		LDTA
	Coefficient	Standard Errors	Coefficient	Standard Errors
Constant	0.7352^***	0.1148	0.3224^***	0.0226
LPM	−0.2856^***	0.0099	−0.0362^***	0.0022
LGDPP	0.0929^***	0.0130	0.0135^**	0.0053
LSH	0.0807^***	0.0083	0.0399^***	0.0042
HSR	0.1101^***	0.0087	0.0430^***	0.0038
LTAP	0.0792^***	0.0073	0.0093^***	0.0009
LITA_t–1	0.6247^***	0.0053
LDTA_t–1			0.9446^***	0.0036
Observations	476		468
AR(1)	0.0318		0.0198
AR(2)	0.7944		0.1085
Sargan test	0.3200		0.6640

Note: LPM = natural logarithm of concentrations of PM_2.5 (PM); LITA = natural logarithm of inbound tourist arrivals (ITA); LDTA = natural logarithm of domestic tourist arrivals (DTA); LGDPP = natural logarithm of GDP per capita; LSH = natural logarithm of the number of star-rated hotels; LTAP = natural logarithm of the total number of air passengers carried in the city. The dummy variable HSR is equal to 1 if a city has a high-speed railway station and 0 if it does not. The values reported for AR(1) and AR(2) tests are the p values for the null hypothesis, respectively; no first-order or second-order serial correlation in the first-differenced residuals. The row for the Sargan test reports is the p values for the null hypothesis of instrument validity.

Significance at the 5% level.

***

Significance at the 1% level.

The coefficients of both lagged LITA and lagged LDTA are positive and statistically significant at the 1% level, suggesting that the current inbound and domestic tourist arrivals depend on their previous level. In addition, the regression coefficients of all four control variables, LGDPP, LSH, HSR, and LTAP, are positive and significant at the 1% level. As mentioned, the economic prosperity of one destination, measured by LGDPP, is expected to enhance tourism attractiveness and competitiveness, thereby contributing to tourism demand. The results of the regression test support this expectation.

Accommodation facilities catering to specific needs of different tourists are central to tourism. The significantly positive coefficients of LSH on LITA and LDTA reveal that both inbound and domestic tourist arrivals rely on the availability of comfortable accommodations. The presence of a high-speed railway network in one destination can enormously improve access to transportation at that destination and hence plays a vital role in attracting tourists. In addition, both inbound and domestic tourist arrivals are positively related to the total number of air passengers carried in the city. The test results found in the study are also consistent with this expectation.

Moreover, like all other GMM estimators, the two-step SYS-GMM estimators can produce consistent estimates only if the moment conditions used are valid. We conduct two diagnostic tests proposed by Arellano and Bond (1991) to check for overidentification restrictions and serial correlation. The results of the Arellano-Bond test for serial correlation in the first-differenced residuals show a first-order serial correlation since the null hypothesis of no first-order serial correlation can be rejected at the 5% significance level. There is no second-order serial correlation because the null hypothesis of no second-order serial correlation cannot be rejected. The Sargan test results of the overidentification restrictions cannot reject the null hypothesis that the overidentification restrictions are valid, suggesting that the instruments are valid. These results support the validity of model specification.

Nonlinear Relationship

Based on the regression test results in Table 3, there is a linear and negative impact of LPM on LITA and LDTA, showing that air pollution can significantly hurt China’s inbound and domestic tourist arrivals. These findings are consistent with the literature (e.g., Deng, Li, and Ma 2017; Wang, Fang, and Law 2018). To test our proposal that there is a nonlinear relationship between air quality and tourist arrivals, this study uses the two-step SYS-GMM estimators to estimate nonlinear models (4a) -(4b).

The regression results are reported in Table 4. As shown in Table 4, the coefficients of LPM and LPM² on LITA are 1.8199 and −0.3142, respectively; both are statistically significant at the 1% level. The positive coefficient of LPM and the negative coefficient of LPM² suggest that the relationship between air quality and inbound tourist arrivals is nonlinear (inverted U–shaped). Similarly, the coefficients of LPM and LPM² on LDTA are 0.3564 and −0.0571, respectively, and statistically significant at the 1% level, indicating an inverted U–shaped nonlinear relationship between air quality and domestic tourist arrivals.

Table 4.

Test Results of Nonlinear Impact of Air Quality on Tourist Arrivals.

	LITA		LDTA
	Coefficient	Standard Errors	Coefficient	Standard Errors
Constant	−2.6126^***	0.3125	−0.4049^***	0.0913
LPM	1.8199^***	0.1574	0.3564^***	0.0355
LPM ²	−0.3142^***	0.0236	−0.0571^***	0.0053
LGDPP	0.0728^***	0.0157	0.0217^***	0.0067
LSH	0.0854^***	0.0072	0.0413^***	0.0038
HSR	0.0747^***	0.0094	0.0337^***	0.0032
LTAP	0.0541^***	0.0041	0.0042^***	0.0009
LITA_t–1	0.6825^***	0.0070
LDTA_t–1			0.9451^***	0.0025

Observations	476		468
AR(1)	0.0423		0.0201
AR(2)	0.5879		0.1216
Sargan test	0.5536		0.4841

Significance at the 5% level.

***

Significance at the 1% level.

The results of the Arellano–Bond test for serial correlation and the Sargan test of the overidentification restrictions all support the validity of the specification of nonlinear regression models (4a)-(4b). Consequently, the test results of the nonlinear effects of air quality on both inbound and domestic tourist arrivals are reliable.

Discussion, Implications, and Conclusion

This study makes a novel contribution to the literature on tourism research by proposing an inverted U–shaped impact of air quality on tourist arrivals. The empirical investigation, based on 58 major tourist cities in China from 2004 to 2015 using the system generalized method of moments (SYS-GMM) approach, comprises two steps. The first step examines the traditional linear effect of air quality on tourist arrivals. The second step tests the new proposal: a nonlinear impact of air quality on tourist arrivals. The test results of two-step SYS-GMM estimators for dynamic panel data models reveal several interesting findings.

The first finding is that air quality measured by PM_2.5 concentrations has a significantly negative impact on both inbound and domestic tourist arrivals. This indicates that an increase in PM_2.5 concentrations would lead to a decrease in both inbound and domestic tourist arrivals. Given that severe air quality is now a problem, Chinese tourism authorities and policymakers must formulate and enforce stricter environmental regulations to control air pollution and reduce its adverse effects on inbound and domestic tourism.

The second finding is that the adverse effect of air quality is stronger on inbound than on domestic tourist arrivals in China. The absolute value of estimated coefficient of LPM on LITA is 7.9 (=0.2856/0.0362) times that of LPM on LDTA. This finding has one important implication: inbound or foreign tourists are much more concerned with China’s air quality than China’s domestic tourists are. A possible explanation for this finding is that international tourists are more safety-conscious (Pizam and Mansfeld 1996), and pay more attention to the corollaries of air pollution, such as disease and traffic accidents. China’s chronic air pollution significantly increases the risks of traveling to that country, and is a deterrent to many international tourists who decide not to go there.

The third finding is that while LPM, LGGDP, LSH, HSR, and LTAP are found to be significant in influencing China’s inbound tourist arrivals, LPM has the highest impact on China’s inbound tourism. As shown in Table 3, the absolute value of estimated coefficient of LPM on LITA is 3.07 (=0.2856/0.0929), 3.54 (=0.2856/0.0807), 2.59 (=0.2856/0.1101), and 3.61 (=0.2856/0.0792) times that of LGDPP, LSH, HSR, and LTAP on LITA, respectively. These results show that air quality plays the most important role in the development of inbound tourism in China.

In comparison, the absolute value of estimated coefficient of LPM on LITA is 2.68 (=0.0362/0.0135), 0.91 (=0.0362/0.0399), 0.84 (=0.0362/0.0430), 3.89 (=0.0362/0.0093) times that of LGDPP, LSH, HSR, and LTAP on LDTA, respectively. That is, air quality is ranked third of the five factors affecting China’s domestic tourist arrivals. These findings also echo the second finding that air quality has a stronger impact on inbound than on domestic tourism, supporting that inbound tourists indeed care more about China’s air quality than China’s domestic tourists do.

The fourth finding is that the impact of air quality on both inbound and domestic tourist arrivals is nonlinear and inverted U–shaped. This finding validates our proposal. The effect of air quality on tourist arrivals is not always negative, as found in the empirical results of linear regression tests in several previous studies. This study shows that as tourist cities continue to develop, the concentration of PM_2.5 increases. At the same time, the tourism infrastructure improves along with the facilities to increase destination attractiveness and competitiveness, thereby raising ITA and DTA. Nevertheless, once the concentration of PM_2.5 exceeds a certain threshold, high PM reduces the safety of tourists and limits both ITA and DTA.

Furthermore, the corresponding thresholds of the effect of LPM on LITA and LDTA can be calculated. As shown in Table 4, the corresponding test results of both LPM and LPM² on LITA and LDTA are given as

\begin{array}{l} L I T A_{t} = - 2.6126 + 1.8199 \times L P M_{t} - 0.3142 \times L P {M_{t}}^{2} + \\ 0.0728 \times L G D P P_{t} + 0.0854 \times L S H_{t} + 0.0747 \times \\ H S R_{t} + 0.0541 \times L T A P_{t} + 0.6825 \times L I T A_{t - 1}, \end{array}

(5a)

and

\begin{array}{l} L D T A_{t} = - 0.4049 + 0.3564 \times L P M_{t} - 0.0571 \times L P {M_{t}}^{2} + \\ 0.0217 \times L G D P P_{t} + 0.0413 \times L S H_{t} + 0.0337 \times \\ H S R_{t} + 0.0042 \times L T A P_{t} + 0.9451 \times L D T A_{t - 1} . \end{array}

(5b)

To compute the corresponding threshold levels of LPM maximizing LITA and LDTA, this study takes the derivative of LITA and LDTA with respect to LPM. The equations then become

0 = 1.8199 - 2 \times 0.3142 \times L P M,

(6a)

and

0 = 0.3564 - 2 \times 0.0571 \times L P M .

(6b)

Accordingly, the threshold of LPM maximizing LITA is found to be 2.896 µg/m³, and the threshold of LPM maximizing LDTA is 3.121 µg/m³ (see Figure 5). In other words, the threshold of concentrations of PM_2.5 maximizing inbound and domestic tourist arrivals are 18.102 µg/m³ and 22.669 µg/m³, respectively. These results indicate that inbound tourists are more concerned with air quality than China’s domestic tourists, which is also in line with the finding of linear tests.

Figure 5.

(A) The inverted U–shaped relationship between LPM and LITA. (B) The inverted U–shaped relationship between LPM and LDTA.

The results from nonlinear regression tests offer another valuable policy implication. LPM has an inverted U–shaped effect on LITA and LDTA, implying that air quality has a positive impact on tourist arrivals up to a threshold, and beyond that critical value, LPM harms LDTA and LITA. More importantly, given that the sample mean of LPM in 58 of China’s tourist cities from 2004 to 2015 (3.547) is higher than the threshold value of LPM maximizing LITA (2.896) (see Figure 5A), implying that poor air quality has damaged China’s inbound tourism. Similarly, the sample mean of LPM (3.575) is higher than the threshold value of LPM maximizing LDTA (3.121) (see Figure 5B). The finding indicates that poor air quality has also hurt China’s domestic tourism.

In conclusion, this study offers useful information for all tourist cities to understand the impact of air quality on their inbound and domestic tourism, which can help sustain their long-term tourism development. Chinese tourism authorities and policymakers can incorporate this study’s findings when they endeavor to control air pollution and mitigate its adverse effects on both inbound and domestic tourism. Specifically to optimize both inbound and domestic tourist arrivals, tourism authorities and policymakers need to improve air quality, lowering the concentrations of PM_2.5 to at least the threshold value (i.e., the concentration of PM_2.5 around the range from 18.102 µg/m³ to 22.669 µg/m³) to decrease the perceived travel risk and generate an image of China as a safe place for international tourists to visit.

Nonetheless, this article has some limitations. First, the data of tourism receipts in all 58 major tourist cities are not available. Were this data to become available, tourism authorities and policy makers would be able to calculate the trade-off between air quality and tourism revenue. Second, given that the optimal values of LPM maximizing LITA and LDTA are averaged annual values of all 58 major tourist cities for the entire sample period from 2004 to 2015, it may not be precisely applicable to each city. If the monthly or quarterly time series data over a long sample period are available, each city could determine its own optimal value of LPM maximizing its LITA and LDTA. Third, although this study points out that the current air quality has deteriorated and that it is therefore necessary for Chinese tourism authorities and policymakers to improve the air quality to optimize both inbound and domestic tourist arrivals, the issue of how to improve air quality is indeed beyond the scope of this study.

Fourth, the data of overnight stays and length of stay of tourists in 58 major tourist cities are not available. Given that tourists might shorten their stay if air quality is very poor, it would be interesting to understand the effects of air quality on overnight stays and length of stays of both inbound and domestic tourists. This impact might be stronger for foreign visitors than for domestic visitors.

In terms of future research, while this study proposes and supports that air quality has a nonlinear impact on inbound and domestic tourist arrivals in China, this may not be true elsewhere. Future researchers can conduct similar examinations based on data from other countries. Empirical findings from additional research could shed more light on the relationship between air quality and tourist arrivals and point to a more general conclusion.

In addition, previous studies indicate that weather conditions, such as temperature (Falk 2014; Michailidou, Vlachokostas, and Moussiopoulos 2016; Smith et al. 2016), rain (Álvarez-Díaz, Otero Giráldez, and González-Gómez 2010; Falk 2014), wind or storm (Álvarez-Díaz, Otero Giráldez, and González-Gómez 2010; Smith et al. 2016), and sunshine (Rosselló, Riera-Font, and Cárdenas 2011; Falk 2014) also have an impact on tourism demand. Future studies can examine if weather conditions significantly affect China’s inbound and domestic tourist arrivals.

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The financial support for this study was provided by the National Natural Science Foundation of China (grant number 71774135), Program of Study Abroad for Young Scholar sponsored by Department of Education of Anhui Province, China (grant number gxfx2017035), Project of Philosophy and Social Sciences in Anhui Province, China (grant number AHSKF2018D13) and Major Project of Humanities and Social Science Research in Colleges and Universities in Anhui Provincial, China (grant number SK2017ZD33).

ORCID iD

Ming-Hsiang Chen

Author Biographies

Liang-Ju Wang, PhD, is a Professor of Department of Tourism Management at Anhui University of Finance and Economics, China. His research areas are tourism economics and management.

Ming-Hsiang Chen, PhD, is a Professor of Hospitality Business Management at the Washington State University, USA. His areas of research are hospitality finance and hospitality philanthropy and tourism economics and management.

References

Agnew

M. D.

Palutikof

J. P

. 2006. “Impacts of Short-Term Climate Variability in the UK on Demand for Domestic and International Tourism.” Climate Research 31 (1): 109–20.

Albalate

Campos

Jiménez

J. L

. 2017. “Tourism and High Speed Rail in Spain: Does the AVE Increase Local Visitors?” Annals of Tourism Research 65:71–82.

Álvarez-Díaz

Otero Giráldez

M. S.

González-Gómez

. 2010. “Statistical Relationships between the North Atlantic Oscillation and International Tourism Demand in the Balearic Islands.” Climate Research 43:207–14.

Amelung

Nicholls

. 2014. “Implications of Climate Change for Tourism in Australia.” Tourism Management 41:228–44.

Arellano

Bond

. 1991. “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations.” Review of Economic Studies 58 (2): 277–97.

Atzori

Fyall

Miller

. 2018. “Tourist Responses to Climate Change: Potential Impacts and Adaptation in Florida’s Coastal Destinations.” Tourism Management 69:12–22.

Becken

Jin

Zhang

Gao

. 2017. “Urban Air Pollution in China: Destination Image and Risk Perceptions.” Journal of Sustainable Tourism 25 (1): 130–47.

Blundell

Bond

. 1998. “Initial Conditions and Moment Restrictions in Dynamic Panel Data Models.” Journal of Econometrics 87 (1): 115–43.

Booth

B. B. B.

Dunstone

N. J.

Halloran

P. R.

Andrews

Bellouin

. 2012. “Aerosols Implicated as a Prime Driver of Twentieth-Century North Atlantic Climate Variability.” Nature 484:228.

10.

Brunekreef

Holgate

S. T

. 2002. “Air Pollution and Health.” Lancet 360 (9341): 1233–42.

11.

Bun

M. J. G.

Windmeijer

. 2010. “The Weak Instrument Problem of the System GMM Estimator in Dynamic Panel Data Models.” Econometrics Journal 13 (1): 95–126.

12.

Butler

R. W.

1991. “Tourism, Environment, and Sustainable Development.” Environmental Conservation 18 (3): 201–9.

13.

Chen

Lin

Hsu

. 2017. “Does Air Pollution Drive away Tourists? A Case Study of the Sun Moon Lake National Scenic Area, Taiwan.” Transportation Research Part D: Transport and Environment 53:398–402.

14.

China Tourism Academy. 2015. Annual Report of China Inbound Tourism Development. Beijing: Tourism Education Press.

15.

Ciari

Nocera

Prettenthaler

Scuttari

. 2017. “The Impacts of Climate Change on Tourist Mobility in Mountain Areas.” Journal of Sustainable Tourism 25 (8): 1063–83.

16.

Clarke

Kapustin

. 2010. “Hemispheric Aerosol Vertical Profiles: Anthropogenic Impacts on Optical Depth and Cloud Nuclei.” Science 329 (5998): 1488.

17.

CNTA (National Tourism Administration of the Peoples Republic of China). 2018. Yearbook of China Tourism Statistics 2017.

18.

De Vita

. 2014. “The Long-Run Impact of Exchange Rate Regimes on International Tourism Flows.” Tourism Management 45:226–33.

19.

Deng

. 2017. “Evaluating Impact of Air Pollution on China’s Inbound Tourism Industry: A Spatial Econometric Approach.” Asia Pacific Journal of Tourism Research 22 (1): 1–10.

20.

Dwyer

Kim

. 2003. “Destination Competitiveness: Determinants and Indicators.” Current Issues in Tourism 6 (5): 369–414.

21.

Environmental Protection Agency. 2013. “National Ambient Air Quality Standards for Particulate Matter; Final Rule.” Federal Register 78 (10): 3086–287.

22.

Falk

2014. “Impact of Weather Conditions on Tourism Demand in the Peak Summer Season over the Last 50 Years.” Tourism Management Perspective 9:24–35.

23.

Fenn

M. E.

Lambert

K. F.

Blett

T. F.

Burns

D. A.

Pardo

L. H.

Lovett

G. M.

Haeuber

R. A.

Evers

D. C.

Driscoll

C. T.

Jeffries

D. S

. 2011. “Setting Limits: Using Air Pollution Thresholds to Protect and Restore U.S. Ecosystems.” Issues in Ecology 14.

24.

Garín-Muñoz

Montero-Martín

L. F

. 2007. “Tourism in the Balearic Islands: A Dynamic Model for International Demand Using Panel Data.” Tourism Management 28 (5): 1224–35.

25.

Goodall

1995. “Environmental Auditing: A Tool for Assessing the Environmental Performance of Tourism Firms.” Geographical Journal 161 (1): 29–37.

26.

Gössling

Bredberg

Randow

Sandström

Svensson

. 2006. “Tourist Perceptions of Climate Change: A Study of International Tourists in Zanzibar.” Current Issues in Tourism 9 (4/5): 419–35.

27.

Grossman

G. M.

Krueger

A. B

. 1995. “Economic Growth and the Environment.” Quarterly Journal of Economics 110 (2): 353–77.

28.

Hamilton

J. M.

Maddison

D. J.

Tol

R. S. J

. 2005. “Climate Change and International Tourism: A Simulation Study.” Global Environmental Change 15 (3): 253–66.

29.

Wall

. 2005. “Environmental Management, Environmental Image and the Competitive Tourist Attraction.” Journal of Sustainable Tourism 13 (6): 617–35.

30.

Kim

H. J.

Chen

Jang

S. S

. 2006. “Tourism Expansion and Economic Development: The Case of Taiwan.” Tourism Management 27 (5): 925–33.

31.

Kim

Lee

Mjelde

J. W

. 2018. “Impact of Economic Policy on International Tourism Demand: The Case of Abenomics.” Current Issues in Tourism 21 (16): 1912–29.

32.

Kozak

Crotts

J. C.

Law

. 2007. “The Impact of the Perception of Risk on International Travellers.” International Journal of Tourism Research 9 (4): 233–42.

33.

Law

Cheung

. 2007. “Air Quality in Hong Kong: A Study of the Perception of International Visitors.” Journal of Sustainable Tourism 15 (4): 390–401.

34.

Pearce

P. L.

Morrison

A. M

. 2015. “The Impact of Smog on Risk Perception and Satisfaction of International and Domestic Tourists in Beijing.” Tourism Tribune 30 (10): 48–59.

35.

Liu

Diamond

. 2005. “China’s Environment in a Globalizing World.” Nature 435:1179–86.

36.

Maddison

2001. “In Search of Warmer Climates? The Impact of Climate Change on Flows of British Tourists.” Climatic Change 49 (1): 193–208.

37.

McKercher

Shoval

Park

Kahani

. 2015. “The [Limited] Impact of Weather on Tourist Behavior in an Urban Destination.” Journal of Travel Research 54 (4): 442–55.

38.

Michailidou

A. V.

Vlachokostas

Moussiopoulos

. 2016. “Interactions between Climate Change and the Tourism Sector: Multiple-Criteria Decision Analysis to Assess Mitigation and Adaptation Options in Tourism Areas.” Tourism Management 55:1–12.

39.

Mihalič

2000. “Environmental Management of a Tourist Destination: A Factor of Tourism Competitiveness.” Tourism Management 21 (1): 65–78.

40.

Peng

Xiao

. 2018. “How Does Smog Influence Domestic Tourism in China? A Case Study of Beijing.” Asia Pacific Journal of Tourism Research 23 (12): 1115–28.

41.

Pizam

Mansfeld

. 1996. Tourism, Crime and International Security Issues. Chichester: Wiley.

42.

Pongkijvorasin

Chotiyaputta

. 2013. “Climate Change and Tourism: Impacts and Responses. A Case Study of Khaoyai National Park.” Tourism Management Perspectives 5:10–17.

43.

Pope

C. A.

Thun

M. J.

Namboodiri

M. M.

Dockery

D. W.

Evans

J. S.

Speizer

F. E.

Heath

C. W

. 1995. “Particulate Air Pollution as a Predictor of Mortality in a Prospective Study of U.S. Adults.” American Journal of Respiratory and Critical Care Medicine 151 (3, Pt 1): 669–74.

44.

Ramanathan

Chung

Kim

Bettge

Buja

Kiehl

J. T.

Washington

W. M.

Sikka

D. R.

Wild

. 2005. “Atmospheric Brown Clouds: Impacts on South Asian Climate and Hydrological Cycle.” Proceedings of the National Academy of Sciences of the United States of America 102 (15): 5326.

45.

Ramanathan

Feng

. 2009. “Air Pollution, Greenhouse Gases and Climate Change: Global and Regional Perspectives.” Atmospheric Environment 43 (1): 37–50.

46.

Ritchie

J. R. B.

Crouch

G. I

. 2010. “A Model of Destination Competitiveness/Sustainability: Brazilian Perspectives.” Revista de Administração Pública 44 (5): 1049–66.

47.

Rosselló

Riera-Font

Cárdenas

. 2011. “The Impact of Weather Variability on British Outbound Flows.” Climate Change 105 (1): 281–92.

48.

Scott

McBoyle

Schwartzentruber

. 2004. “Climate Change and the Distribution of Climatic Resources for Tourism in North America.” Climate Research 27 (2): 105–17.

49.

Smith

J. W.

Seekamp

McCreary

Davenport

Kanazawa

Holmberg

Wilson

Nieber

. 2016. “Shifting Demand for Winter Outdoor Recreation along the North Shore of Lake Superior under Variable Rates of Climate Change: A Finite-Mixture Modeling Approach.” Ecological Economics 123:1–13.

50.

Wang

Fang

Law

. 2018. “Effect of Air Quality in the Place of Origin on Outbound Tourism Demand: Disposable Income as a Moderator.” Tourism Management 68:152–61.

51.

World Bank. 2018. World Development Indicators. Washington, DC: World Bank.

52.

World Health Organization. 2006. Air Quality Guidelines-Global Update 2005. Geneva: WHO Press.

53.

World Health Organization. 2016. Ambient Air Pollution: A Global Assessment of Exposure and Burden of Disease. Geneva: WHO Press.

54.

World Health Organization. 2018. World Health Statistics 2018: Monitoring Health for the SDGs. Geneva: WHO Press.

55.

Zhang

Zhong

Wang

Dang

. 2015. “Tourists’ Perception of Haze Pollution and the Potential Impacts on Travel: Reshaping the Features of Tourism Seasonality in Beijing, China.” Sustainability 7 (3): 2397–414.

56.

Zhang

Gao

Ding

. 2017. “Impacts of Haze on the Intention to Visit China of Overseas Tourists: Based on Destination Image and Risk Perception Theories.” Tourism Tribune 32 (12): 58–67.