**1. Introduction**

According to the World Health Organization (WHO), 91% of the world's population lives in areas where air pollution exceeds the safety limits [1]. Air pollution negatively affects both human's daily life, such as emotional and physical health, and sustainable economic growth, such as labor productivity and tourism (e.g., [2–5]). However, air pollutants are not always visible, which might lead to the public being unaware of pollution. Prior studies have indicated that the public perception of air pollution might be inconsistent with objective air quality, which was evaluated using scientific indices such as PM2.5 and PM10 [6]. A potential reason is that individuals' perceived air quality could differ on the basis of their sociodemographic status, including gender, age, education, knowledge, and health status (e.g., [7,8]). Therefore, perhaps there exists a gap between objective and subjective measures of air quality.

There are two important reasons why the relationship between the actual level of air pollution and residents' concern about air pollution should be examined. On the one hand, ignorance or underestimation of the severity of air pollution potentially poses a threat to residents' health since it could increase the probability of long term exposure to air pollutants. For instance, it was reported that 90% of residents in Hong Kong would not stop their outdoor activities in the face of poor air quality [9]. Whether residents' concern about air pollution could correctly reflect the actual pollution level has become a critical question. On the other hand, a high level of public awareness regarding air pollution might contribute to political and social enthusiasm for the enforcement of environmental protection behaviors and policies. Larson [10] reported that the concurrent rise of online platforms in China has become a positive force for environmental data transparency in China. Kay et al. [11] provided further evidence showing that social media are capable of empowering the governmen<sup>t</sup> to respond to the air pollution problem to ensure social stability. More recently, Lu et al. [12] stressed that public concern about air quality might have a more direct impact than perceived air pollution in influencing people's behaviors and the actions of the community, production sectors, and the government. Therefore, how Chinese residents responded to air pollution should be carefully examined. In particular, more studies are needed to analyze the relationship between the actual level of air pollution and residents' concern about air pollution, given the importance of public concern in shaping public health and environmental regulation practices.

The purpose of this study is to examine whether and to what extent actual air pollution is correlated with residents' concern about air quality in Shanghai, China. Shanghai was chosen as the target city in this study since it is one of the largest and most developed cities in China. Although Shanghai has a high level of economic and social development, its air pollution issue is considerable. According to the Shanghai Environmental Bulletin published by the local government, during the year 2018, air quality was classified as "good" on only 93 days. In 2017 and 2016, the numbers were 58 and 78, respectively. By the end of 2018, Shanghai's air pollution was even reported to be worse than that of Beijing [13], which is well known for its air pollution problem. However, it seems that most attention has been centered on air pollution issues in Beijing in recent years. Public opinion on air pollution from Shanghai residents should be examined since it might maintain pressure on governments to roll out environmental regulations to reduce air pollution effectively and efficiently. From another perspective, understanding Shanghai residents' concern about pollution is valuable in providing residents with useful advice about public health and environmental protection.

In this study, actual air quality was measured by the air quality index (AQI) reported by environmental monitoring stations. Residents' concern about air pollution was measured using the Baidu index for the online search keywords "Shanghai air quality". Baidu index data were provided by Baidu, which is the most popular Internet search engine in China. The index was calculated on the basis of the search volume for a specific search item on a daily basis within a specific region. The Baidu index has been widely used to predict public health issues (e.g., [14–17]) and tourism flows (e.g., [18–20]) in China. Similar to its usage in public health and tourism studies, the Baidu index is also applicable in measuring the degree of public concern about air pollution [12]. This point will be discussed in detail later in the literature review section.

Figure 1 shows the logarithmic values of the actual AQI (blue curve) in Shanghai and the Baidu index (yellow curve) for "Shanghai air quality". The sample period spanned from 2 December 2013 to 31 July 2019, covering 2068 days. The 30-day moving average values of the variables are also displayed to more clearly show varying trends. It can be observed that both the AQI and the Baidu index followed a similar cyclical pattern with apparent fluctuations. The correlation coefficient between logarithmic AQI and the Baidu index was 0.432, indicating a statistically significant positive correlation.

Although Figure 1 provides preliminary visual evidence on the relationship between actual air pollution and residents' concern about air pollution, a basic correlation analysis is not sufficient. To better understand the relationship, the following three research questions are proposed. (1) To what extent and how soon could the actual air pollution level influence public concern about air quality? (2) Does a decline in air quality in another major city, such as Beijing, affect public concern about air quality in Shanghai? (3) Could public concern about air quality and the level of actual air pollution reciprocally influence each other? A vector autoregression (VAR) model was applied to answer these research questions due to its high ability to capture linear interdependencies among multiple variables over time.

It is expected that this study could add to the air pollution literature by examining the reciprocal interactions between public concern about air pollution and the actual degree of air pollution. From a practical perspective, it could help the municipal governmen<sup>t</sup> of Shanghai better understand the degree of public concern about air quality and better assess the current environmental managemen<sup>t</sup> practices. This study also provides insights into the spillover effects that the actual air pollution in other major tourist cities in China might have on public concern about air quality in Shanghai.

**Figure 1.** Actual air pollution and residents' concern about air pollution in Shanghai from 2 December 2013 to 31 July 2019.

The rest of this paper is organized as follows. Section 2 presents a literature review and develops the hypotheses. Section 3 describes the empirical model and the data used in the analyses. The estimated results of the empirical model are reported in Section 4. Section 5 discusses the main findings and implications of the results. Section 6 concludes the paper and talks about limitations and directions for future research.

#### **2. Literature Review and Hypothesis Development**

#### *2.1. Air Pollution: Actual Level, Perceived Level, and Public Concern*

Actual air quality levels are measured by scientific techniques, and they are reported in the forms of different air pollutant indicators such as PM2.5 (particulate matter with a size of 2.5 micrometers or less) and SO2 (sulfur dioxide), or comprehensive indices such as AQI and API (air pollution index) constructed using the air pollutant indicators. Generally speaking, there is not much controversy on the proper measurement of actual air quality level, and scientists can measure it objectively and accurately

People's perception of air quality is subjective. Traditionally, researchers have used questionnaire surveys to collect data on perceived air quality level and people's displeasure with air pollution (e.g., [21–23]). Some of the literature reported that perceived and actual levels of air quality are strongly correlated. For example, Atari et al. [24] conducted a community health survey in the "Chemical Valley" in Sarnia, Ontario, Canada, and they observed a significant correlation between odor annoyance scores and modeled ambient pollution. Peng et al. [22] analyzed the data for over 5000 valid respondents in 62 cities, on the basis of the China Social Survey 2013, and reported a congruence of perceived and actual air pollution. Similar findings were reported by some other studies, such as for Northeast England [25], for Switzerland [26], and for China [27].

However, some of the literature found that perceived air quality is not strongly related to actual air quality. For instance, Egondi et al. [21] focused on two slums in Nairobi, Kenya, on the basis of a cross-sectional study of 5317 individuals. They reported that the majority of respondents were exposed to air pollution. However, the perceived air pollution level was low among residents. This implied the need for promoting public awareness on air pollution. Focusing on the city of Wuhan in China, Guo et al. [7] reported that most persons they surveyed believed that air quality had become worse, though statistics of measured air pollutants over the period of 2010–2014 did not actually show that trend. During the period of 2005–2006, Semenza et al. [23] surveyed the residents of two U.S. cities, Portland and Houston. They reported that the residents' perception of poor air quality was not related to PM2.5 or ozone indicators. On the basis of a sample of 200 people interviewed in London in 1999, Williams and Bird [28] reported that public perception of air quality was not a reliable indicator of actual air pollution levels in the investigated areas. The inconsistency between actual and perceived level of air quality was also reported by Berezansky et al. [29] for Haifa, Israel, by Brody et al. [30] for Texas in the U.S., and by Kim et al. [31] for Seoul, South Korea.

A notable recent study was by Lu et al. [12]. They used the correlation-analysis technique to study the relationship between actual PM2.5 concentration and public concern about haze in five large Chinese cities for the period between 2013 and 2017, where public concern was measured by the Baidu index. They reported that short term fluctuations in actual pollution level and concern about pollution were strongly and positively correlated, but long term annual trends of these two variables were opposite.

Overall, the literature has not provided a consensus on whether the public could correctly perceive air quality and whether public concern about air pollution concurs with the actual level of air pollution. There are two possible explanations of this disagreement in the literature. First, different studies were based on different samples. Since social and individual characteristics strongly affect people's viewpoints [31–33], it is natural to see that perceptions of some groups are consistent with actual air quality, but some other groups tend to over- or under-estimate the level of air pollution. Second, public perceptions vary over time. Along with the development of information technology, improvements in education, and the propagation of mass media, the public receives increasing levels of information on the current situation and importance of air quality. Therefore, many people might not have perceived and been concerned with air pollution problems in the past, but they have changed their minds in recent years.

In this study, we focus on the city of Shanghai. Shanghai is one of the most important and developed cities in China. In 2018, Shanghai's GDP per capita exceeded 20,000 U.S. dollars, which was close to that of Greece and Slovakia. Shanghai has a large population size of over 24 million residents, close to the population of a middle sized country such as Australia. As the financial center of China, many companies have their headquarters in Shanghai. Shanghai has good educational resources, including 64 universities. As an international city, Shanghai is visited by approximately nine million (person-times) inbound tourists annually. Therefore, on average, Shanghai residents have the

knowledge and ability to perceive the importance of clean air and to worry about pollution if they feel that air quality has declined. Accordingly, the first research hypothesis in this study was established:

#### **Hypothesis 1.** *Deprivation of air quality in a local area raises the public concern about the air pollution problem.*

Furthermore, since the Internet has allowed information to flow swiftly across regions, news about air pollution in one major city could quickly propagate and might cause mental or even physical responses of residents in another city. In this study, Beijing, the capital of China, was used as a study area because it is notorious for its severe air pollution (it is also important to note that other major tourist cities, including Nanjing and Guangzhou, are also examined in the robustness check section). If Beijing's AQI rose to an unhealthily high value, the residents in Shanghai might also want to check the air quality in their local area after they receive the information about Beijing. Thus, the second research hypothesis was built:

**Hypothesis 2.** *Deprivation of air quality in another major city raises public concern about the air pollution problem in the local city.*

#### *2.2. Importance of Air Pollution Awareness and Concern*

People's concern about air pollution is largely due to the health risks it has [21,34,35]. People's concern about air quality promotes people's avoidance of the polluted environment. For instance, people tend to limit their social activities on polluted days [36]; people are unwilling to buy houses located in polluted areas [29,37,38]; and tourists flow from polluted to less polluted cities [39,40].

Public concern about air pollution generates pressure on the governmen<sup>t</sup> to implement policies to promote better air quality. Environmental perception and concern also promote more pro-environmental behaviors [41,42]. Tam and Chan [43] further examined the association between environmental concern and behavior across 32 countries (China was not included), and they indicated that the association was weaker in countries with stronger distrust and belief in external control. To better understand the strength of this association in the case of Shanghai, the third research hypothesis was built:

**Hypothesis 3.** *Public concern about the air pollution problem pushes people or governments to take actions to reduce air pollution.*

#### *2.3. Applicability of the Baidu Search Index to Measure Residents' Concern about Air Pollution*

Internet search engines enable quick access to information. Do et al. [44] indicated that tracking online search behavior using relative search volumes was an effective way to gauge public interest. A number of researchers have used the search index as a proxy of issue salience or attention in the public. The majority of search-volume data are extracted from two dominant engines, Google and Baidu. For example, Caputi et al. [45] noticed a significant public interest in heat-not-burn tobacco products on the basis of Google trends data. Do et al. [44] utilized Google trends data to assess public awareness on protected wetlands in South Korea. Mellon [46] found that Google trends data could be used to measure the salience of four issues (fuel prices, economy, immigration, and terrorism) in the United States. Search index data have also been used to predict tourism flows (e.g., [18–20,47]).

Given the fact that the Baidu search engine occupies a dominant market share in China, the Baidu search index has become a useful proxy for Chinese residents' interest and concerns. In Li et al. [16], Baidu search query data were found to be a reliable indicator for monitoring and predicting the HIV/AIDS epidemic in China. Liu et al. [48] used the Baidu index to assess people's awareness of avian influenza A(H7N9) in Zhejiang, China. Using the Baidu index, Yang et al. [49] found that the daily average PM2.5 concentration had a weak impact on public awareness of lung cancer risk in China. Lu et al. [12] used the Baidu index to measure public concern about haze in five large Chinese cities. Given that the use of the search index has been extensively applied in multiple disciplines as a method for surveillance, monitoring, and measuring interest on and concern about specific topics, this study adopted Baidu index data to measure residents' concern about air pollution in China.
