Air pollution, investor behavior and firm value：
evidence from China
MSc International Financial Management
Zexian Xu S3168727 Supervisor: Dr. S. Homroy Co-Assessor: Dr. M. A. Lamers June, 2018 Abstract
1. Introduction
Traditional finance sees market participants as homo economicus who make rational decisions for their best interests and have the ability to process all the available information. It ignores the nature of homo sapiens whose decisions are usually driven by their emotions and feelings (Baker and Nofsinger, 2010). Whereas behavioral finance sees the assumption of homo economicus is false, suggesting investors are irrational who may make irrational decisions. It combines psychology with finance to study the effects of human behavior on assets pricing.
Among the large body of behavioral finance literature, researchers have documented the effects of a variety of mood variables on stock returns. Saunders (1993) finds that stock returns are higher in sunny days and lower in cloudy days, while the impact of totally sunny days is more pronounced. Temperature is also found to be negatively correlated with stock returns as higher temperature induces apathy which diminishes investors’ risk appetite (Cao and Wei, 2005). Kramstra, Kramer and Levy (2000) document a daylight saving effect, suggesting stock markets experience losses after the change of the time, indicating that though the one-hour time loss is minor, it can induce anxiety and errors in judgement which impact stock prices. Other factors such as wind, seasonal affective disorder are also found to affect stock returns (Kamstra, Kramer, and Levi, 2003; Shu and Hung, 2009)
This thesis investigates the effect of air pollution on investor behavior and the value of firms which are exposed to it, focusing on the Chinese stock market. Air pollution has received great attention from the Chinese society in recent years due to the increasing frequency and intensity of the pollution among the whole country. Despite an extensive body of Chinese literature gives introduction of the air pollution and the strategies to prevent and overcome it, the studies about its financial impacts are quite scarce. Therefore, the study of interest could contribute to the scarce researches for the Chinese stock market.
The typical type of air pollution in China is the smog. It appears when the pollutants such as 𝑃𝑀#.%, 𝑃𝑀&', 𝑆𝑂#, 𝑁𝑂#, 𝑂+ and 𝐶𝑂1 exceed the standard level, among which 𝑃𝑀
#.% is considered the most direct and important driver for smog. Air pollution is caused either by natural environment such as dust, forest fires and volcanic ash, or by human activities such as car exhaust and burning fossil fuels. Polluted air threatens human health and impairs mood. The toxic substances could penetrate
1 𝑃𝑀
#.% is fine particulate matter that has a diameter of less than 2.5 micrometers; 𝑃𝑀&' is the particulate matter that has a diameter
deep into respiratory tract and lungs which cause illness such as acute bronchitis and acute rhinitis. They also induce lung cancer if people are exposed in such environment chronically (Stafford, 2015). Air pollution has significant impacts on the economy. The mortality caused by heavy air pollution in three major Chinese cities has reached 7000 in 2010, resulting in a total loss of 5.5 billion Chinese yuan (Zhang, 2017). During the days of heavy air pollution, the visibility on the road is minimized which causes problems of public transportation. Air pollution is found to reduce worker productivity, both outdoor and indoor workers, since some elements could easily enter buildings (Chang, Zivin, Gross, and Neidell, 2016b; Zhang and Jin, 2017; Zivin and Neidell, 2012). The impaired worker productivity may influence firms’ profitability.
Some researchers have documented a negative impact of air pollution on stock market returns (Hayes, Neidell, and Saberian, 2016; Levy and Yagil, 2011, 2013; Li and Peng, 2016), but their studies focus on the market index, instead of individual stocks. This thesis takes individual stocks as the object, studying how exposure to air pollution could influence firms’ value. It may provide some insights to firms since they may bear additional cost of capital. The underlying mechanism of air pollution on firm value is as follows. First of all, Other than the health effects of air pollution, it induces anxiety, depression, annoyance and aggression (Evans, Colome, and Shearer, 1988; Rotton, 1983). Risk aversion increases when people are in bad mood which influences the evaluation of prospective risks. Cognitive problems also arise. People tend to make decisions by comparing choices in isolation rather than as a group (Hayes et al., 2016). The increased risk aversion results in a shift of investment from stocks to less risky assets. Secondly, though the direct effects of air pollution are on local investors, investors are found to subject to home bias. They hold more local stocks in their portfolios (Grinblatt and Keloharju, 2001; Huberman, 2001). Therefore, the value of local firms can be affected during the days of heavy air pollution. Thirdly, since investors have strong tendency of herding (Banerjee, 1992), the impacts can be aggravated.
contrary, state ownership could result in more negative impacts as firms with state ownership is correlated with diminished corporate governance and transparency (Gul, Kim, and Qiu, 2010). They are more likely to hire small local audit firms in order to manipulate earnings to achieve operating performance (Hou, Kuo and Lee, 2012). Meanwhile, government may suppress information which undermines political dealings or reveal political corruption (Cannizzaro and Weiner, 2018). Cross-listing is also expected to aggravate the negative impacts. Though there are quite a few benefits of cross-listing, they are not permanent (Sarkissian and Schill, 2009). The volatility increases and liquidity decreases in the home market after cross-listing and domestic investors tend to react to the difference between home market and destination market (Podpiera, 2001).
This thesis employs an event study to examine the change of firms’ value during the days of heavy air pollution. The sample consists 88 heavy air pollution events, 74 firms located in 26 cities across China and a total of 296 observations, from December 2013 to December 2016. The results show a significant loss of 0.44% on the event day, and a cumulative loss of 0.75% in the 3-day event window. Furthermore, the cross-sectional analysis suggests firms with state ownership and are cross-listed in foreign markets experience more negative impacts on firm value, whereas no evidence is found for the mitigation effects of foreign institutional ownership due to relatively small shareholdings. Meanwhile, larger firms and firms with lower leverage have smaller negative abnormal returns, and firms in heavily polluted cities suffer significantly more loss than those located elsewhere.
2. Literature review and hypotheses
2.1 Air pollution and stock returns
2.1.1 The effects of air pollution on health and mood
Air pollution has immediate effects on physical health. It irritates eyes and noses and causes symptoms such as dizziness, nausea and fatigue, more seriously, inducing asthma, respiratory diseases, skin rash, fever and angiocardiopathy (Seaton, MacNee, Donaldson, and Godden, 1995; Stafford, 2015). These symptoms can arise shortly after exposure and may last for several days. On the other hand, the increased level of air pollution is associated with greater anxiety, depression, annoyance and aggression (Evans et al., 1988; Rotton, 1983). Chronic exposure to air pollution increases one’s stress perception, and impairs mood and emotion (Bullinger, 1989). Increased visits to psychiatric unit and psychiatric related 911 emergency phone calls reflect the significant damage of air pollution on mental health (Oudin et al., 2018; Rotton and Frey, 1984). From the view of physiology, polluted air and perceived stress increase the level of cortisol in human body. The normal level of cortisol has powerful cognitive and emotional effects, however, the excessive cortisol dampens immune system, alters mood, induces anxiety, increases risk aversion and impairs the ability to make rational decisions (Coates and Herbert, 2008; Tomei et al., 2003).
which reduces the daily number of calls handled per worker with the increasing air pollution level. Zhang and Jin (2017) test the effect of air pollution on worker productivity by using the level of PM2.5. They document an overall of 0.14% decrease in productivity with one unit increase in PM2.5. Moreover, they find the impact is more pronounced to highly skilled workers such as managers and technical staff than front-line workers since highly skilled workers require both physical and cognitive support.
2.1.2 Mood and decision making
Psychologist have studied the effects of a wide range of mood and emotions on decision making. As Loewenstein (2000) document that visceral factors drive people to react differently to situations compared with how they normally behave that may affect their future welfare. Extreme actions can be taken under strong emotions. The mood-as-information theory states that mood can be regarded as a kind of information to influence one’s judgements and thus guide decision making (Schwarz and Clore, 1988). Lucey and Dowling (2005) develop a mood misattribution theory, stating that mood can even affect the decisions which are not related to the cause of the mood, for example, buying or selling of stocks. Risk aversion increases when people are in bad mood, which influences the evaluation of prospective risks (Slovic, Finucane, Peters, and MacGregor, 2007). As bad air quality causes health problems and induces bad mood, investors’ risk appetite is likely to decrease and the assessment of risk of future events may be biased. Moreover, Forgas (1995) documents that mood has stronger impacts on one’s judgement when the information is limited. The investment decision in stock market is relatively complex and risky which leads to more reliance on one’s mood during the decision making.
2.1.3 Empirical evidence of mood on stock returns
Cao and Wei (2005) study the temperature effects on stock returns on a world-wide base and find that stock returns are lower when the temperature gets higher as higher temperature induces apathy which diminishes risk appetite. Other studies provide evidence of the effects of mood-related variables on stock returns such as wind, daylight saving and seasonal affective disorder (Kamstra et al., 2000, 2003; Shu and Hung, 2009)
Regarding the evidence of the effects of air pollution on stock returns, Levy and Yagil (2011) show that the stock index returns are lower under unhealthy condition than healthy condition, and the negative relationship is weakened when the stock exchange is far away from the polluted area. Levy and Yagil (2013) further extend their research into several other countries such as Canada, China, Australia and Netherlands. They find that the results hold for those countries. In addition, they document that pollution-related companies in USA and Canada bear more negative effect. Hayes, Neidell and Saberian (2016) document that air pollution affects stock market returns by the change of investors’ risk aversion. They find a significant positive correlation between pollution level and risk aversion. Investors tend to switch their investment to lower risky stocks with lower returns due to increased risk aversion induced by air pollution. Li and Peng (2016) study this relationship by taking the average air quality of 16 big cities in China. They find a strong negative relationship between the level of air pollution and stock returns, and this relationship is even stronger after 2013 when the Chinese government implement a stricter measure and policy of air quality. They also find a negative two-day lagged effects of air pollution on stock returns which implies the effects of pollutants on mood does not disappear immediately.
2.1.4 Investor behavioral biases in stock market
they further identify that firm’s geographical proximity with where fund managers located is one of the dominant factors of their investment decisions. After analyzing the shareholding of 7 regional bell operating companies across 48 states, Huberman (2001) documents that local investors hold overwhelming more shares of local bell operating companies than those located in other states, and the amount of capital invested in local companies are much more than non-local companies. All these studies provide evidence that local investors have strong preference for local stocks in their portfolios which may indicate the impact of local investor mood on stock returns of local firms.
Moreover, Chinese stock market is comprised of a large proportion of individual investors, they still are the biggest trading body and capital provider (Tan and Chen, 2012). Banerjee (1992) shows that individual investors have the strong tendency of herding. Herding reflects the behavior bias that individual investors have when making investment decisions by following others’ strategy, or imitating others’ behavior, expecting to benefit from others’ superior information, instead of making decisions based on their own information. The significant herd behavior has been observed in Chinese stock market (Cheng, Jiang, Chen, and Wu, 2004; Li and Zeng, 2005; Yang and Zhi, 2006). Operating under frequent intervention of the government and relatively higher information asymmetry, Chinese stock market presents a severer herding than other developed markets.
To sum up, the negative returns of local stocks are expected around the days of heavy air pollution through the following pattern: air pollution affects the mood of local investors with increased risk aversion, pushing them to shift their investment to less risky assets, which lowers the stock returns of local firms. The negative impacts could be aggravated by herding in the stock market.
𝐻&: The reaction of stock prices of Chinese local firms to air pollution is negative.
2.2 Factors influencing air pollution-induced abnormal returns
In the following section, I will discuss the possible firm level determinants of air pollution-induced abnormal stock performance.
2.2.1 Foreign institutional ownership
return volatility (Chen, Hong and Stein, 2002; Gul et al., 2010). For Chinese listed firms, the information they disclose should satisfy both domestic and foreign investors when absorb foreign capital, thus facilitates the transparency of the information. Air pollution, therefore, has less power to induce the negative expectation of domestic individual investors to firms in polluted cities since those investors obtain enough relevant information. On the other hand, the high threshold for foreign institutional investors to enter the Chinese stock market assures those who qualified are large in scale, highly skilled and have adequate experience and long investment horizons (Li and Han, 2013). Compared with individual investors, they have better evaluation of firm value based on the ability to obtain and process information. Overvalued stocks are sold when markets are rising whereas undervalued stocks are bought when markets are falling which could help to stabilize stock prices (Lipson and Puckett, 2006). Moreover, concerning the high transaction costs of foreign stocks, foreign institutional investors are less likely to change their portfolio frequently. Meanwhile, they also get trust from and provide great guidance to domestic investors who follow their investment decisions (Li and Han, 2013). Therefore, foreign ownership is expected to reduce the negative abnormal returns around the days of heavy air pollution.
𝐻#: Foreign institutional ownership could reduce the negative abnormal returns induced by heavy air pollution
2.2.2 State ownership
corruption, government is likely to incentivize management to suppress such information. The diminished corporate governance and information environment may exacerbate the mispricing of stock prices during the days of heavy air pollution since investors obtain limited information. Moreover, a large proportion of the management in those firms is comprised of government officials or former politicians with the lack of business expertise that leads to poorer financial performance (Zou and Adams, 2008). Therefore, investors may sell more of these stocks in the days of heavy air pollution due to relatively poorer financial performance.
𝐻+: State ownership is negatively related to heavy air pollution-induced abnormal returns.
2.2.3 Cross listing
There is an increasing number of Chinese firms cross-listed on foreign stock exchanges over the recent decades. The benefits of cross-listing have been substantially discussed by researchers. According to Pagano, Roell and Zechner (2002), cross-listing could overcome market segmentation which caused by investment barriers of different markets. The removal of the investment barriers implies that investors could diversify their portfolio and disperse the investment risks which result in higher stock prices and lower expected returns. Cross-listing also enables firms to access foreign capital market and seek for lower cost capital sources. Furthermore, same as foreign ownership, cross-listing improves firms’ information environment if the chosen foreign market is strictly regulated, such as strict requirements for disclosure, corporate governance and investor protection. However, the benefits are not permanent (Sarkissian and Schill, 2009). In the case of emerging markets where the markets are opaque or semi-opaque, the price volatility increases and the liquidity decreases after cross-listing, especially for those stocks open to foreign ownership. This is due to order flow migration which is the transfer of traders to the markets which are more transparent, considering the existence of market segmentation (Domowitz, Glen, and Ananth, 1998). Podpiera (2001) further supports this finding. He supplements that investors always react to the pricing differences between local markets and foreign markets. Therefore, I expect Chinese investors react more negatively to those cross-listed stocks to avoid potential risks.
3. Methodology
3.1 Event study
In order to examine the change in the value of firms exposed to air pollution, an event study is applied. According to MacKinlay (1997), estimation window normally do not overlap event window to avoid estimation bias for parameters, and the event window is often expanded to multiple days surrounding the event day to capture the potential effect of events. Therefore, the event window is set to be [-1,3], considering the forecast of air pollution, as well as the effect of air pollution on mood could last at most for four days (Bullinger, 1989). Day 0 is the event date when the heavy air pollution occurs. The estimation window is set to be 150 days prior to the event window. The market model is employed to obtain abnormal returns:
𝐴𝑅34 = 𝑅34− 𝛼3 − 𝛽3𝑅94 (1)
where 𝐴𝑅34 and𝑅34 are the abnormal return and return for stock i at time t, respectively; 𝛼3 and 𝛽3 are the parameters calculated through the 150-day estimation window; 𝑅94 is the corresponding market return at time t, which is the return of Shanghai A share index. The returns are calculated by taking the natural logarithm of today’s closing price divided by yesterday’s closing price:
𝑅34 = ln <=>
<=>?@ (2)
Test statistics
Adjusted standardized cross-sectional test
event-induced variance but prone to cross-sectional dependence. The test statistics are all based on standardized abnormal returns (SARs). For detailed steps please refer to Appendix 1.
The test statistic is:
𝜃& = BBC> D
E ∗
&GH
&I(DG&)H (3) where 𝐴𝐴𝑅4 is the average standardized abnormal returns at time t; N is the sample size; 𝑟 is the average cross-correlation of excess returns in the estimation window. The test statistic for cumulative average standardized abnormal returns is:
𝜃# =MBBCEN D∗ &I(DG&)H&GH (4) where 𝐶𝐴𝐴𝑅O is the cumulative average standardized abnormal returns at event period 𝝉; N is the sample size; 𝑟 is the average cross-correlation of excess returns in the estimation window.
Generalized rank test
A generalized rank test is employed to check the robustness of the parametric test results, since the above parametric test assumes the abnormal returns are normally distributed. The generalized rank test is developed by Kolari and Pynnonen (2011). It immunes to cross-sectional dependences of excess returns and event-induced volatility. It is also based on SARs. This method ranks SARs both in the estimation window and event window. For detailed steps please refer to Appendix 1. The final rank is conducted as follows:
𝑈3O =CBDR SEBC=>
TI& − 0.5 (5) where 𝑅𝐴𝑁𝐾 𝐺𝑆𝐴𝑅34 is the rank of generalized abnormal returns which takes 𝑆𝐴𝑅 in the estimation window and takes standardized cumulative abnormal returns for cumulative event day in the event window as the cumulative event day is considered to be one time point. T is the length of the estimation window plus one cumulative event day. The test statistics for zero abnormal returns is:
𝑈4 = D&
> 𝑈a4
D>
3b& (9) where 𝑇 is the estimation window plus one cumulative event day; 𝑁4 is the number of GSARs for the sample at time point 𝒕, including both estimation window and cumulative event day; 𝑁 is the sample size. 𝑈' is the 𝑈4 at cumulative event day.
3.2 Cross-sectional analysis
I further examine whether foreign institutional ownership, state ownership and cross-listing influence the value change in heavy air pollution events since the magnitude of value change may differ under different characteristics. I use Ordinary Least Squares (OLS) (MacKinlay, 1997) to regress CARs on those firm characteristics to identify the determinants of the abnormal returns. The heteroscedasticity and autocorrelation consistent (HAC) standard errors are used for the possible autocorrelation and heteroscedasticity of the residuals. The regression model is:
𝐶𝐴𝑅34=𝛽'+𝛽&𝐹𝐼𝑂34+𝛽#𝑆𝑇𝑂34+𝛽+𝐶𝑅𝐿34+𝛽0𝑆𝑖𝑧𝑒34+𝛽%𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒34+𝛽n𝑅𝑂𝐴34+𝛽o𝑃𝑜𝑙𝑙𝑢𝑡𝑒𝑑 𝑐𝑖𝑡𝑦34+ 𝑌𝑒𝑎𝑟 𝑓𝑖𝑥𝑒𝑑+𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑓𝑖𝑥𝑒𝑑+𝜀34 (10)
The dependent variable CAR is the 3-day cumulative abnormal return. It is chosen as the proxy of firm value since the returns are the direct measure of shareholder wealth and investors’ future expectations. It reflects the stock market reaction around the days of heavy air pollution. Though there is no direct literature provides references to the measure of the air pollution-induced change in firm value, other studies in finance such as M&A effects on acquirers’ value largely use excess returns (CARs) as the proxy of firm value to examine the potential determinants of the change in value (Du and Boateng, 2015). The advantage of stock returns is that it is a relatively unbiased proxy compared with other objective measures since it does not differ under different accounting policies (Du and Boateng, 2015). Therefore, the CARs in the event window [-1,1] is employed as the dependent variable.
Size, leverage, ROA and polluted cities are the control variables. The size of the firms is measured by taking the natural logarithm of total assets. Firm size is negatively related to stock returns as smaller firms are subject to higher risks (Duy and Huu Phuoc, 2016). Firms with large size are also associated with lower stock return volatility (Zou and Adams, 2008). Therefore, investors may react more negatively to smaller firms due to air pollution-induced risk aversion. Moreover, large firms disclose more information than small firms which may mitigate investor pessimism for future profitability of firms exposed to air pollution. The leverage is measured as the ratio of long-term debt divided by total equity. According to Jiang and Wang (2010), the effects of bad mood on abnormal returns is more pronounced to firms with higher leverage ratio since firms with higher leverage ratio increases the probability of bankruptcy. This is consistent with lowered risk propensity under bad mood. ROA is measured by dividing net income by total assets. The higher profitability indicates more dividends distributed to investors which attracts more investors (Hatem, 2015). Therefore, the negative abnormal returns could be mitigated by the profitability of firms. In other words, the higher the profitability, the lower the abnormal returns induced by heavy air pollution. Highly polluted cities indicate that investors suffer more days with poor air quality which enlarges the damage of air pollution on mental and physical health. The symptoms under chronic exposure to air pollution may appear much severer. Moreover, the news about the air pollution of those cities is covered more frequently by the media. Investors may show increased antipathy and pessimism. Hence, I assume firms in heavily polluted cities suffer more negative abnormal returns in the days of heavy air pollution. I mark cities with more than 5 air pollution events as the heavily polluted cities and assign them as 1, and 0 otherwise. The actual days of heavy air pollution are more than the events included in this study due to the merging and exclusion of the events during the data screening.
4. Data
4.1 Event study data
higher the AQI, the heavier the air pollution is. In the days when AQI is larger than 300, there are significant impacts on health. Therefore, I employ the AQI larger than 300 as the event of air pollution. The time span of the study is from December 2013 to December 2016. The AQI is obtained from Aqistudy database (www.aqistudy.cn). Since the China Ministry of Environmental Protection no longer provides historical AQI data of cities, the aqistudy database becomes the most popular and reliable source for AQI. This database provides not only daily AQI, but also other pollutants such as PM2.5, PM10, CO, etc, and covers the majority of cities in China.
The sample firms are derived from Shanghai 180 index component stocks. The index comprises 180 A-share stocks which are the most representative stocks traded in Shanghai stock exchange. There are two types of tradable shares in Chinses stock market, which are A-shares and B-shares. A-shares are denominated and traded in Chinese yuan. B-shares are denominated in Chinese yuan but traded in foreign currencies. A-share market is the main trading market in China while the market of B-shares is relatively small and less liquid (Zhang, 2014). The component stocks are adjusted every six months with maximum 10% of the total amount. The sample firms used in this study is obtained in February 2018 from Shanghai stock exchange. The stock prices of firms are obtained from Datastream.
Table 1 Summary statistics of events by city
This table presents summary statistics of the heavy air pollution events included in the sample by city. The heavy air pollution events are defined as the days when Air Quality Index (AQI) is larger than 300. The first column is the name of the city, followed by the number of events, firms and total observations. The sample period is from December 2013 to December 2016.
City Events Firms N City Events Firms N
Beijing 9 31 196 Lanzhou 3 1 3 Shanghai 1 13 13 Lianyungang 4 1 4 Baotou 1 2 2 Nanjing 2 1 2 Baoding 15 1 15 Ningbo 2 2 4 Chengdu 1 1 1 Qingdao 1 2 2 Ordos 2 1 2 Yantai 1 1 1 Harbin 6 1 6 Changzhi 2 1 2 Hangzhou 1 1 1 Shenyang 3 1 3 Hefei 1 1 1 Tianjin 6 1 6 Langfang 4 1 4 Wuhan 3 2 4 Hohhot 2 1 2 Urumchi 5 2 9 Jinan 3 1 3 Xian 4 2 4 Nantong 1 1 1 Zhengzhou 5 1 5 Total 26 88 74 296
4.2 Cross-sectional model data
cross-listing is obtained from Shanghai stock exchange. Firms are cross-listed in New York, London, Japan and HongKong, among which HongKong is the most popular destination. The data for size, leverage and ROA are obtained from CSMAR. The descriptive statistics of variables are presented in Table 2.
Table2 Descriptive statistics of variables
This table presents the means, medians, maximum values, minimum values and the standard deviations of variables in the regression. CAR is the cumulative abnormal returns in the event window [-1,1], and it is the dependent variable. FIO is the foreign institutional ownership; STO is the state ownership; CRL represents whether a firm is cross-listed or not. FIO, STO and CRL are the independent variables of interest, and they are dummy variables. Size, leverage, ROA and polluted city are control variables. Polluted city is a dummy variable indicates whether the city is heavily polluted.
Variable Mean Median Maximum Minimum Std.Dev.
CAR [-1,1] -0.008 -0.005 0.127 -0.155 0.033 FIO 0.239 0 1 0 0.428 STO 0.243 0 1 0 0.429 CRL 0.237 0 1 0 0.426 Size 24.673 24.636 28.504 20.186 1.599 Leverage 0.424 0.236 2.352 0 0.474 ROA 0.035 0.023 0.193 -0.090 0.041 Polluted City 0.811 1 1 0 0.392 5. Results 5.1 Event study
Panel B in Table 3 shows the cumulative average abnormal returns (CAAR) for 3 different event windows, [-1,1], [-1,2] and [-1,3]. The CAAR for event period [-1,1] is -0.75% at 5%, 10% significance level under generalized rank test and parametric test respectively. The largest CAAR appears for the event period [-1,3] with -0.83%, but only significant at 10% level under generalized rank test.
Table 3 (Cumulative) average abnormal returns
This table presents the average abnormal returns (AAR) and cumulative average abnormal returns (CAAR) of firms in the days of heavy air pollution. The market model is applied. The event window is from one day before the occurrence of the heavy air pollution to three days after, which are 5 days in total. The AARs in the event window are presented in Panel A. The CAARs for event period [-1,1], [-1,2] and [-1,3] are showed in Panel B. The adjusted standardized cross-sectional test is the parametric test, and the generalized rank test is the nonparametric test. The Jarque-Bera test shows none of the abnormal returns and cumulative abnormal returns in the event window are normally distributed. ***, **, and * denote statistical significance at 1%, 5% and 10% level respectively.
Panel A -1 0 1 2 3 AAR -0.13% -0.44% -0.19% -0.02% -0.06% Adjusted standardized cross-sectional T -0.09 -3.97*** -1.56* 0.79 -0.34 Generalized rank T -0.32 -2.92*** -0.71 0.57 -0.62 Jarque-Bera 57.14*** 44.20*** 61.10*** 95.09*** 251.45*** Panel B [-1,1] [-1,2] [-1,3] CAAR -0.75% -0.77% -0.83% Adjusted standardized cross-sectional T -1.60* -0.42 -0.52 Generalized rank T -1.81** -1.26 -1.55* Jarque-Bera 138.90*** 45.42*** 37.87***
5.2 Results of Cross-sectional model
Table 4 Correlation metrics
This table presents the correlation between variables. CAR is the cumulative abnormal returns in the event period [-1,1], and it is the dependent variable; FIO is foreign institutional ownership; STO is state ownership; CRL represents whether a firm is cross-listed or not. FIO, STO and CRL are the independent variables of interest. Size, leverage, ROA and polluted cities are control variables. ***, **, * denote statistical significance at 1%, 5% and 10% level respectively. 1 2 3 4 5 6 7 8 1. CAR 1.000 2. FIO 0.058 1.000 3. STO -0.189*** 0.032 1.000 4. CRL -0.089 0.153*** -0.056 1.000 5. Size 0.011 0.162*** 0.123** 0.553*** 1.000 6. Leverage -0.071 -0.039 0.205*** 0.164*** 0.544*** 1.000 7. ROA 0.069 0.174*** -0.164*** -0.039 -0.215*** -0.415*** 1.000 8. Polluted city -0.172*** 0.009 0.053 0.208*** 0.245 0.157*** -0.135** 1.000
The regression results of CARs on firm characteristics are showed in Table 5. The coefficient of foreign institutional ownership is 0.005, indicating that foreign institutional ownership mitigates the negative abnormal returns induced by heavy air pollution, which is consistent with the hypothesis. However, it is not statistically significant which means the impact is not different from 0. Therefore, I cannot conclude that firms with foreign institutional ownership suffer less loss than those without foreign institutional ownership. One possible explanation for the insignificant result is the relatively small shareholdings of foreign institutional shareholders (Zou and Adams, 2008). Foreign institutional ownership is a dummy variable in this study, but in Zou and Adams’s study, they have the mean of 2.5% for foreign ownership of Chinese listed firms. It may imply that the advantages of foreign institutional ownership are diminished in China. The small shareholdings do not alter the downward stock price movement due to heavy air pollution, and the improvements of the quality of information reported by firms are limited.
information which undermines political dealings or reveal political corruption. Therefore, the information available to investors is limited and altered which exacerbates the mispricing of stock prices under heavy air pollution and results in more negative abnormal returns.
The relationship between cross-listing and CARs is found to be negative and highly significant, suggesting that cross-listed firms have more negative abnormal returns. This can be viewed as the benefits of cross-listing quickly disappear and the stock volatility increases after cross-listing for Chinese listed firms driven by the transfer of capital to the destination market where the environment is more transparent, and domestic investors always react to the pricing difference between home market and destination market. Therefore, investors react more negatively to those high equity risk stocks in the days of heavy air pollution.
Of the control variables, the signs of the coefficients are all consistent with the prediction. The coefficients of size and leverage are positive and negative respectively, both are highly significant, suggesting that firms which are larger in size and lower in leverage have smaller negative abnormal returns. ROA is positively related to abnormal returns but not significant. Firms in highly polluted cities generate more negative abnormal returns since investors are more frequently affected by heavy air pollution both physically and mentally.
5.3 Robustness check
As reviewed in the literature, one might argue that it is other factors such as weather conditions that influence investor mood which lead to the fluctuation in the stock prices and the change in firm value. Following the study of Li and Peng (2016), I regress the 3-day CARs on temperature, humidity, wind speed, Monday dummy and January dummy to test whether these factors are correlated with the negative excess returns in the days of heavy air pollution. The data of event day 0 is taken. The meteorological data is obtained from Weather Underground Corporation (www.wunderground.com). The results are presented in Appendix 3-Table 7.
days of heavy air pollution is indeed caused by the air pollution-induced bad mood, and that bad mood is not caused by other factors.
Table 5 The effect of firm characteristics on firm value
This table presents the estimate of OLS regression with year and industry fixed effects. The dependent variable is the 3-day cumulative abnormal returns obtained from event study. Foreign institutional ownership, state ownership and cross-listing are dummy independent variables, they take 1 when the firms have foreign institutional ownership, state ownership and are cross-listed, and 0 otherwise. Size, leverage, ROA and polluted city are the control variables. Standard errors are in parentheses. ***, **, * denote statistical significance at 1%, 5% and 10% level respectively.
Variables Y=CAR[-1,1]
Predicted sign Coefficient
Foreign Institutional Ownership + 0.0044
(0.0037) State Ownership - -0.0178*** (0.0062) Cross-listing + -0.0145*** (0.0049) Size + 0.0046*** (0.0016) Leverage - -0.0101** (0.0044) ROA + 0.0111 (0.0721) Polluted City - -0.0143** (0.0059) Constant -0.0891** (0.0404)
Fixed Effects Year, Industry
Observations 296
Adjusted R-squared 0.0656
6. Conclusion
This study examines the impact of heavy air pollution on the value of local firms which are exposed to it in the period of 2013-2016. An event study is employed to detect the abnormal returns. By collecting the data of 74 firms dispersed in 26 cities across China, the evidence shows that in general, those firms suffer a loss of 0.44% on the day when the heavy air pollution occurs, and a total loss of 0.75% within the three days around the occurrence of heavy air pollution. The significant 3-day cumulative abnormal returns are further regressed on firm characteristics to examine the determinants of the negative abnormal returns. State ownership is found to result in more negative abnormal returns, as well as cross-listing, whereas foreign institutional ownership does not have impact on the abnormal returns due to relatively small shareholdings of those foreign institutional investors. Larger firms and firms with lower leverage are found to suffer less loss from the heavy air pollution, and firms located in more polluted cities are affected more by the air pollution.
This thesis highlights the importance of local environmental factors and investor behavior in firm value changing. Air pollution could cause significant health problems, cognitive problems, and impair mood. The cognitive problems affect the information processing, and the impaired mood is associated with increased risk aversion. Investors are likely to have biased evaluations of prospective risks under increased risk aversion, and thus shift their investments to less risky assets which lowers the stock returns which represents the value of firms. The local air pollution can affect local firms’ value through its effect on local investors’ mood. Since investors are found to hold a large proportion of local stocks in their portfolio, their reaction to air pollution can be reflected in the local stock prices. And this effect can be aggravated by herding in the stock market.
References
Baker, H. K., Nofsinger, J., 2010. Behavioral finance: investors, corporations, and markets. John Wiley & Sons, New Jersey.
Banerjee, A., 1992. A simple model of herd behavior. Quarterly Journal of Economics, 107(3), 797– 817.
Boehmer, E., Musumeci, J., Poulsen, A., 1991. Event-study methodology under conditions of event-induced variance. Journal of Financial Economics, 30(2), 253–272.
Brown, S. J.,Warner, J.,1980. Measuring security price performance. Jouranl of Financial Economics, 8, 205–258.
Brown, S. J., Warner, J., 1985. Using daily stock returns. Journal of Financial Economics, 14, 3–31. Bullinger, M., 1989. Psychological effects of air pollution on healthy residents-A time-series
approach. Journal of Environmental Psychology, 9, 103–118.
Cannizzaro, A. P., Weiner, R. J., 2018. State ownership and transparency in foreign direct investment. Journal of International Business Studies, 49, 172–195.
Cao, M., Wei, J., 2005. Stock market returns: A note on temperature anomaly. Journal of Banking and Finance, 29, 1559–1573.
Chang, T., Zivin, J. G., Gross, T., Neidell, M., 2016a. Particulate pollution and the productivity of pear packers. American Economic Journal: Economic Policy, 8(3), 141–169.
Chang, T., Zivin, J. G., Gross, T., Neidell, M., 2016b. The effect of pollution on worker productivity: evidence from call-center workers in China. Discussion paper. University of Southern
California, Los Angeles.
Chen, J., Hong, H., Stein, J. C., 2002. Breadth of ownership and stock returns. Journal of Financial Economics, 66, 171–205.
Chen, Y., Hung, M., Wang, Y., 2018. The effect of mandatory CSR disclosure on firm profitability and social externalities: Evidence from China. Journal of Accounting and Economics, 65, 169– 190.
Cheng, X., Jiang, X., Chen, M., Wu, G., 2004. An empirical analysis on the portfolios herd behavior in China stock market. Systems Engineering—Theory & Practice, 12, 34–38.
Coates, J. M., Herbert, J., 2008. Endogenous steroids and financial risk taking on a London trading floor. Proceedings of the National Academy of Sciences, 105(16), 6167–6172.
Crocker, T. D., Horst, R. L., 1981. Hours of work , labor productivity , and Environmental conditions : a case study. The Review of Economics and Statistics, 63(3), 361–368. Domowitz, I., Glen, J., Ananth, M., 1998. International cross-listing and order flow migration :
evidence from an emerging market. The Journal of Finance, 53(6), 2001–2027.
Du, M., Boateng, A., 2015. State ownership, institutional effects and value creation in cross-border mergers & acquisitions by Chinese firms. International Business Review, 24, 430–442.
Duy, N. T., Huu Phuoc, N. P., 2016. The relationship between firm sizes and stock returns of service sector in Ho Chi Minh city stock exchange. Review of European Studies, 8(4), 210–219. Evans, G. W., Colome, S. D., Shearer, D. F., 1988. Psychological reactions to air pollution.
Environmental Research, 45, 1–15.
Forgas, J. P., 1995. Mood and judgment: The affect infusion model (AIM). Psychological Bulletin, 117(1), 39–66.
Grinblatt, M., Keloharju, M., 2001. How distance , language , and culture influence stockholdings and trades. The Journal of Finance, 56(3), 1053–1073.
Gul, F. A., Kim, J. B., Qiu, A. A., 2010. Ownership concentration, foreign shareholding, audit quality, and stock price synchronicity: evidence from China. Journal of Financial Economics, 95, 425–442.
Hatem, B. S., 2015. What determines cumulative abnormal returns? An empirical validation in the French market. International Business Research, 8(12), 89–95.
Hayes, A., Neidell, M., Saberian, S., 2016. The effect of air pollution on investor behavior: evidence from the S&P 500. Unpublished working paper. University of Ottawa, Ottawa.
Hirshleifer, D., Shumway, T., 2003. Good day sunshine : stock returns and the weather. The Journal of Finance, 58(3), 1009–1032.
Hou, W., Kuo, J., & Lee, E. 2012. The impact of state ownership on share price informativeness: the case of the Split Share Structure Reform in China. The British Accounting Review, 44, 248-261. Huberman, G., 2001. Familiarity breeds investment. The Review of Financial Studies, 14(3), 659–
680.
Jiang, Y., Wang, M., 2010. Investor sentiment and stock returns：An empirical study on aggregate effects and cross-section effects. Nankai Business Review, 3, 150–160.
Kamstra, M. J., Kramer, L. A., Levi, M. D., 2000. Losing sleep at the arket: the daylight saving anomaly. The American Economic Review, 90(4), 1005–1011.
Kolari, J. W., Pynnonen, S., 2011. Nonparametric rank tests for event studies. Journal of Empirical Finance, 18, 953–971.
Kolari, J. W., Pynnonen, S., 2010. Event study testing with cross-sectional correlation of abnormal returns. Review of Financial Studies, 23(11), 3996–4025.
Levy, T., Yagil, J., 2011. Air pollution and stock returns in the US. Journal of Economic Psychology, 32, 374–383.
Levy, T., Yagil, J., 2013. Air pollution and stock returns – extensions and international perspective. International Journal of Engineering, Business and Enterprise Applications, 4(1), 1–14.
Li, L., Han, L., 2013. Value investing or value creating? A comparative study between foreign and domestic institutional investors. China Economic Quarterly, 13(1), 351–372.
Li, P., Zeng, Y., 2005. Study of inventory and herding in security markets. Chinese Journal of Management, 5, 606–608.
Li, Q., Peng, C. H., 2016. The stock market effect of air pollution: evidence from China. Applied Economics, 48(36), 3442–3461.
Lipson, M., Puckett, A., 2006. Volatile markets and institutional trading. Unpublished working paper. Darden Graduate School of Business Administration, Charlottesville.
Loewenstein, G., 2000. Emotions in economic theory and economic behavior. The American Economic Review, 90(2), 426–432.
Lucey, B. M., Dowling, M., 2005. The role of feeelings in investor decision-making. Journal of Economic Surveys, 19(2), 211–238.
MacKinlay, A. C., 1997. Event studies in economics and finance event studies in economics and finance. Journal of Economic Literature, 35(1), 13–39.
Oudin, A., Åström, D. O., Asplund, P., Steingrimsson, S., Szabo, Z., Carlsen, H. K., 2018. The association between daily concentrations of air pollution and visits to a psychiatric emergency unit: a case-crossover study. Environmental Health, 17(4), 1–9.
Pagano, M., Röell, A. A., Zechner, J., 2002. The geography of equity listing : why do companies list abroad ? The Journal of Finance, 57(6), 2651–2694.
Podpiera, R., 2001. International cross-listing: the effects of market fragmentation and information flows. Unpublished working paper. Academy of Science of the Czech Republic, Prague. Roosenboom, P., van Dijk, M. A., 2009. The market reaction to cross-listings: Does the destination
market matter? Journal of Banking and Finance, 33(10), 1898–1908.
Rotton, J., Frey, J., 1984. Psychological costs of air pollution : atmospheric conditions, seasonal trends , and psychiatric emergencies. Population and Environment, 7(1), 3–16.
Sarkissian, S., Schill, M. J., 2009. Are there permanent valuation gains to overseas listing? Review of Financial Studies, 22(1), 371–412.
Saunders, E. M., 1993. Stock prices and wall street weather. The American Economic Review, 83(5), 1337–1345.
Schwarz, N., Clore, G. L., 1988. How do I feel about it? The informative function of affective states. In: Fiedler, K., Forgas, J. (Ed.), Affect, cognition and social behavior. Hogrefe, Toronto, pp. 44–62.
Seaton, A., MacNee, W., Donaldson, K., Godden, D., 1995. Particulate air pollution and acute health effects. Lancet, 345(8943), 176–178.
Shu, H., Hung, M., 2009. Effect of wind on stock market returns: evidence from European markets. Applied Financial Economics, 19, 893–904.
Slovic, P., Finucane, M. L., Peters, E., MacGregor, D. G., 2007. The affect heuristic. European Journal of Operational Research, 177, 1333–1352.
Stafford, T. M., 2015. Indoor air quality and academic performance. Journal of Environmental Economics and Management, 70, 34–50.
Tan, S., Chen, Y., 2012. Investor experience and return rates. Journal of Financial Research, 5, 164– 178.
Tomei, F., Rosati, M. V., Ciarrocca, M., Baccolo, T. P., Gaballo, M., Caciari, T., Tomao, E., 2003. Plasma cortisol levels and workers exposed to urban pollutants. Industrial Health, 41, 320–326. Yang, Y., Zhi, X., 2006. The empirical study on herd behavior and stock returns. The Border
Economy and Culture, 5, 74–75.
Zhang, J., Jin, H., 2017. A study on the difference between the impacts of haze on the labor productivity among different skilled employees: an empirical analysis based on CEES Data. Journal of Macro-Quality Research, 5(3), 101–118.
Zhang, L., 2017. Air pollution prevention and economic growth. Economic Vision, 1, 25–30. Zhang, S., 2014. The Market Efficiency of the Chinese A-B-share Market. Unpublished working
paper. Stanford University, Stanford.
Zivin, J. S. G., Neidell, M. J., 2012. The impact of pollution on worker productivity. American Economic Review, 102(7), 3652–3673.
Appendix 1
Event study methodology
Adjusted standardized cross-sectional test
First of all, abnormal returns of individual stocks in the event window are standardized by the standard deviation of excess returns in the estimation window:
𝑆𝐴𝑅34 =BC=>
}~•= (11) Then the average standardized abnormal returns are calculated by dividing the sum of standardized abnormal returns on the event day 𝒕 by the sample size N:
𝐴𝐴𝑅4 =D& D 𝑆𝐴𝑅34
3b& (12) The test statistics is
𝜃& = BBCE> D∗ &I(DG&)H&GH (13) where
𝑆# = &
DG& (𝑆𝐴𝑅34 D
3b& − 𝐴𝐴𝑅4) (14) where 𝑟 is the average cross-correlation of excess returns in the estimation window.
The standardized abnormal returns can be aggregated to cumulative standardized abnormal returns for multiple days in the event window:
𝑆𝐶𝐴𝑅3O = OObO[ @𝑆𝐴𝑅34 (15)
And the average cumulative standardized abnormal returns are then calculated by dividing the cumulative standardized abnormal returns by the sample size N:
𝐶𝐴𝐴𝑅O =D& D 𝑆𝐶𝐴𝑅3O
3b& (16) and the test statistics is
𝜃# =MBBCN D
E ∗
&GH
𝑆# = &
DG& (𝑆𝐶𝐴𝑅3O D
3b& − 𝐶𝐴𝐴𝑅O) (18) where 𝑟 is the average cross-correlation of excess returns in the estimation window.
Generalized rank test
The calculation of standardized abnormal returns and standardized cumulative abnormal returns are the same as in eq. (11) and (15), respectively. They are then re-standardized by the cross-sectional standard deviation to account for the event-induced variance in the event window:
𝑆𝐶𝐴𝑅∗ = EMBC=N
E€•~• N (19)
The calculation of the denominator is the same as in eq. (18). The generalized abnormal returns (GSAR) take 𝑆𝐴𝑅 in the estimation window and take 𝑆𝐶𝐴𝑅∗ for cumulative event day as the cumulative event day is considered to be one time point. The rank of GSARs including all the SARs in the estimation window and the CAR-period. Then the GSARs are ranked as:
𝑈3O =CBDR SEBC=>
TI& − 0.5 (20) where T is the length of the estimation window plus one cumulative event day. The test statistics for zero abnormal returns is:
Appendix 2
Table 6 Summary statistics of events by time
This table presents summary statistics of the heavy air pollution events included in the sample by time. The heavy air pollution events are defined as the days when Air Quality Index (AQI) is larger than 300. The first column is the event time, followed by the number of events. The sample period is from December 2013 to December 2016.
Event time Number of events Event time Number of events
Appendix 3
Table 7 Robustness check of other potential explanatory factors of the excess returns
This table presents the explanatory power of other potential mood-related variables on the excess returns during the days of heavy air pollution. Temperature, humidity and wind speed are weather variables; Monday and January are calendar anomalies, they are dummy variables which take 1 when the days that heavy air pollution occurs are on Monday and in January respectively, and take 0 otherwise. The data of the event day 0 is taken. The standard errors are reported in parentheses. ***, **,* denote statistical significance at 1%, 5% and 10% level respectively.
Variables Y=CAR[-1,1] Coefficient Temperature 0.0003 (0.0003) Humidity -0.0061 (0.0067) Wind speed -0.0018 (0.0011) Monday 0.0085 (0.0062) January -0.0081 (0.0058) Constant 0.0273 (0.0316)
Fixed effects Year
Observations 296
Adjusted R-squared 0.0169
