Does Extreme Weather Impact Performance in Capital Markets? Evidence from China

Abstract

No form of economic activity is unaffected by climate change, which has emerged as a new risk factor impacting financial market stability and sustainable development. This study examines the impact of extreme weather on the stock returns of A-share listed companies in China. Utilizing a decade-long dataset, we construct monthly proportions of extreme high-temperature days and extreme humid days using a percentile comparison approach. The findings reveal a significant negative impact of extreme weather on stock returns. Specifically, each standard deviation increase in the monthly proportion of extreme high-temperature days and extreme humid days corresponds to a decrease in annualized returns by 0.09% and 0.15%, respectively. The mediation analysis suggests that extreme weather primarily affects stock returns through its influence on investor sentiment, impacting economic decision making, with minimal direct effects on corporate performance. Additionally, the sensitivity of stock returns to extreme weather varies notably among different types of companies. Larger, more profitable, and less risky firms show lower sensitivity to extreme weather. The impact is observed not only in heat-sensitive industries but also in non-heat-sensitive industries and remains significant even after excluding company announcement days. This study offers new insights and relevant recommendations for businesses and policymakers on sustainable development and financial stability.

1. Introduction

As global climate warming intensifies, extreme weather events have become more frequent, and daily weather patterns have become increasingly erratic. These phenomena not only disrupt the balance of ecosystems but also profoundly impact human production, lifestyles, and decision-making processes, posing significant threats to global sustainable development. Climate change is now recognized as a major risk to financial market stability. Consequently, authoritative institutions worldwide have acknowledged the importance of integrating climate factors into macroeconomic and microeconomic forecasts and analyses. This integration aims to accurately assess the climate risk exposure of financial assets and inform risk management and investment decisions. This paper examines the impact of climate risk on capital market performance from the perspective of the stock market. As the world’s second-largest economy, China has actively advanced climate change policies in recent years, setting goals for emission peak and carbon neutrality (dual carbon goals) to reduce greenhouse gas emissions and promote green low-carbon development. Given China’s vast territory, diverse climate, uneven regional economic development, varied industries, and uneven distribution of listed companies, it provides a rich dataset for studying the impact of climate risk on its stock market. Furthermore, the large number of small and medium investors in the Chinese stock market means that their investment decisions significantly affect market returns and household wealth. Studying the impact of climate risk on stock returns can help these investors better identify and manage climate risks, thereby enhancing their investment security and stability.

Existing research often focuses on the general trends of climate change, overlooking the dynamics of relative changes. This paper innovatively introduces the proportion of extreme weather days as an indicator, aiming to reveal trends in climate risk through historical quantile comparisons. Although relative weather changes may be imperceptible in the short term, their cumulative effects can lead to frequent extreme weather events and even compound disasters [1,2]. This perspective not only enriches the methodology of climate risk research but also highlights the potential risks posed by climate change, urging businesses and investors to take proactive measures. Although scholars have established a correlation between extreme temperature and humidity and stock market performance [3,4], their studies have not further explored the mechanisms involved nor thoroughly tested the heterogeneity and robustness of their results. Consequently, current research lacks detailed explanations of how climate variables specifically impact stock returns. Furthermore, studies focusing on the impact of extreme temperatures on stock returns tend to concentrate on this single variable [5], potentially neglecting the comprehensive influence of other climate factors such as humidity. On the one hand, humidity and temperature are considered significant factors affecting human psychology [6]. Large-scale data studies indicate that both heat and high humidity are associated with deteriorating emotional expression [7], a phenomenon observed across different countries [8]. Additionally, prolonged exposure to hot and humid environments impairs cognitive abilities [9]. On the other hand, extreme high temperatures and humidity are direct evidence of global warming [10], and such compounded extreme conditions significantly drive heat stress [11], posing threats to physical and mental health and causing substantial economic impacts [12,13]. Therefore, considering both extreme temperature and humidity while examining the impact of climate on capital markets is crucial. This approach holds significant policy and practical implications, encouraging stakeholders to recognize and mitigate potential risks and capitalize on emerging opportunities.

This study establishes a multivariate OLS model with fixed effects for companies, years, and quarters, with stock returns as the dependent variable, and the proportion of extreme hot days and extreme humid days as the core independent variables, to investigate the impact of extreme weather on stock prices of listed companies in China. The results indicate that extreme weather significantly and negatively affects stock returns. Specifically, a one-standard-deviation increase in the proportion of extreme high-temperature days reduces annualized stock returns by 0.09%, and a one-standard-deviation increase in the proportion of extreme humidity days reduces annualized stock returns by 0.15%, surpassing the average annualized returns of 90% of the stocks in the sample.

This paper explores whether investor sentiment from the perspectives of psychology and behavioral finance can partially explain why extreme weather affects stock returns. Extreme weather affects human physiological and psychological states, thereby influencing cognition and emotion. These emotional fluctuations can lead to changes in extrapolative expectations, attribution bias, and risk preferences, ultimately affecting economic decisions and, due to home bias, the stock returns of companies headquartered in affected areas. This paper also employs revenue, net profit, and productivity as dependent variables to verify whether extreme weather affects stock prices through corporate performance. However, it finds no significant impact of the proportion of extreme weather on the performance of listed companies. Even after excluding companies headquartered in Beijing and Shanghai to avoid significant distance between their primary operations and headquarters, the above results hold. In the heterogeneity analysis, this paper investigates which types of companies are most sensitive to extreme weather in terms of total asset size, return on assets, and volatility. Companies with smaller assets, weaker profitability, and higher volatility are more affected by extreme weather in stock returns. Particularly, the negative impact of extreme humidity on low-profitability companies is nearly 45% greater than that on high-profitability companies. Robustness checks are conducted by varying control variables, the construction of dependent variables, and sample selection. Even after including firm characteristics and controlling for regional macroeconomic factors, the negative impact of extreme weather variables on stock returns remains significant. Regardless of whether they belong to thermally sensitive industries, the main results persist; even after removing samples following company announcement dates to prevent event-driven stock price changes, the impact remains. Finally, this paper extends the analysis by exploring the interaction between extreme humid weather and extreme high-temperature weather, offering new perspectives for future research.

The structure of this paper is as follows. Section 2 reviews the relevant literature and raises hypothesis. Section 3 outlines the model specifications. Section 4 presents the primary findings. Section 5 conducts a mechanism analysis. Section 6 performs a heterogeneity analysis. Section 7 conducts a robustness check. Section 8 provides a discussion for further study. Section 9 offers concluding remarks.

2. The Literature Review and Hypothesis

2.1. The Related Literature

The earliest research on the relationship between weather and the stock market used data from 1927 to 1989, examining New York City’s weather and the NYSE index. The study found that lower cloud cover is associated with higher stock returns [14]. Since the beginning of the 21st century, global climate warming has exacerbated environmental issues, prompting international scholars to delve deeper into the impact of climate variables on the economy. Some researchers have focused on the financial stock market to capture climate risk. From the perspective of extreme climate events, the exposure to catastrophic climate events (including droughts, hurricanes, floods, snowstorms, tornadoes, and hail) significantly impacts a company’s systemic risk, equity cost, stock trading volume, and stock returns [15,16,17,18,19]. Another common approach to measuring the impact of climate risk on stock prices involves using official indices. The Climate Policy Uncertainty Index (CPU) is utilized to proxy for climate risk exposure and there is a premium in the level of exposure [20]; the ND-GAIN Climate Index is utilized and found to have an impact on the stock returns of fossil fuel firms [21]. Additionally, some scholars have employed three complementary but distinct climate risk indices—Climate Policy Uncertainty (CPU), Climate Change News (CCN), and Negative Climate Change News (NCCN)—to study the long-term volatility and correlation of daily returns for green and brown energy stocks [22]. A similar approach to this study involves measuring climate risk by calculating daily weather data. By using MA models or first-difference models to distinguish extreme weather from normal conditions, researchers have investigated the impact of extreme temperatures and humidity on stock returns [3,4]. Alternatively, comparing daily maximum temperatures with a benchmark of 30 °C (86 °F) reveals the presence of a high-temperature effect. Specifically, when the daily maximum temperature exceeds this benchmark, trading volume significantly decreases [23]. Unlike previous studies, this paper focuses on the impact of weather changes relative to historical norms on capital market performance. Analyzing daily weather trends is crucial because most natural and compound disasters are driven by weather system characteristics, which increase their severity [1,2,24,25]. For example, rising temperatures can increase the frequency of heatwaves, while humidity, air flow, and other weather factors can exacerbate their effects, leading to heat stress [26]. To generate the measurement indicators, this study compares weather conditions with the 90th percentile of historical distributions to determine the occurrence of extreme weather and calculate its proportion. This paper also includes average weather variables such as temperature, wind speed, and sunlight in the control variables, and accounts for company, year, and quarter fixed effects to minimize the influence of other factors on the results.

In examining why extreme weather impacts various economic variables, international scholars primarily attribute this to investor sentiment. Using Google search volume as a measure of public attention, it has been found that interest in climate change increases during periods of unusually high local temperatures. As a result, investors adjust their views on global warming, leading them to buy stocks that are less sensitive to climate and sell those that are more sensitive to climate changes [27]. As a company’s exposure to high temperatures increases, the discrepancy between analysts’ estimates and actual financial performance, as well as earnings announcement returns, becomes larger [28]. Extreme weather may also impact investors’ overconfidence, leading to changes in both return rates and turnover rates [29]. From a risk-seeking perspective, the hysteria and apathy induced by high temperatures can reduce investors’ willingness to take risks, leading to a significant negative correlation between temperature and returns [30]. Section 5.1 of this paper will further detail how weather impacts stock returns through investor sentiment. Beyond investor sentiment, the previous literature has explored the direct impact of climate risk on corporate performance. Increased exposure to extreme heat reduces company revenue and operating income [28]. Extreme temperatures significantly affect a firm’s market value due to inherent risk-taking levels [31], possibly because of impacts on worker productivity and attendance [32] or the efficiency of production equipment [33]. However, one study using data from U.S. firms has found that neither sales nor worker productivity are affected by extreme temperature shocks. Regardless of whether the industry is temperature-sensitive, a company’s quarterly sales do not significantly change due to extreme temperature shocks [34]. While the aforementioned studies separately examine the effects of weather on investor sentiment or corporate productivity, this paper integrates both aspects, utilizing the extensive theoretical literature and empirical data to analyze the mechanism.

2.2. Research Hypothesis

Greater climate risk impacts stock returns through investor sentiment and corporate performance.

Firstly, the increase in extreme weather events significantly affects investor sentiment. Research shows that extreme weather heightens feelings of frustration and hostility among investors [30,35]. These negative emotions lead investors to downgrade their beliefs about a firm’s profitability and performance and their expectations for its future development [36,37,38]. Additionally, low investor sentiment decreases risk tolerance [39,40], influencing investment decisions.

Home bias causes investors to hold more local stocks [41,42]. This preference increases their focus on local companies, amplifying the impact of sentiment changes on the stock market [43]. Furthermore, investor ambiguity aversion makes them more sensitive to local firms’ stocks [44,45], causing stronger reactions to stocks of firms headquartered locally.

Secondly, from an operational perspective, increased extreme weather significantly impacts corporate performance. Extreme weather affects workers’ psychological and physical health, reducing motivation and productivity and increasing the risk of workplace injuries [32]. Additionally, the efficiency of production equipment decreases due to heat and humidity, affecting product quality [33]. These factors collectively reduce a firm’s revenue and net profit, ultimately impacting the stock returns of firms headquartered locally.

Based on these observations, this paper proposes the following research hypothesis:

H1. 

The bigger the increase in extreme weather, the lower the stock returns of firms headquartered in the affected areas.

3. Research Design

3.1. Data Resources

This study focuses on A-share companies listed on the Shanghai and Shenzhen stock exchanges from 2015 to 2019 (Our study primarily utilizes data spanning the period from 2015 to 2019 for its analysis. The rationale behind this choice lies in the fact that, subsequent to the onset of the COVID-19 pandemic in 2020, both the epidemic and weather changes serve as exogenous variables capable of influencing investor sentiment and exerting effects on the dynamics of supply and demand for commodities, which, in turn, have indirect implications for firm performance. Consequently, both contribute to heightened turbulence in the stock market [46,47,48]. Therefore, this research investigates the influence of weather changes on stock returns through an examination of data predating the year 2019). Data pertaining to stock returns, registered provinces and cities, and company announcement dates were obtained from the CSMAR database (https://data.csmar.com/, accessed on 17 April 2021). Financial indicators and other relevant data were sourced from both the Tonghuashun database (https://data.10jqka.com.cn/, accessed on 30 April 2021) and the Wind database (https://www.wind.com.cn/, accessed on 26 April 2021) (The Tonghuashun database and the Wind database are both comprehensive and widely utilized economic and financial databases, known for their extensive data coverage and broad applicability). Weather data, including temperature, sunshine duration, relative humidity, and wind speed, for the years 2010 to 2019 were sourced from the China Meteorological Administration (https://data.cma.cn/, accessed on 15 May 2021) (Utilizing data from the period between 2010 and 2014, we computed monthly weather variables with seasonal effects removed and distinguished between extreme weather and average weather variables, as elaborated in the following sections).

Upon acquiring the raw data, the following filtering criteria were applied:

• Exclusion of companies in the financial and real estate sectors, specifically by removing companies with industry codes starting with “J” and “K” (The core dependent variable in this study is stock returns. Financial and real estate industries, compared with other sectors, are capital-intensive, highly leveraged, and subject to government regulation. Consequently, their stock prices are influenced by additional specific factors. Therefore, data from these industries were excluded from the analysis. This exclusion method is widely used in numerous studies related to the stock market [49,50]).
• Exclusion of companies with the designation “ST” or “ST*”, ensuring that all sample companies maintained good and normal financial conditions.
• Removal of suspicious, erroneous, and missing weather data, specifically data with weather quality control codes of 1, 2, and 8, and data where all meteorological elements were recorded as 32,766.
• Matching the location of the meteorological station identification number (“QuZhanHao”) with the cities where the listed companies were based and excluding samples where the geographical location did not align with the name of city (This study follows the approach consistent with the previous literature [51,52]).

3.2. Processing of Variables

A plethora of scientific evidence indicates a growing prevalence of extreme weather events. In certain regions, the occurrence of heatwaves has more than doubled, while the frequency and intensity of winter storms continue to escalate. Investigating the impact of extreme weather variables on the stock market is particularly crucial. This study employs the proportion of extreme hot days and extreme humid days to represent these variables. This study focuses on analyzing weather trend changes over time. Given China’s vast latitudinal span, diverse climate zones, and significant geographical differences, it adopts a relative standard for defining extreme weather, rather than absolute thresholds like 30 °C or 0 °C for extreme temperatures [5,34]. Herein, extreme hot days per month are defined as those surpassing the 90th percentile of the daily temperature distribution for each corresponding month from 2010 to 2014. Similarly, extreme humid days per month are defined as those exceeding the 90th percentile of the daily relative humidity distribution for each corresponding month from 2010 to 2014.

Due to the correlation between extreme weather variables and average weather variables, this study selects four common weather factors, namely temperature, sunshine hours, relative humidity, and wind speed, as control variables. While weather variables exhibit strong seasonal effects, stock market returns also demonstrate seasonal characteristics, with the month being one of the influencing factors on portfolio expected returns [53]. Therefore, to mitigate spurious correlations, this study adopts a method inspired by past scholars to seasonally adjust the weather variables [27,54,55]. Firstly, the daily weather variables for each station are averaged to obtain monthly weather variables. Subsequently, the monthly weather variables for each station are differenced from the average daily weather variables for each month over a five-year period from 2010 to 2014 to derive seasonally adjusted average monthly weather variables.

After data cleaning and variable processing, this study conducts the main regression analysis based on a sample of 100,864 observations.

3.3. Model Setup

To investigate the impact of extreme weather on the stock prices of listed companies in China, this study employs stock returns as the dependent variable. The core independent variables include the proportion of extreme hot days and extreme humid days, while controlling for monthly average weather variables adjusted for seasonality such as temperature, sunshine duration, relative humidity, and wind speed. Additionally, to mitigate the potential influence of past stock returns on the results, the model incorporates lagged one-period stock returns (The purpose of including lagged one-period stock returns as a dependent variable in the study is to take into account the serial dependence [3,5,56]). The analysis utilizes a multivariate Ordinary Least Squares (OLS) regression model with fixed effects for companies, years, and quarters.

$$R_{it}=\alpha +{\theta }_1{pctT}_{i,t}+{\theta }_2{pctH}_{i,t}+\beta {Controls}_{i,t}+{\delta }_{i,y,q}+{\epsilon }_{i,t}$$

(1)

The variable names and definitions are listed in

Table 1. Variable definitions and names.

Variable Definitions	Variable Names
$R_{i,t}$	Monthly stock return (%)
pctTi,t	Proportion of extreme hot days (proportion of extreme hot days is the number of extreme hot days over number of days per month)
${pctH}_{i,t}$	Proportion of extreme humid days (proportion of extreme humid days is the number of extreme humid days over number of days per month)
DTi,t	De-seasonalized average monthly temperature (°C)
${DS}_{i,t}$	De-seasonalized average monthly sunshine hours (h)
${DH}_{i,t}$	De-seasonalized average monthly relative humidity (%)
${DW}_{i,t}$	De-seasonalized average monthly wind speed (m/s)

.

3.4. Descriptive Statistics

Table 2. Descriptive statistics.

Variables	Mean	Sd	Min	Max	P10	P90
Proportion of extreme hot days	0.140	0.115	0	0.839	0	0.300
Proportion of extreme humid days	0.143	0.130	0	0.968	0	0.323
Monthly temperature (°C)	17.335	9.633	−20.087	32.142	4.117	28.303
Monthly sunshine hours (h)	5.339	1.916	0.132	12.367	2.800	7.910
Monthly relative humidity (%)	69.906	13.283	18.645	97.033	49.226	83.533
Monthly wind speed (m/s)	2.214	0.503	0.506	6.458	1.650	2.806

Notes: Descriptive statistics are calculated based on 100,864 observations. Sd represents the standard deviation. P10 and P90 denote the 10th quartile and 90th quartile, separately. Source: China Meteorological Administration; CSMAR database; Tonghuashun database; Wind database.

presents the descriptive statistics for each weather variable. Regarding extreme weather, both the maximum values of the proportion of extreme hot days and extreme humid days approach 1, indicating that, in certain months, the temperature or relative humidity remains consistently above the 90th percentile value of the respective month from 2010 to 2014, with the 90th percentile approximately at 0.3. The notable disparity between this percentile and the maximum values suggests a moderate increase rather than a sharp rise in extreme hot and humid days from 2015 to 2019 compared with the reference period. In terms of monthly average weather, the mean temperature is 17.34 °C and the mean monthly humidity is 69.91%, reflecting to some extent the concentration of listed companies in the sample in the southern regions of China, particularly in the East and South China regions, which collectively constitute over 50% of the sample and belong to the monsoon climate zone. The mean monthly sunshine duration is 5.34 h, with the longest sunshine duration recorded in June 2019 in Chifeng City, Inner Mongolia, reaching 12.37 h.

4. Impact of Extreme Weather on Stock Returns of Listed Companies

The results of this study demonstrate that extreme weather variables have significant negative impacts on stock returns, confirming the hypothesis. The first column of

Table 3. Effect of extreme weather variables on stock returns.

Variables	Monthly Stock Return
	(1)	(2)	(3)	(4)
Proportion of extreme hot days	−0.0986 ***	−0.0996 ***	−0.0934 ***	−0.0688 ***
	(0.005)	(0.006)	(0.006)	(0.006)
Proportion of extreme humid days	−0.1136 ***	−0.1174 ***	−0.0974 ***	−0.0960 ***
	(0.006)	(0.006)	(0.006)	(0.006)
Constant	0.0210 ***	0.0201 ***	0.0169 ***	0.0158 ***
	(0.001)	(0.001)	(0.001)	(0.001)
Controls	YES	YES	YES	YES
Firm FE		YES	YES	YES
Year FE			YES	YES
Quarter FE				YES
Observations	100,864	100,864	100,864	100,864
R-squared	0.022	0.033	0.076	0.088

Notes: * p < 0.10; ** p < 0.05; *** p < 0.01. Controls means that we control for lagged one-period stock return and monthly de-seasonalized average weather variables including temperature, sunshine duration, relative humidity, and wind speed. Standard errors clustering at firm level are reported in parentheses. Source: China Meteorological Administration; CSMAR database; Tonghuashun database; Wind database.

shows the regression results with control variables included in Model 1, while the second and third columns progressively add company, year, and quarter fixed effects. Overall, the coefficients of the variables do not vary significantly across the three columns. Using the final column as the baseline, the coefficients for the proportion of extreme high-temperature days and extreme humidity days are −0.069 and −0.096, respectively. This indicates that a one-standard-deviation increase in the proportion of extreme high-temperature days reduces the annualized stock return by 0.09%, whereas a similar increase in the proportion of extreme humidity days decreases the annualized stock return by 0.15%. The impact of extreme humidity is 1.7 times greater than that of extreme high temperatures, exceeding the average annualized stock return for 90% of the stocks in the sample. The significant difference between the two coefficients suggests that the impact of relative humidity on financial markets should receive more attention, although few studies currently address this factor [3,4,56]. Compared with the de-seasonalized average weather variables, the impact of extreme weather on stock returns is considerably larger. For instance, the coefficients for de-seasonalized monthly average temperature and monthly average sunlight are only −0.013 and −0.04, respectively. This indicates that the increase in extreme weather proportions has a stronger negative effect on stock returns than routine weather changes.

These results align with previous research findings, which also demonstrate a significant negative impact of climate risk on stock returns [5,20,21,57]. This study extends the existing literature, particularly regarding the construction of indicators. For instance, previous studies have used a Climate Policy Uncertainty index (CPU) based on U.S. newspapers to examine its impact on the Chinese stock market [20] and have utilized the ND-GAIN index, a comprehensive climate index with an annual frequency, to capture climate risk [21]. This study defines extreme weather by the differences in temperature and humidity between the sample period and the reference period at meteorological stations in Chinese cities, capturing the impact of daily weather trends on capital market performance. Due to the strong exogeneity of extreme temperature and humidity proportions, stock returns do not react to weather, eliminating reverse causality. Additionally, controlling for lagged stock returns, company fixed effects, and time fixed effects minimizes the influence of extraneous variables correlated with both extreme weather and stock returns. Thus, the size and significance of the combined variables’ coefficients confirm that extreme weather significantly negatively impacts stock returns of listed companies. The underlying mechanisms will be discussed in detail in the following section.

5. Mechanism Analysis

As established earlier, extreme weather significantly affects the stock returns of listed companies. In this section, the paper delves into the mechanisms underlying the impact of extreme weather on stock returns. It examines the impact from the perspectives of both investors and firms, categorizing them into investor sentiment and corporate performance and substantiates their existence through theoretical and empirical methods.

5.1. Investor Sentiment

Most of the psychological literature elucidates the idea that human emotions and feelings influence behavior, and that emotions themselves are influenced by environmental factors such as weather [30,58,59]. Research indicates that good weather fosters positive emotions through symbolic associations [60], while bad weather enhances depressive and hostile feelings [30,35,61]. Among these weather variables, sunlight, temperature, and humidity have the most significant psychological impact on humans [6]. To quantify this impact, previous research has combined social media data with spatial geographic data. For instance, using Chinese social media data, it was found that for each standard deviation increase in extreme high temperatures, the emotional index decreases by approximately 0.161 units [62]. Combining remote sensing imagery with statistical yearbook data to depict the spatial and temporal patterns of temperature’s effect on residents’ irritability, it was observed that, even after controlling for temperature, humidity also significantly affects individuals’ psychological and emotional states [63]. Furthermore, changes in humidity alter the predictive power of temperature on emotions across all temperature levels [64]. According to the largest survey to date, analyzing over 3.5 billion social media posts, both heat and high humidity are associated with deteriorating emotional expression [7]. In addition to affecting personal emotions, prolonged exposure to hot and humid environments also deteriorates cognitive abilities [9].

Investor sentiment plays a crucial role in economic decision making. Emotional fluctuations affect both the reception of information [65,66] and the processing of new information [67,68]. These fluctuations also influence personal interests and preferences [69,70], ultimately shaping behavior [71]. In the stock market, how does extreme weather impact stock returns through investor sentiment [3,5,72]. First, emotions may cause investors to mistakenly attribute “good” or “bad” moods to a company’s performance or other factors, often without realizing that their decisions are being influenced by their emotions. This can alter investors’ expectations of listed companies and their sensitivity to new information [37,38]. For instance, analysts experiencing bad weather respond more negatively and slowly to earnings announcements compared with those experiencing good weather [73]. Second, changes in investor sentiment can alter risk preferences, affecting risk-taking behavior [74]. For example, people in a good mood are more willing to tolerate risk and may hold higher risk stocks in their portfolios [39]. Emotional states also influence the credit approval rates of lower level financial managers and are linked to seasonal affective disorder in financial market risk aversion behavior [40,75]. Additionally, research indicates that changes in the natural environment can shift attention, leading to the disposition effect [63].

Many investors exhibit an overwhelming home bias when selecting financial products, showing a preference for domestic stocks, especially those headquartered nearby. Home bias is widespread globally [76]. For instance, U.S. investors allocate 98% of their equity portfolios to domestic stocks, despite the U.S. market comprising only 36.4% of global stock market capitalization [41]. Similarly, in the UK, these figures are 78.5% and 10.3%, respectively, and in Japan, 86.7% and 43.7%. Similar phenomena have been observed among fund managers. For instance, in the investment portfolios of American fund managers, at least one out of every ten companies is located in the same city as the fund manager [42]. Additionally, fund managers hold 12% more stocks from their own state compared with their peers [77]. When local preferences arise from an informational advantage, they can yield positive returns for investors. The annualized return from local holdings outperformed non-local holdings by 3.2%, likely due to local investors leveraging local knowledge [78]. Analysts geographically closer to company headquarters provide more accurate forecasts [79,80]. However, if local bias stems from limited or irrational behavior, it can negatively impact stock returns. For example, less capable investors prefer local stocks [81]. Due to limited attention, investors respond more quickly and sensitively to information released by local firms [43]. Overconfident investors may overestimate their ability to predict the performance of familiar assets [82]. Pure aversion to ambiguity also plays a role [44,45]. Domestic investors perceive foreign stocks as riskier simply because they are foreign [83]. Moreover, even if domestic investors have access to the same information as foreign investors, they may choose to remain uninformed [84].

Based on the literature regarding investor sentiment and home bias, it can be inferred that extreme weather conditions may lead to changes in investors’ beliefs about companies and their own risk preferences. Given the prevalent home bias, where investors tend to hold more local stocks, this results in decreased stock returns for companies headquartered locally. Previous studies have also explored the impact of environmental conditions on financial markets. For example, studies utilizing data from Beijing-listed companies and local weather conditions found that smog pollution significantly negatively affects stock returns through investor sentiment [85]. When the analysis was extended from Beijing to a national scale, it was discovered that nearly all air pollution proxy variables had a significant negative impact on the firm-level returns of companies headquartered in the Shanghai Stock Exchange Index [51]. Additionally, researchers measuring investor sentiment using media indices assessed the average air pollution levels in 364 Chinese cities, weighting them by the corresponding city’s Baidu Search Index [52].

Drawing from these studies, we think that a higher proportion of days of extreme heat and humidity intensifies investors’ negative sentiments, leading to downward revisions in company beliefs and reduced risk tolerance. Due to the presence of home bias, this results in a significant negative impact on the stock returns of companies headquartered locally.

5.2. Corporate Performance

In addition to investor sentiment, weather changes can potentially influence stock returns through their impact on firm performance. By adopting the approach of the previous literature, this study assesses firm profitability both directly and indirectly, employing gross revenue, net profit, and productivity as key metrics [34]. Productivity—an indicator denoted as the ratio of a firm’s gross operating income to its number of employees—is utilized in an effort to investigate the potential link between weather changes and firm profitability by examining their influence on labor motivation. Given that Chinese listed companies disclose financial reports quarterly, the model differs from the main regression in that quarterly averages of each variable are included. The equations are as follows:

$${GR}_{i,q}=\alpha +{\theta }_1{pctT}_{i,q}+{\theta }_2{pctH}_{i,q}+\beta {Controls}_{i,q}+{\delta }_{i,y,q}+{\epsilon }_{i,q}$$

(2)

$${NP}_{i,q}=\alpha +{\theta }_1{pctT}_{i,q}+{\theta }_2{pctH}_{i,q}+\beta {Controls}_{i,q}+{\delta }_{i,y,q}+{\epsilon }_{i,q}$$

(3)

$${PR}_{i,q}=\alpha +{\theta }_1{pctT}_{i,q}+{\theta }_2{pctH}_{i,q}+\beta {Controls}_{i,q}+{\delta }_{i,y,q}+{\epsilon }_{i,q}$$

(4)

where ${GR}_{i,q}$ denotes the logarithm of a firm’s gross revenue in a given quarter, ${NP}_{i,q}$ denotes the logarithm of a firm’s net profit in a given quarter, ${PR}_{i,q}$ denotes the logarithm of a firm’s productivity in a given quarter. The control variables include seasonally adjusted average weather variables similar to Model 1 and lagged one-period dependent variables. (Through the incorporation of preceding period gross revenue/net profitability/productivity as a control variable, we enhance the ability to isolate the impact of weather variables on current period firm profitability, independent of historical profitability levels. This method aids in discerning whether the identified effects are genuinely associated with changes in weather and not attributable to other variables or historical profitability levels.) ${GR}_{i,q-1}$, ${NP}_{i,q-1}$ and ${PR}_{i,q-1}$ denotes the one-period lagged terms of gross revenue, net profit, and productivity, respectively. But, given that publicly traded companies exclusively disclose financial indicators on a quarterly basis, this study integrates quarterly weather variables into the analysis. To align with the reporting frequency, quarterly averages of the weather variables are computed for inclusion in the model.

Table 4. Effect of weather variables on net profit and productivity.

Variables	Corporate Performance
	Log of Firm’s Gross Revenue	Log of Firm’s Net Profit	Log of Firm’s Productivity
Proportion of extreme hot days	0.0332	−0.2512 *	−0.0183
	(0.076)	(0.147)	(0.082)
Proportion of extreme humid days	0.0161	−0.1036	−0.0251
	(0.058)	(0.122)	(0.061)
Constant	1.7810 ***	2.8609 ***	−0.5857 ***
	(0.157)	(0.061)	(0.053)
Controls	YES	YES	YES
Firm, year, quart FE	YES	YES	YES
Observations	32,234	24,781	30,726
R-squared	0.995	0.809	0.943

Notes: * p < 0.10; ** p < 0.05; *** p < 0.01. Controls means that we control for lagged one-period dependent variable and monthly de-seasonalized average weather variables including temperature, sunshine duration, relative humidity, and wind speed. Standard errors clustering at firm level are reported in parentheses. Source: China Meteorological Administration; CSMAR database; Tonghuashun database; Wind database.

presents the results of regressing company profitability onto extreme weather variables while controlling for fixed effects and other variables. Columns 1, 2, and 3 respectively employ the logarithm of gross revenue, net profit, and productivity as dependent variables. Extreme weather variables do not exhibit significant effects on the performance of listed companies, except for the coefficient of extreme hot days’ proportion being significant at the 10% level in the regression of the logarithm of net profit. Even with such a large sample size and utilizing three measures to gauge company performance, robust conclusions regarding the significant impact of extreme weather on listed company performance cannot be drawn. This finding is consistent with the results of previous studies [5,34].

To access the latest information and attract top talent, most companies establish their headquarters in Beijing or Shanghai. However, their primary business operations may not be located in these cities. To prevent such biases from affecting the results, this study excludes companies headquartered in Beijing and Shanghai and separately regresses stock returns and corporate performance. The results in

Table 5. Regressions excluding firms headquartered in Beijing or Shanghai.

Variables	Monthly Stock Return	Corporate Performance
		Log of Firm’s Gross Revenue	Log of Firm’s Net Profit	Log of Firm’s Productivity
Proportion of extreme hot days	−0.0902 ***	−0.1130	−0.1159	0.0167
	(0.006)	(0.100)	(0.157)	(0.091)
Proportion of extreme humid days	−0.0740 ***	0.0772	0.0367	0.0025
	(0.006)	(0.100)	(0.133)	(0.064)
Constant	0.0164 ***	5.7524 ***	2.7676 ***	−0.5557 ***
	(0.001)	(0.052)	(0.074)	(0.065)
Controls	YES	YESP	YES	YES
Firm, year, quarter FE	YES	YES	YES	YES
Observations	74,180	19,349	18,130	22,531
R-squared	0.092	0.922	0.801	0.943

Notes: * p < 0.10; ** p < 0.05; *** p < 0.01. Controls means that we control for lagged one-period dependent variable and monthly de-seasonalized average weather variables including temperature, sunshine duration, relative humidity, and wind speed. Standard errors clustering at firm level are reported in parentheses. Source: China Meteorological Administration; CSMAR database; Tonghuashun database; Wind database.

indicate that weather changes still significantly affect the stock returns of companies headquartered outside Beijing and Shanghai. The coefficient magnitudes and directions align with the main regression results in Table 3. However, the coefficient for the proportion of extreme high-temperature days is slightly larger than that for extreme humidity days. This discrepancy may be attributed to two factors. First, as China’s financial centers, Beijing and Shanghai host more sophisticated investors and a higher proportion of institutional investors, who are more likely to recognize and manage the emotional fluctuations caused by extreme heat. In contrast, investors outside these cities are more susceptible to the impact of extreme heat, leading to more pronounced market reactions. Additionally, the average proportion of extreme high-temperature days is slightly higher outside Beijing and Shanghai. Second, areas outside Beijing and Shanghai have lower population densities, resulting in weaker depressive effects from extreme humidity. From a corporate performance perspective, even after excluding Beijing and Shanghai, weather changes do not significantly affect companies’ operating revenue, net profit, or productivity, consistent with the results in Table 4.

6. Heterogeneity Analysis

Researchers developed a novel investor sentiment index (BW index) and found that fluctuations in investor sentiment significantly impact companies with subjective valuation and limited arbitrage opportunities. This phenomenon is observed not only in the United States but also in non-U.S. markets such as Canada, Japan, and the United Kingdom [86,87,88]. Therefore, in this section, we further investigate the impact of weather changes on stock returns for various types of publicly traded companies in terms of asset size, profitability, and risk level. To differentiate between companies with large and small asset sizes, we calculate the average total assets of sample companies for each quarter from 2015 to 2019 and define companies below the 30th percentile as “small asset size companies” and those above the 70th percentile as “large asset size companies”. Similarly, we use Return On Assets (ROA) to represent a company’s profitability. Employing a similar approach, we categorize companies as “strong profitability companies” and “weak profitability companies”. Volatility serves as a measure of risk, and we determine “low volatility companies” as those falling below the 30th percentile in average monthly volatility for each sample company from 2015 to 2019, and “high volatility companies” as those exceeding the 70th percentile.

Columns 1 and 2 of

Table 6. Regressions on stock returns for firms sorted by asset size, profitability, and volatility.

Variables	Monthly Stock Return
	Asset Size		Profitability		Risk Level
	Small Asset	Large Asset	Low Profitability	High Profitability	High Risk	Low Risk
Proportion of extreme hot days	−0.0786 ***	−0.0245 **	−0.0560 ***	−0.0519 ***	−0.0792 ***	−0.0318 ***
	(0.013)	(0.010)	(0.013)	(0.010)	(0.011)	(0.009)
Proportion of extreme humid days	−0.1067 ***	−0.0930 ***	−0.1046 ***	−0.0723 ***	−0.1196 ***	−0.0531 ***
	(0.012)	(0.011)	(0.013)	(0.011)	(0.011)	(0.008)
Constant	0.0126 ***	0.0138 ***	0.0152 ***	0.0129 ***	0.0208 ***	0.0051 ***
	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)	(0.001)
Controls	YES	YES	YES	YES	YES	YES
Firm, year, quarter FE	YES	YES	YES	YES	YES	YES
Observations	25,018	33,260	27,674	26,448	44,615	25,066
R-squared	0.0958	0.0548	0.0649	0.0813	0.0861	0.0702

Notes: * p < 0.10; ** p < 0.05; *** p < 0.01. Controls means that we control for lagged one-period stock return and monthly de-seasonalized average weather variables including temperature, sunshine duration, relative humidity, and wind speed. Standard errors clustering at firm level are reported in parentheses. Source: China Meteorological Administration; CSMAR database; Tonghuashun database; Wind database.

, respectively, present the regression results for small and large asset size companies. The coefficients of each weather variable are significant at the 1% level and consistent with those in Table 2. Notably, across both columns, the impact of extreme wet days on stock returns is significantly greater than that of extreme hot days. It is evident that the impact of extreme weather variables on stock returns is significantly amplified for small asset size companies relative to large asset size companies. Particularly noteworthy is that the coefficient of extreme hot days proportion for small asset companies is three times that of large asset companies. Specifically, when controlling for other factors, a 10% increase in the proportion of extreme hot days leads to an additional change of 0.06% in annualized returns for small asset companies, which is close to the monthly stock return value at the 75th percentile within the sample. Compared with the regression results based on firm size categories, the impact of extreme hot days on stock returns of publicly listed companies across different profit levels shows minimal variation. The coefficients of the core independent variables do not exhibit significant differences between the two columns. In contrast, the negative impact of extreme wet days on companies with low profitability is nearly 45% higher than on those with high profitability. The impact of weather changes on stock returns differs notably for companies with different levels of volatility. Both the proportion of extreme hot days and extreme humid days exhibit significantly lower sensitivity to weather for companies with lower volatility compared with those with higher volatility, with the coefficients differing by at least a factor of two. This implies that, for companies with higher volatility, a 20% increase in the proportion of extreme hot days (or extreme humid days) results in a decrease of 0.11% (0.16%) in annualized stock returns compared with companies with lower volatility, a magnitude that exceeds (or matches) the average annualized returns within the entire sample.

In

Table 7. Regressions on stock returns for firms sorted considering three features together.

Variables	Monthly Stock Return
	Weak Companies	Strong Companies
Proportion of extreme hot days	−0.0738 **	−0.0371 *
	(0.036)	(0.019)
Proportion of extreme humid days	−0.1566 ***	−0.0420 **
	(0.032)	(0.021)
Constant	0.0209 ***	0.0155 ***
	(0.006)	(0.004)
Controls	YES	YES
Firm, year, quarter FE	YES	YES
Observations	3538	3817
R-squared	0.114	0.091

Notes: * p < 0.10; ** p < 0.05; *** p < 0.01. Controls means that we control for lagged one-period stock return and monthly de-seasonalized average weather variables including temperature, sunshine duration, relative humidity, and wind speed. Standard errors clustering at firm level are reported in parentheses. Source: China Meteorological Administration; CSMAR database; Tonghuashun database; Wind database.

, this study concurrently considers three categories of characteristics: asset size, profitability, and risk level. Panel A samples are defined as companies that satisfy both small asset size, weak profitability, and high volatility (referred to as “weak companies”). Panel B samples, on the other hand, encompass companies with large asset size, strong profitability, and low volatility (referred to as “strong companies”). These results, akin to previous findings, demonstrate significant variations in the impact of extreme weather on stock returns from heterogeneous companies [86]. Large asset size, strong profitability, and low risk companies exhibit lower sensitivity to extreme weather. From the perspective of investor sentiment, investors affected by weather changes are more likely to deviate from rational beliefs [87] and their risk preferences also shift [89]. Stronger large asset firms with relatively less investment risk and relatively better disclosure, beliefs, and preferences are more likely to act on firms with small asset sizes, weak profitability, and higher volatility [90,91], thus affecting the stock returns of listed firms.

7. Robustness Check

This paper conducts robustness tests primarily on several fronts. Firstly, this study mitigates the potential influence of other asset pricing factors on the main results by incorporating additional enterprise-specific control variables into the model [92,93,94]. Specifically, quarterly total assets, book-to-market ratio, and Return On Equity (ROE) are included to measure firm size, relative value, and profitability, respectively. As shown in the first column of

Table 8. Regressions on stock returns for robustness check (method 1 and method 2).

Variables	Monthly Stock Return			Adjusted Stock Return
	(1)	(2)	(3)	(4)
Proportion of extreme hot days	−0.0694 ***	−0.0702 ***	−0.0751 ***	−0.0064
	(0.006)	(0.006)	(0.006)	(0.005)
Proportion of extreme humid days	−0.0941 ***	−0.0946 ***	−0.0969 ***	−0.0154 ***
	(0.005)	(0.005)	(0.005)	(0.005)
Constant	0.0147 ***	0.0207 ***	0.1587 ***	0.0045 ***
	(0.001)	(0.002)	(0.004)	(0.001)
Controls for average weather	YES	YES	YES	YES
Firm FE	YES	YES	YES	YES
Year, quarter FE	YES	YES	YES
Controls for firm features	YES		YES
Controls for province		YES	YES
Year*Month FE				YES
Observations	98,236	98,236	98,236	98,236
R-squared	0.093	0.093	0.118	0.217

Notes: * p < 0.10; ** p < 0.05; *** p < 0.01. Controls for average weather means that we control for monthly de-seasonalized average weather variables including temperature, sunshine duration, relative humidity, and wind speed. Controls for firm features means that we control for quarterly total assets, book-to-market ratio, and Return On Equity (ROE). Controls for province means that we control for province GDP and province fix effect. Standard errors clustering at firm level are reported in parentheses. Source: China Meteorological Administration; CSMAR database; Tonghuashun database; Wind database.

, these additions yield results consistent with the primary findings. Furthermore, to mitigate the impact of external macroeconomic factors, province fixed effects and province quarterly GDP are added to Model 1, as indicated in the second column. The robustness of the results persists in the third column of Table 8, which considers both types of factors. Secondly, the study uses market-adjusted returns as the dependent variable, calculated as the difference between stock returns and market returns, and incorporates year*month fixed effects to control for risk factors. The results are presented in the fourth column of Table 8. Although the coefficients of the core independent variables are somewhat attenuated in this regression, the coefficient for extreme wet days remains highly significant and notably larger than that for extreme hot days. Columns 1 to 3 of

Table A1. Extended regressions for robustness check method 2.

Variables	Market-Adjusted Stock Return
	(1)	(2)	(3)
	FF3	FF4	FF5
Proportion of extreme hot days	−0.0000	0.0001	−0.0000
	(0.005)	(0.005)	(0.005)
Proportion of extreme humid days	−0.0125 **	−0.0123 **	−0.0121 **
	(0.005)	(0.005)	(0.005)
Constant	0.0006	0.0014 *	0.0015 *
	(0.001)	(0.001)	(0.001)
Controls	YES	YES	YES
Firm FE	YES	YES	YES
Year * Month FE	YES	YES	YES
Observations	98,236	98,236	98,236
R-squared	0.026	0.026	0.027

Notes: * p < 0.10; ** p < 0.05; *** p < 0.01. Controls means that we control for lagged one-period stock return and monthly de-seasonalized average weather variables including temperature, sunshine duration, relative humidity, and wind speed. Standard errors clustering at firm level are reported in parentheses. Source: China Meteorological Administration; CSMAR database; Tonghuashun database; Wind database.

estimate the weights of influencing factors under the Fama–French three factor/four-factor/five-factor models based on data from 2010 to 2014, and calculate the expected returns during the main regression analysis period (2015–2019). The dependent variable is computed by subtracting original stock returns from expected returns. Similar to the first column, the coefficient for extreme humid weather percentage remains robustly and significantly negative.

Then, it examines whether this effect is exclusive to industries sensitive to weather changes. Therefore, in the first two columns of

Table 9. Regressions on stock returns for robustness check (method 3 and method 4).

Variables	Monthly Stock Return
	Weather-Sensitive Industry	Weather-Insensitive Industry	Delete Announcement Dates
Proportion of extreme hot days	−0.0719 ***	−0.0593 ***	−0.0920 ***
	(0.007)	(0.017)	(0.006)
Proportion of extreme humid days	−0.0884 ***	−0.1395 ***	−0.0671 ***
	(0.006)	(0.014)	(0.006)
Constant	0.0149 ***	0.0204 ***	0.0153 ***
	(0.001)	(0.002)	(0.001)
Controls	YES	YES	YES
Firm, year, quarter FE	YES	YES	YES
Observations	82,401	18,463	100,864
R-squared	0.084	0.116	0.088

Notes: * p < 0.10; ** p < 0.05; *** p < 0.01. Controls means that we control for lagged one-period stock return and monthly de-seasonalized average weather variables including temperature, sunshine duration, relative humidity, and wind speed. Standard errors clustering at firm level are reported in parentheses. Source: China Meteorological Administration; CSMAR database; Tonghuashun database; Wind database.

, all industries are divided into weather-sensitive and weather-insensitive categories for analysis (According to the China Securities Regulatory Commission’s industry types in 2012, this study categorizes agriculture, forestry, animal husbandry, fisheries, mining, manufacturing, electric power production and supply, construction, wholesale and retail trade, transportation, accommodation, and food services as weather-sensitive industries). The results indicate that both types of industries are significantly affected by weather risk, indirectly validating that extreme weather does not influence stock returns through corporate performance (If extreme weather affects stock returns through corporate performance, we would expect weather variables to have significant effects only on stock returns of weather-sensitive industries or larger absolute coefficient values. However, as indicated in the table above, even the coefficient for the proportion of extremely wet weather is larger in absolute terms in the sample of non-heat-sensitive industries). Secondly, to prevent the influence of stock price due to event-driven factors, rather than weather fluctuations, the analysis excludes samples from dividend announcement days (Samples with the following types of corporate announcements made between 2015 and 2019 have been excluded from the analysis: IPO, periodic reports, rights issues and capital increases, convertible bond financing, other forms of financing, equity changes or mergers and acquisitions, business transactions, shareholder meetings, equity pledge freezes, litigation and arbitration, changes in board and executive leadership, unexpected events, regulatory violations, and donations. After excluding samples on announcement days, we followed the methodology outlined in Section 3.2 to preprocess the variables and obtain monthly data, which were then incorporated into the regression model). It is found that the impact of extreme weather on the stock returns of publicly listed companies remains robust and significant, as shown in the third column of Table 9.

8. Discussion

The results from the previous chapters indicate that both the proportion of extreme heat days and extreme humid days have a significant negative impact on stock returns. By comparing the coefficients in Table 1, it can be observed that a one standard deviation increase in the proportion of extreme humid days leads to a 0.15% decrease in annualized stock returns, which is 1.7 times the impact of extreme high-temperature days. Regardless of a company’s asset type, profitability, and risk level, the impact of extreme humid weather on monthly stock returns is greater than that of extreme high-temperature weather. Specifically, for weaker companies, the effect of a one-unit change in the proportion of extreme humid days on stock returns is approximately twice that of a one-unit change in extreme heat days. Robustness checks using various methods consistently demonstrate the significant negative impact of the proportion of extreme humid days on stock returns. This evidence highlights the importance of considering extreme humid weather [95].

Currently, much of the research on climate risks or extreme weather impacts on financial markets focuses primarily on temperature [5,23,96], with relatively little attention given to humidity. Although scholars have established a link between humidity and physical health, this phenomenon has not garnered sufficient attention from mental health researchers [97]. In fact, humidity affects cognitive abilities [98,99], consumer spending [100], life satisfaction [101], and extreme behaviors [102]. Emotions and psychological states may mediate these effects [100,102,103,104]. With global warming, high temperatures often coincide with high humidity [10,11], and humidity frequently influences the relationship between temperature and emotional and psychological conditions [63,64].

Building on the main findings of this study and the previous literature, we further explore the combined effects of extreme high-temperature days and extreme humid days on stock returns. To this end, we incorporate an interaction term between the proportion of extreme high-temperature days and extreme humid days into Model 1. As shown in

Table 10. Effect of interactive extreme weather variables on stock returns.

Variables	Monthly Stock Return
	(1)	(2)	(3)	(4)
Proportion of extreme hot days	−0.0615 ***	−0.0661 ***	−0.0775 ***	−0.0431 ***
	(0.007)	(0.007)	(0.008)	(0.008)
Proportion of extreme humid days	−0.0714 ***	−0.0785 ***	−0.0792 ***	−0.0661 ***
	(0.007)	(0.008)	(0.008)	(0.007)
Proportion of extreme hot * humid days	−0.3227 ***	−0.2946 ***	−0.1400 ***	−0.2275 ***
	(0.030)	(0.030)	(0.030)	(0.030)
Constant	0.0151 ***	0.0148 ***	0.0144 ***	0.0118 ***
	(0.001)	(0.001)	(0.001)	(0.001)
Controls	YES	YES	YES	YES
Firm FE		YES	YES	YES
Year FE			YES	YES
Quarter FE				YES
Observations	100,864	100,864	100,864	100,864
R-squared	0.022	0.033	0.076	0.088

Notes: * p < 0.10; ** p < 0.05; *** p < 0.01. Controls means that we control for lagged one-period stock return and monthly de-seasonalized average weather variables including temperature, sunshine duration, relative humidity, and wind speed. Standard errors clustering at firm level are reported in parentheses. Source: China Meteorological Administration; CSMAR database; Tonghuashun database; Wind database.

, the regression results from columns 1 to 4 indicate that the direction and magnitude of the core independent variable coefficients remain stable. Specifically, column 4 reveals that, after controlling for firm, year, and quarter fixed effects, both the proportion of extreme high-temperature days and extreme humid days have individual negative impacts on stock returns, with the latter having a slightly larger absolute value. The interaction term coefficient is significantly negative, indicating that a one standard deviation increase in the proportion of extreme humid days amplifies the negative impact of the proportion of extreme high-temperature days on annualized stock returns by 0.35%. This combined effect is greater than the individual impacts of either variable alone. Consistent with previous results, a higher proportion of extreme humid days correlates with lower stock returns. Furthermore, extreme humid weather modulates the effect of extreme high temperatures on stock returns, likely due to increased emotional volatility under high-temperature and high-humidity conditions.

This analysis not only underscores the need for governments and businesses to consider often-overlooked environmental variables (such as humidity) and their interactions when assessing the impacts on financial markets but also highlights the importance of developing comprehensive and effective policies to address the multifaceted challenges posed by climate change. Additionally, it opens new avenues for future research, such as investigating stock return volatility and trading volume fluctuations in high-temperature and high-humidity environments, examining changes in risk preferences and belief updates under these conditions, and assessing the potential impacts on corporate production and operational activities.

9. Conclusions

An increasing number of scholars have begun to focus on the economic and financial implications of climate change, aiming to integrate climate into analytical frameworks and explore its impact channels. This study specifically investigates the influence of extreme weather on stock market returns, employing theoretical and empirical analyses from both investor and corporate perspectives to verify underlying mechanisms. Using daily weather variables for A-share listed companies across various cities from 2015 to 2019, we construct proportions of extreme hot days and extreme humid days, incorporating multiple seasonally adjusted monthly weather variables. We establish a multivariate OLS model controlling for firm-level, yearly, and quarterly fixed effects, revealing a negative impact of proportions of extreme hot days and extreme humid days on stock returns. For every standard deviation increase in proportions of extreme hot days and extreme humid days, stock annualized returns decrease by 0.09% and 0.15%, respectively. Additionally, there exists a relatively small yet significant correlation between average monthly weather control variables and stock returns.

To explore how extreme weather affects stock prices, we conduct mediation analysis. At the individual level, weather changes primarily affect investor sentiment through beliefs and preferences, thereby influencing stock price movements. At the corporate level, changes in the proportions of extreme hot days or extreme humid days do not significantly alter total revenue, productivity, or net profit, even after excluding samples headquartered in Beijing and Shanghai. Investor sentiment serves as the primary mediator linking weather and stock returns. Furthermore, in the heterogeneity analysis, results akin to the previous literature are obtained [86]. Companies with large asset size, strong profitability, and low volatility are less affected by extreme weather compared with those with small asset size, weak profitability, and high volatility. Specifically, for every standard deviation increase in the proportions of extreme temperature days and extreme humid days, the former’s annual stock returns decrease by 0.05% and the latter’s by 0.18%. Finally, to mitigate the dominance of heat-sensitive industries in the sample and the impact of corporate announcements on stock prices, this study reanalyzes the data after excluding relevant samples. Moreover, various methods such as incorporating additional control variables or altering the form of the dependent variable are employed in the models, all of which pass robustness checks.

Climate risks pose new considerations for individuals, businesses, and governments. Firstly, investors should be aware that extreme weather can influence emotions, affecting information interpretation, corporate expectations, and risk preferences. It is advisable for investors to manage their emotions and psychological states, avoiding impulsive decisions during extreme weather conditions, and exercising caution when entering the market. Secondly, prudent asset allocation should involve rational analysis of local and non-local enterprises, avoiding irrational influences on asset allocation. Diversification strategies can mitigate climate risks. The previous literature indicates that weather changes affect not only individual investors but also institutional investors [36]. Institutional investors, managing substantial funds, may experience greater impacts on their trading behaviors from extreme weather. Therefore, institutional investors must manage their emotions effectively, regularly assess climate change impacts on portfolios, and develop corresponding risk mitigation strategies.

For enterprises, strategic planning should include careful consideration of climate change impacts on capital markets. Incorporating climate factors into decision making, company valuation, and stock price forecasts is crucial. Establishing climate risk assessment mechanisms to periodically monitor operational and financial performance impacts is recommended. Furthermore, companies should recognize that extreme weather can influence stock prices through investor sentiment. Timely disclosures, stock repurchases, or other measures can mitigate investor decisions during extreme weather events, reducing market uncertainty. Additionally, analyzing investor behavior with a focus on home biases in company analysis should be prioritized. Different weighting strategies should be applied based on company heterogeneity, particularly for smaller companies with relatively weaker profitability, enhancing resilience against weather-related risks through climate insurance purchases or emergency funds. While this study finds no significant impact of extreme weather on corporate productivity, businesses should remain vigilant against climate risks such as extreme disasters threatening operations. Precautionary measures should be taken in advance.

For governments, this study confirms the impact of extreme hot days on stock returns and underscores the significant influence of extreme humidity. This highlights the importance of not neglecting relative humidity and other weather indicators, with necessary periodic tracking of daily weather trends. For instance, real-time collection and analysis of various weather data through climate monitoring systems can provide scientific bases for policy formulation. Government attention should also prioritize social sustainability, enhancing climate risk management systems and policies. This includes guiding financial institutions to increase focus on climate risks, improving market information transparency, and reducing information asymmetry. Incentive policies encouraging climate risk management by companies and supervising climate risk management practices are crucial. Additionally, governments should raise investor awareness of climate risks, promote investor mental health, and encourage emotional management.

Future research can further utilize the methods for measuring climate risks in this study to verify their impact on corporate CSR and ESG indicators and explore underlying mechanisms. Given data availability, exploring how extreme weather affects individual investor sentiment and consequently stock prices is essential. Although foreign scholars have identified significant impacts of cloud cover on institutional investor trading behaviors and stock returns, further exploration is needed in China—where individual investors constitute a majority—to understand how extreme weather affects individual investor sentiment.