(Author: Hu Xiaoli, PhD student, National University of Singapore Business School; Li Zhen, Professor, National University of Singapore Business School, Director of China Business Center; Lin Yupeng, Associate Professor, Business School, City University of Hong Kong)
National University of Singapore Business School Professors and students discussed the impact of air pollution on investor behavior for the first time during a brainstorming session in a literature discussion class. At first glance, the two seem to be unrelated. However, studies on the capital market have shown that changes in the environment may affect investor behavior through the effect on emotions and have an impact on the stock market. For details, please see below:
At the end of 2013, domestic and foreign media continuously reported the air pollution situation in China, and photos and stories of smog on social networking platforms such as Weibo were also quickly reposted. In a brainstorming session in a literature discussion class, we discussed this question for the first time: Does China’s air pollution affect the behavior of investors? At first glance, the two seem to be unrelated. However, studies on capital markets have shown that changes in the environment (such as weather (Saunders, 1993, Hirshleifer and Shumway, 2005), sunshine duration (Kamstra, Kramer and Levi, 2003) , Temperature (Cao and Wei, 2005), etc.), may affect investor behavior through the effect on emotions and have an impact on the stock market.
Investigations and experimental studies have shown that air pollution not only affects physical health, but also affects mood. Air pollution can cause anxiety, annoyance, and even affect people's stock price of works of art (Evans et al., 1988, Jones, 1978, Stone et al., 1979, Rotton, 1983). So, will the impact of air pollution on people's mood be transmitted to China's stock market? In order to study this issue, we used the historical air quality data provided by the Ministry of Environmental Protection of China to conduct a quantitative statistical regression analysis.
About the data
Currently, the Ministry of Environmental Protection website provides real-time and historical air quality data for major cities across the country. Before 2013, air quality was measured by API (Air Pollution Index); since 2013, cities have successively adopted the new air quality standard AQI (Air Quality Index). Although the measurement standards of the two indexes are different, the classification standards of the two indexes are similar. The range of 0-50 indicates good air quality, the range 51-100 indicates good air quality, and the above 100 indicates that the air is polluted. The higher the pollution index, the higher the pollution. The more serious the index exceeds 300, it means the air is heavily polluted, and the upper limit of the index is 500. Table 1 lists the scope and time period of air quality data currently available directly from the Ministry of Environmental Protection website.
Table 1: The air quality data range of the data center announced by the Ministry of Environmental Protection
Standard and measurement frequency
Number of cities covered
New standard air quality data (AQI), hourly
Old standard air quality data (API), dailyJanuary 2014 Month 1-present 161 June 5, 2000-June 4, 2001 42 June 5, 2001-June 3, 2004 47 June 4, 2004-December 31, 2005 84 January 1, 2006 -February 10, 2011 86 February 11, 2011-January 14, 2013 120 January 15, 2013-March 26, 2013 68 March 27, 2013-April 19, 2013 64 April 2013 From 20th to 31st December, 2013 62
Since the implementation time of the new air quality index is relatively short, we mainly analyzed the air quality data under the old standard. We manually downloaded all the air quality data from 2000.6.5 to 2013.12.31 from the website, a total of 381050 observations. Our research is based on January 1, 2001As the sample period from January to December 31, 2012, in addition to testing the relationship between air pollution and the capital market, we also used these data to analyze and test China’s air pollution from 2001 to 2012 (see appendix).
Stock market returns and stock market air quality
First of all, referring to previous studies on the relationship between air quality and capital markets, we will compare the A shares of the Shanghai and Shenzhen stock markets The daily return rate regressed the air pollution index of the two places on the same day. It was found that after controlling for the impact of the year, month and week, there was no significant correlation between the daily return rate of the stock market and the air quality. At the same time, the air quality When the pollution index is on both sides of the pollution threshold (100), there is no significant difference in the return rate of the stock market.
The above results seem to indicate that the haze has not had a significant impact on investors' investment behavior. But this may also be related to the failure of the Air Pollution Index to measure how people perceive air quality. On the one hand, existing studies have pointed out that air quality reports may be manipulated. For example, our data also shows that there are frequency breakpoints in air quality reports (see appendix), which may affect people’s confidence in the air quality index; on the other hand, Research on the impact of air quality on people’s psychological effects points out that sometimes what people “perceived” the degree of air pollution will play a role (Zeidner and Shechter, 1988). According to the "relative happiness theory" in sociology, people's perception of their own happiness comes from comparison, including comparison with historical stages and comparison with others. Do people compare the local air pollution index with other cities to get a judgment on the degree of pollution? Starting from this idea, we subtracted the air quality of the capital Beijing from the local air quality to calculate the "relative air pollution" indicator. This approach may seem arbitrary, but it is not without its merits. First, this is a very simple method of comparison. Beijing’s air pollution index is often listed at the top of the report, so people can easily compare it. On the other hand, the media is very concerned about Beijing’s air quality. It is also in the area with the most serious air pollution in my country. If the local air pollution index is higher than Beijing, it is easy to give people the impression of "very serious air pollution".
We found that although Beijing’s air pollution is generally more serious than Shanghai and Shenzhen, about 30% of all trading days in Shanghai are more serious than Beijing’s air pollution, while the proportion in Shenzhen is 16.7%; At the same time, although the absolute air pollution index has declined in recent years (see appendix), there is no similar trend in relative air pollution. Figures 1 and 2 show the distribution of air quality in Shanghai and Shenzhen relative to Beijing by quarter and year.
Number of days in Shanghai relative to Beijing’s air quality by quarter
By quarter Dividing the number of days when the air quality in Shenzhen is better or worse than that in Beijing
The proportion of days when the local (Shanghai) air pollution is more serious than Beijing
Percentage of days where local (Shenzhen) air pollution is more serious than Beijing
Using relative air pollution measurement, we found that after controlling the impact of year, month and week, air pollution There is a significant negative correlation between the relative value of and the market return of A-shares in Shanghai and Shenzhen stock markets. The more serious the air pollution in the stock market is relative to that in Beijing, the lower the daily return of the stock market. This relationship is also true when we use the weekly and monthly average stock returns and the relative air pollution index to test. The relative air quality has a considerable impact on the market value of the stock market: regression analysis shows that after the local pollution index rose by 10 points relative to Beijing, the stock market return fell by an average of 0.008%, which is about 1/4 of the average daily return of the stock market during the sample period. Based on the total A-share market value of Shanghai and Shenzhen (15 trillion and 3.5 trillion respectively) as of the end of 2013, the 0.008% rate of return has fallen. The date has caused the Shanghai and Shenzhen A-share markets to shrink by RMB 1.2 billion and RMB 30 million respectively.
The impact of relative air pollution on stock market returns
When local air pollution is slightly lower than Beijing and slightly higher than Beijing, although The difference between the two is very small. If people define the two in terms of "better than Beijing's air" and "worse than Beijing's air," the emotional response may be significantly different. We found that when the local air pollution index changes from slightly lower than Beijing to slightly higher than Beijing, the daily return of the stock market will drop significantly. When the relative air pollution is in the range of -15 to 15 as an example, the average stock market in this range Daily return is-0.06%, and when the pollution index exceeds Beijing, the average daily return will be reduced by 0.47%, corresponding to the decline in the market value of Shanghai and Shenzhen to about 70 billion and 16 billion yuan respectively.
Local air pollution exceeds Beijing’s impact on stock market returns
Local air pollution Exceeding the impact of Beijing on stock market returns
In addition, we also found that when relative air pollution rises, the trading volume of A shares in the Shanghai and Shenzhen stock markets will also drop significantly.
Appendix: Overview of Air Pollution
Based on the data we collected, we calculated the air pollution index from January 1, 2001 to December 31, 2012 The data during the date was statistically analyzed. For the sake of space and representativeness, we mainly analyze the air pollution index of 32 cities in the provincial capital and Shenzhen.
The closer the distance, the more relevant the air quality
Through the correlation analysis of the air pollution index of the two cities, we found that the correlation coefficient exceeds 0.5 Cities are usually relatively close. For example, Tianjin and Shijiazhuang have a correlation coefficient of more than 0.5 with Beijing's air pollution index; Nanjing and Hangzhou (0.6577) have a correlation coefficient with Shanghai of more than 0.5; Guangzhou and Haikou have a correlation coefficient with Shenzhen of more than 0.5. The air pollution index of similar cities is highly correlated, which also shows that the air quality analysis of provincial capital cities is representative.
Regional distribution of air pollution: the north is more serious, the south air is better
Seasonality of air pollution: the most serious in the first and fourth quarters, and the lightest in the third quarter
Finding four: the air pollution index has decreased Trends
From the average pollution index, the air pollution index of most cities decreased significantly in the second half of the observation period (except Hefei and Haikou). The proportions of Beijing, Shanghai and Shenzhen whose annual air pollution index exceeds 100 (pollution) have also shown a downward trend during the observation period.
We have performed an autoregression on the air pollution index. This analysis helps us understand the predictability of air pollution. Taking Beijing, Shanghai and Shenzhen as examples, we found that the air pollution index of these three cities on the day was significantly positively correlated with the air pollution index of the previous day. After controlling the air pollution index of the previous day, the previous air pollution index of Beijing and Shanghai Two days, the first three days of the air pollution index has no additional predictive effect, and the first two days and the first three days of Shenzhen's air pollution index and the air pollution index of the day are still correlated, but the degree of correlation is relatively weak. In terms of predictability, Beijing’s air pollution has the lowest predictability, with the model’s interpretation strength being only about 24%, Shanghai’s air pollution’s predictability is slightly higher, the model’s interpretation strength is about 32%, and Shenzhen’s air pollution is the most predictable. More than 50%.
Air quality report: frequency has a breakpoint
Some studies have pointed out that my country’s air quality report may have a downward trend at the critical value of pollution. Manipulation (Andrews, 2008, Chen, Jin, Kunar and Shi, 2012, Ghanem and Zhang, 2014). Our sample period was longer than the above-mentioned study, and similar evidence was also found. For example,
We drew a frequency chart of Beijing’s daily air pollution index from 2001 to mid-2012:
The gray square in the figure represents the number of days when the air pollution index is greater than or equal to 96 and less than or equal to 100; the white square represents the number of days when the air pollution index is greater than or equal to 101 and less than or equal to 105, and the frequency of the gray square is 3.5 times that of the white square. We used the PM2.5 index published by the US consulate to draw a similar frequency chart (not shown in this article). The data period is from April 8, 2008 to December 31, 2012. It should be noted that the US consulate’s PM2.5 The index is different from the air pollution index published by my country in the calculation method and measurement location, but also uses 100 as the critical point of pollution. In addition, the data of the US consulate is published every hour. For the above reasons, it is not possible to directly combine the two sets of data. ofThe frequency distribution is compared, but we hope to provide independent source data for reference. We found that the frequency distribution of the PM2.5 index published by the US consulate has no obvious breakpoints in the entire range from 21 to 300. The two sets of data have different time spans. We also drew the frequency chart of the Beijing Air Pollution Index from April 8, 2008 to December 31, 2012 (not shown in this article). The frequency gap still exists.
Saunders, E. 1993. Stock price and wall street weather. The American Economic Review 83, 1337-1345.
Hirshleifer, D ., and T. Shumway. 2003. Good day sunshin: stock returns and the weather. The Journal of Finance 58,1099-1032.
Kamstra, M., L. Kramer and M. Levi. 2003 . Winter blues: a sad stock market cycle. The American Economic Review 93,324-343.
Cao, M and J. Wei. 2005. Stock market returns: a note on temperature anomaly. Journal of Banking and Finance 29, 1559-1573.
Evans, G., S. Colome and D. Shearer. 1988. Psychological reactions to air pollution. Environmental Research45, 1-15.
Hu, X., OZ Li., and Y. Lin. 2015. Particles, Pollutions and Prices. Working paper. Available at SSRN:
Jones, J. 1978. Adverse emotional reactions of nonsmokers to secondary cigarette smoke. Environmental Psychology & Nonverbal Behavior 3, 125-127.
Stone, J., S. Breidenbach and N. Heimstra. 1979. Annoyance response of nonsmokers to cigar ette smoke. Perceptual and motor skills 49, 907-916.
Rotton, J. 1983. Affective and cognitive conse of malodorous pollution. Basic and Applied Social Psychology 4, 171-191.
Zeidner, M and M. Shechter. 1988. Psychological responses to air pollution: some personality and demographic correlates. Journal of Environmental Psychology 8, 191-208.
Andrews, S. 2008. Andrews S Q . Inconsistencies in air metrics:\'Blue Sky\'days and PM10 concentrations in Beijing. Environmental Research Letters 3. 1-14.
Chen, Y., GZ Jin., N. Kunar and G. Shi. 2012. Gaming in Air Pollution Data ? Lessons from China. The BE Journal of Economic Analysis & Policy 12. 1935-1682.
Ghanem, D and J. Zhang. 2014. Effortless Perfection: Do Chinese Cities Manipulate Air Pollution Data?. Journal of Environmental Economics and Management 68. 203-225.
For the original text, please refer to: Particles, Pollutions and Prices, Hu, Li and Lin,