Perceived pollution and inbound tourism for Shanghai: a panel VAR approach

ABSTRACT

Although there exists sizeable literature studying the impact of tourists’ risk perceptions, less attention has been devoted to studying the difference between the impact of measured and perceived risks. We apply a panel vector autoregression (PVAR) to a dataset of inbound tourism in Shanghai, China to study the dynamic relationship between pollution and tourism. Particularly, we distinguish between measured pollution level and perceived pollution (measured by Google Trends search data) to test which one has the larger impact on tourists’ decisions. Our results show that pollution (measured or perceived) can be a powerful deterrent to potential international tourists.

KEYWORDS:

• Pollution
• perceived risk
• international tourism
• panel vector autoregression
• Shanghai

Disclosure statement

No potential conflict of interest was reported by the author(s).

ORCID

Markum Reed http://orcid.org/0000-0002-0072-475X

Notes

1 $P{M}_{2.5}$ refers to atmospheric particulate matter that has a diameter less than 2.5 micrometers, which is often responsible for causing air pollution and related health issues. More information can be found at https://www.epa.gov/pm-pollution/particulate-matter-pm-basics.

2 Statistics of $P{M}_{2.5}$ used in this paper were reported by the US Shanghai Consulate. Data portal: www.stateair.net. Accessed on December 12, 2016.

3 Data can be found at http://lyw.sh.gov.cn/. Accessed on December 2, 2016.

4 http://www.who.int/phe/health_topics/outdoorair/databases/en/

5 Total tourism in this instance is the combination of inbound tourism from Australia, Canada, Great Britain, and the United States.

6 In the case of the current paper, we look at search results for ‘Pollution in China’, ‘Pollution in Beijing’, and ‘Shanghai Pollution’. For example, the top 3∼5 search results presented by Google were the same. People searching for these terms would find similar links to only the first 3 webpages shown by Google. Subsequent webpages could differ. Therefore, the data from one search term could be considered a representative of all of the search results since initial search results are the same.

7 We use the estimation package developed by Abrigo, Love, & others (2015).

8 MAIC suggests a second-order PVAR model. Including a second-order lag does not change our main results.

9 World Health Organization Global Urban Ambient Air Pollution Database (WHO AAP) http://www.who.int/phe/health_topics/outdoorair/databases/cities/en/