Does Weather-Related Disaster Affect the Financing Costs of Enterprises? Evidence from Chinese Listed Companies in the Mining Industry

Abstract

In this paper, we test the impact of weather-related disasters on the individual firm’s equity financing cost based on Chinese listed companies in the mining industry. We collect data from the China Meteorological Disaster Yearbook and CSMAR database. Using direct economic loss associated with extreme weather-related events to quantitate meteorological disasters and regression analysis, we find that weather-related disasters significantly increase a firm’s equity financing cost. This result is robust compared to alternative measurements of equity financing cost, such as the two-way fixed effect model, severe disaster dummy variable, and instrumental variable regression. Further research shows that cash holdings and managerial ability can mitigate the impact of meteorological disasters on the equity financing cost. Our study provides significant implications for firms and policymakers. Firms and policymakers should carefully evaluate the risk of weather-related disasters.

1. Introduction

The frequency of extreme weather events such as severe precipitation and serious drought is on the rise and this trend will continue in the future Tachiiri et al. [1]. The increase in extreme weather events poses a huge threat to human society. For example, the global extreme weather events from 1999 to 2018 caused more than 500,000 casualties and more than USD 3.5 trillion in damage Kreft et al. [2]. Although researchers have an almost consistent opinion on the consequence of extreme weather, the impact of meteorological disasters on a firm’s equity financing cost has, so far, not been well studied. To the best of our knowledge, there are only two studies on the effect of climate risk on the equity financing cost. However, their conclusions are mixed. (Huynh et al., 2020) found a statistically significant and positive correlation between drought risk and a firm’s equity financing cost [3]. (Kling et al., 2021) found no evidence that climate risk affects a firm’s equity financing cost using panel data from 71 countries [4]. Therefore, whether climate risk affects equity financing cost needs more empirical tests. Furthermore, neither of the above two studies examined the Chinese market. To fill this research gap, we aim to present fresh evidence by examining the impact of meteorological disasters on equity financing cost based on Chinese listed companies in the mining industry.

The economic activities of the mining industry are directly dependent on weather and, hence, the most vulnerable to meteorological disasters. Meteorological disasters bring damage to the construction of the mining industry (Pearce [5]). Meteorological disasters increase the physical difficulty of mining operations and therefore affect mineral resource supply. In the context of the warming climate, the decarbonization target of many industries will lead to adverse effects on the commodity demand of the mining industry. The impacts of climate change on the mining industry have received attention (Pearce, 2011, Sharma et al., 2013) [5,6]. However, there are few studies focusing on the impacts of meteorological disasters on the equity financing cost in the mining industry. The equity financing cost is one of the key factors for firms to achieve sustainable development. Examining the relation between meteorological disasters and the equity financing cost should assist managers in the resource industry in obtaining a better understanding about the consequence of climate risks and encourage a firm’s manager to become more concerned about climate change.

Our interest in the impact of meteorological disasters on equity financing costs is inspired by the following literature. First, equity financing cost is the required rate of return given investors’ perceptions of a firm’s risk Ghoul et al. [7]. Recent research suggests that climate risk negatively affects the firm’s financial performance Sun et al. [8]. Therefore, meteorological disasters affect investors’ perceptions of firm’s risk. Consequently, firms suffering from meteorological disasters should bear higher equity financing costs. Second, theoretical disaster models indicate that the disaster risk premium will likely be low and smooth during normal times but rise substantially after a disaster Barro, 2006; Chen et al. [9,10]. Therefore, the weather risk premium should be higher after a meteorological disaster event. As a result, the equity financing cost for those firms affected by meteorological disasters is higher. Third, meteorological disasters are scattered over time and the impacts are concentrated on specific companies in the affected areas. Therefore, weather risk is not an indivisible systemic risk. According to traditional asset pricing theory, weather risk can be dispersed across the whole market portfolio. Thus, there should be no risk premium for individual firms. Therefore, whether climate risk affects the cost of equity capital is an empirical issue.

Our paper show that weather-related disasters significantly increase a firm’s equity financing cost. Further research shows that cash holdings and managerial ability can mitigate the impact of meteorological disasters on the equity financing cost. Our paper not only enriches the academic research on the consequences of climate change, but also provides feasible suggestions for enterprises on how to mitigate the adverse effects of disaster shocks.

The rest of the paper is organized as follows: Section 2 reviews the relevant prior literature. Section 3 discusses variable definitions and samples. Section 4 presents the empirical results. Section 5 is the conclusion.

2. Literature Review

Disasters related to extreme weather can cause a rise in a firm’s operation risk owing to damage to the operating assets and production output of firms. Prior studies have presented evidence of climate risk having a negative economic impact. For example, stronger storms result in an increase in fatality and sizable losses in the economy (Yang [11]). Climate change disrupts related economic activities (Fuss [12]). Unfavorable weather conditions bring about lower economic growth rates and even economic losses Hsiang et al. [13]. If the economic impacts of climate change are as large as the above studies have suggested, the impacts of climate risk on individual firms could also be significant because individual firms constitute the main body of macro-economic activities. Extreme weather events can directly destroy or accelerate the depreciation of a firm’s fixed assets. Therefore, scholars recently began to pay attention to the consequence of climate change at the firm level. For example, (Huang, et al., 2018) found that firms whose location is in disaster-affected countries are more likely to have lower and more volatile earnings and cash flows [14]. (Sun et al., 2020) showed that rain waterlogging, high temperature, and cryogenic freezing risks lead to a decline in a firm’s financial performance in the mining industry [8]. (Zhou et al., 2021) found evidence that storms and flooding negatively impact on a firm’s sales growth [15].

We hold the opinion that there will be a higher level of information asymmetry for companies affected by disasters due to earnings management. Firm managers often resort to manipulating earnings in order to achieve target earnings. Managers have motivations to engage in earnings management when they are concerned about their performance in the current period e.g., Stein, 1989; Rodriguez-Perez et al. [16,17]. The adverse effects of meteorological disasters on a firm’s operation would give managers an impulse to undertake earnings manipulations due to the poor performance in bad times. For example, firm managers have a stronger willingness to engage in earnings management during an economic recession (Strobl [18]). The motivation to engage in earnings management will be more significant if the manager expects a firm’s profit to be lower than the average profit of the industry Wang et al. [19]. In the years of a disaster and after, managers may achieve certain earnings goals (e.g., earnings specified in a manager’s compensation contract) through earnings management. The effects of natural disasters on earnings management are most predominant during the first three years following disasters Wu et al. [20].

Previous studies have given consistent conclusions on the impact of earnings management on equity financing costs. (Lambert et al., 2007) and (Kim et al., 2013) suggest that the purpose of earnings management is to hide a firm’s real earnings performance and exaggerate the quality of information used by outside investors [21,22]. (Easley et al., 2004) examined the effect of information on firms’ equity financing costs and found that the return required by outside investors is positively related to the level of information asymmetry [23]. This higher required return illuminates the fact that information asymmetry increases the risk to hold that stock for outside investors. Consequently, the equity financing cost is positively correlated with the degree of earnings management (Lambert, et al., 2007; Aboody et al., 2005; Kim et al., 2013) [21,22,24]. (Bertomeu et al., 2016) provide a review on accounting information as a determinant of the equity financing cost [25].

It can be seen from the above literature that when a disaster occurs, the business performance of firms in that region will be affected. Managers are motivated to conduct earnings management, which affects the quality of a firm’s earnings. With high information asymmetry, outside investors demand a higher risk premium, which eventually induces the equity financing cost to rise.

Experience with disasters alerts investors to climate risk and induces a growing cognition of climate risk Choi et al. [26]. Global warming is a long-term trend often ignored by people. However, a local meteorological disaster is more visible to an investor. Because investor attention is not unlimited, they may pay more attention to striking disaster events. The recent literature on climate risk has also looked at the consequences of a personal experience of disaster events. Zaval et al. (2014) indicate that people’s consciousness about climate risk would rise due to their personal experience of a meteorological disaster [27]. This is also confirmed by Broomell et al. [28]. (Konisky et al., 2016) found extreme weather events (such as droughts and flooding) increased investor concern about climate risk [29]. (Li et al., 2011) further point out that people’s perception of deviations from normal temperatures not only changes their beliefs but also leads to actions [30]. If investors are increasingly aware of climate change risks, they may buy stocks not affected by disasters and sell stocks affected by disasters, such that the former outperform the latter. This reflects the fact that investors have higher risk premium expectations for stocks affected by the disaster when other conditions are equal.

By reviewing the above literature, we have learned that scholars have studied the consequences of meteorological disasters from a firm’s perspective. However, the existing literature mainly focuses on the impact of meteorological disasters on a firm’s performance, earnings management, and risk premium. Only two studies (Huynh et al., 2020; Kling et al., 2021) discussed the impact of meteorological disasters on corporate financing costs [3,4]. It is worth noting that these two studies did not discuss the Chinese market. Our research attempts to fill this gap.

How to mitigate the adverse effects of meteorological disasters on enterprises? The existing literature on managerial ability and cash holdings provides us with ideas.

Firm managers are more knowledgeable about their business. A manager can affect a firm’s earnings through her/his ability and effort (Bhagat et al., 2011) [31]. Managerial ability is positively related to investment opportunities (Lee et al., 2018), earnings quality (Demerjian et al., 2013), and firm value (Yung et al., 2020) [32,33,34]. The ability of managers is conducive to enterprises when firms face outside threats (Yung et al., 2020) [34]. When high-ability managers use earnings management, their earnings management is associated with better firm performance Huang et al. [35]. Motivated by the literature on managerial ability, we examine the role of managerial ability on mitigating the impact of meteorological disasters in this study.

Why do firms hold cash or cash equivalents? Firms can hedge the risk of future cash shortage by reserving cash. This is the precautionary motive for holding cash. For example, Opler et al. (1999) showed that firms tend to hold more cash if their cash flow volatility is higher [36]. (Han et al., 2007) modeled the precautionary motive for a firm’s cash holdings [37]. Higher cash holdings can reduce the risk refinancing Harford et al. [38]. Cash precautionary reserves are beneficial to the firm during the period of adverse climate impact. Therefore, we can expect that cash reserves enable firms to mitigate the adverse effects of meteorological disasters.

3. Methodology and Sample

3.1. The Equity Financing Cost

We refered to existing literature in accounting and finance to calculate the ex ante equity financing cost implied in current stock prices and analyst forecasts. The three models introduced by (Ohlson et al., 2005) (i.e., OJN model) [39] and (Easton, 2004) (i.e., PEG and MPEG model) were used to calculate an individual firm’s equity financing cost [40]. We designated these resulting equity financing costs as OJN, PEG and MPEG, respectively. In order to eliminate the influence of measurement error of individual estimates, we calculated the average cost of equity based on the three models. This yielded COE, which is the implied equity financing cost that we used as our dependent variable. Specific variables were defined as follows:

$$PEG=\sqrt{\left(EP{S}_{t+2}-EP{S}_{t+1}\right)/P_t},$$

$$MPEG=\sqrt{\left(EP{S}_{t+2}+R_s\times DP{S}_{t+1}-EP{S}_{t+1}\right)/P_t},$$

$$OJN=A+\sqrt{A^2+EP{S}_{t+1}/P_t\times \left((EP{S}_{t+2}-EP{S}_{t+1}\right)/EP{S}_{t+1}-\left(\gamma -1\right))},$$

$$COE=\left(PEG+MPEG+OJN\right)/3,$$

where $A=\left(\gamma -1+DP{S}_{t+1}/P_t\right)/2$, EPS is the earnings per share of the firm, P is the stock closing price, DPS is dividend per share, and γ is long term earnings growth rate.

3.2. Quantitating Meteorological Disasters

We used the direct economic loss (denoted by Loss) associated with extreme weather-related events to quantitate meteorological disasters by province. Due to the inability to obtain detailed data at the plant level, we followed (Huynh et al., 2020) and assumed that the headquarters and the main production facilities of a firm are often located in the same province [3]. Firms headquartered in provinces influenced by climate risk are exposed to higher business risk, and therefore the cost of capital rises. So, we used the province where a firm’s headquarters are located to decide whether the firm suffers from the disaster.

3.3. Control Variables

Referring to the existing literature (e.g., Ohlson et al., 2005; Easton 2004; Huynh et al., 2020; Ghardallou, 2023; Ghardallou, 2022) on the equity financing cost [3,39,40,41,42], this paper selected the firm’s total assets (Size), return on total assets (ROA), financial leverage (Lev), book-to-market ratio (BM), systemic risk (Beta), and year dummy variables as the control variables.

3.4. Baseline Regression Model

To examine the impact of meteorological disasters on the equity financing cost, the following static panel model was constructed

$$CO{E}_{i,t}={\beta }_0+{\beta }_{1}Los{s}_{i,t-1}+\delta Control{s}_{i,t-1}+Year+{\epsilon }_{i,t-1}$$

(1)

where COE is the average equity financing cost of model OJN, PEG, and MPEG. Loss is the meteorological disaster loss in the province where the firm’s headquarters are located at year t. Year is the year dummy variable.

3.5. Sample

China is one of the countries most affected by climate change. From 1998 to 2017, global meteorological disasters caused losses of more than USD 2 trillion, among which China’s losses ranked second in the world Feng et al. [43]. The Chinese government is already aware of the consequences of climate change risks. However, firms in China have an insufficient understanding of the consequences of climate change. For example, Chinese mining companies are still lagging behind in addressing climate change risks. Therefore, we used China as a typical case to warn relevant enterprises that they cannot ignore the financial risks brought about by climate change. The sample of this paper was the 94 firms of the mining industry with shares listed on Shanghai and Shenzhen A from 2004 to 2019. The direct economic losses were manually compiled according to the China Meteorological Disaster Yearbook from 2005 to 2020. These direct economic losses were related to drought, rainstorm, flood, hail and other disasters caused by meteorological factors. Other data were from CSMAR database.

4. Empirical Results

4.1. Descriptive Statistics

Table 1. Descriptive statistics.

Variables	Mean	Std. Dev.	Min	Max
OJN	0.126	0.0428	0.0324	0.303
MPEG	0.108	0.0434	0.0159	0.265
PEG	0.098	0.0416	0.0084	0.274
COE	0.116	0.0424	0.0189	0.288
Loss (Billion)	9.756	10.61	0	83.77
ROA	1.245	5.193	−26.46	78.38
Lev	45.36	20.08	2.027	116.4
BM	0.672	0.250	0.0877	1.273
Beta	1.192	0.323	0.134	2.429
lnSize	23.19	1.685	19.70	28.52

Notes: This table reports mean, standard deviation, min. and max of all dependent variables, independent variables of interest, and control variables. OJN, MPEG, and PEG represent different equity financing costs from three models. COE was the average cost of equity calculated based on the three models. Loss means the direct economic loss associated with extreme weather-related events (The calculation unit is billion). ROA, Lev, BM, Beta and Insize mean return on total assets, financial leverage, book-to-market ratio, systemic risk, and the logarithm of a firm’s total assets, respectively.

and

Table 2. Correlation coefficients.

	PEG	MPEG	OJN	COE	lnLoss	ROA	lnSize	Lev	BM	Beta
PEG	1
MPEG	0.967	1
OJN	0.958	0.971	1
COE	0.967	0.996	0.987	1
lnLoss	0.040	0.049	0.044	0.047	1
ROA	−0.004	−0.018	−0.033	−0.024	−0.011	1
lnSize	0.102	0.139	0.094	0.129	0.002	−0.232	1
Lev	0.135	0.116	0.102	0.113	0.089	−0.162	0.450	1
BM	0.164	0.222	0.137	0.200	−0.070	−0.213	0.533	0.362	1
Beta	−0.048	−0.022	−0.038	−0.026	−0.035	−0.052	0.191	0.018	0.054	1

Notes: This table reports correlation coefficients of the variables. OJN, MPEG, and PEG represent different equity financing costs from three models. COE was the average cost of equity calculated based on the three models. Loss means the direct economic loss associated with extreme weather-related events (The calculation unit is billion). ROA, Lev, BM, Beta and Insize mean return on total assets, financial leverage, book-to-market ratio, systemic risk, and the logarithm of a firm’s total assets, respectively.

list descriptive statistics and correlation coefficients of variables, respectively. Table 1 shows that the average equity financing cost (COE) is 11.6%. The cost of PEG model is the smallest, while that of the OJN model is the largest. The standard deviation of the three cost measures is about 4%. The mean Loss of the sample is CNY 9.756 billion and the standard deviation is CNY 10.61 billion. In 2016, the biggest economic loss of meteorological disasters (mainly floods) in Hubei province is CNY 83.77 billion yuan, with 23.31 million people affected and 117 dead.

4.2. Baseline Regression Results

To examine the impact of meteorological disasters on the equity financing cost, the following static panel model was constructed

$$CO{E}_{i,t}={\beta }_0+{\beta }_{1}Los{s}_{i,t-1}+\delta Control{s}_{i,t-1}+Year+{\epsilon }_{i,t-1}$$

(2)

where COE is the average equity financing cost of models OJN, PEG, and MPEG. Loss is the meteorological disaster loss in the province where the firm’s headquarters are located at year t. Year is the year dummy variable.

Table 3. Regression results.

	(1)	(2)	(3)
Variables	COE	COE	COE
Loss (×10−5)	4.82 **	6.03 **	6.13 **
	(2.251)	(2.443)	(2.513)
ROA (×10−4)		−9.07 **	−10.10 **
		(−2.122)	(−2.512)
lnSize (×10−3)		6.53 **	4.99 *
		(2.433)	(1.909)
Lev (×10−4)		3.28 **	1.39
		(2.078)	(0.900)
BM (×10−2)		−4.69 ***	−0.42
		(−3.652)	(−0.264)
Beta (×10−2)		−0.91	−1.53
		(−1.027)	(−1.635)
Constant	0.103 ***	−0.012	0.008
	(6.890)	(−0.216)	(0.152)
Year	Yes	No	Yes
R-squared	0.158	0.101	0.241

Notes: This table reports the regression results. COE is the average cost of equity calculated based on the three models. Loss means the direct economic loss associated with extreme weather-related events (The calculation unit is billion). ROA, Lev, BM, Beta and Insize mean return on total assets, financial leverage, book-to-market ratio, systemic risk, and the logarithm of a firm’s total assets, respectively. *, **, *** indicate significant at the 10%, 5%, and 1% levels, respectively, with t-statistics in parentheses. The meaning of 10-i is scientific notation. For example, 4.82 × 10⁻⁵ is 4.82 × 0.00001.

presents the regression results relating to the effect of meteorological disasters on the equity financing cost. In all columns, we find the coefficients of Loss to be significantly positive (at level 5%), indicating that higher economic losses caused by meteorological disasters are significantly related to higher equity financing cost. In terms of economic significance, the estimated coefficient from Model (1) shows that an increase in Loss by one unit leads to an increase of 0.613 bps in a firm’s equity financing cost. The estimated coefficient also shows that an increase in the standard deviation of Loss means an increase of 6.5 basis points (=0.613 × 10.61) in the equity financing cost.

Our result is inconsistent with Kling et al. [4], but consistent with finding of Huynh et al. [3]. (Huynh et al., 2020) found that the cost of equity capital is significantly and positively correlated with drought risk in the U.S [3]. Our result supports the argument of Huynh et al. [3]. Meanwhile, there are at least two differences between our paper and Huynh et al. [3]. First, we provide empirical evidence on the Chinese market. Second, we use the economic losses related to meteorological disasters to measure climate risk, while only drought is examined in Huynh et al. [3]. We also find that the impact of climate risk on the financing cost of Chinese firms is smaller than that of the U.S. This finding shows that there are differences between Chinese and American market investors in their understanding of the impact of climate risk on enterprises. This may be related to the difference in the investor structure of the two markets. As we all know, the Chinese market is dominated by individual investors while the American market is dominated by institutional investors.

4.3. Robustness Test

4.3.1. Meteorological Disaster and the Individual Equity Financing Cost

We next used each of the three measures of the equity financing cost as the first robustness test.

Table 4. Individual equity financing cost and disaster.

	(1)	(2)	(3)
Variables	PEG	MPEG	OJN
Loss (×10−5)	6.29 ***	6.75 ***	6.10 **
	(2.626)	(2.785)	(2.447)
ROA (×10−4)	−8.48 **	−9.59 **	−13.2 ***
	(−2.104)	(−2.364)	(−3.195)
lnSize (×10−3)	5.46 **	7.79 ***	4.02
	(2.207)	(3.049)	(1.505)
Lev (×10−4)	1.12	0.17	1.27
	(0.827)	(0.120)	(0.803)
BM (×10−3)	3.58	−5.59	−4.86
	(0.241)	(−0.362)	(−0.296)
Beta (×10−2)	−1.51 *	−1.50	−1.54
	(−1.675)	(−1.621)	(−1.612)
Constant (×10−2)	−2.24	−5.63	3.65
	(−0.449)	(−1.102)	(0.674)
R-squared	0.173	0.232	0.208

Notes: This table reports the individual equity financing cost and disaster. OJN, MPEG, and PEG represent different equity financing costs from three models. Loss means the direct economic loss associated with extreme weather-related events (The calculation unit is billion). ROA, Lev, BM, Beta and Insize mean return on total assets, financial leverage, book-to-market ratio, systemic risk, and the logarithm of firm’s total assets, respectively. *, **, *** indicate significant at the 10%, 5%, and 1% levels, respectively, with t-statistics in parentheses. The meaning of 10-i is scientific notation. For example, 6.29 × 10⁻⁵ is 6.29 × 0.00001.

presents the regression results.

The coefficients on Loss in Columns (1) to (3) are positive and statistically significant at least at the level of 5%. The coefficients on Loss in Columns (1) to (3) are 0.0000629, 0.0000675, and 0.0000610, respectively. These coefficients show that an increase in one standard deviation of Loss over the last year translates to a 6.67 bps, 7.16 bps, and 6.47 bps increase in PEG, MPEG, and OJN, respectively. The results in Table 4 are consistent with that in Table 3, which suggest that disasters are associated with an increase in the equity financing cost.

4.3.2. Two-Way Fixed Effect Model

To mitigate concerns that our results might be affected by omitted firm-level characteristics, we included both time and firm fixed effects in the baseline model and re-ran the regression analysis.

Table 5. Estimation results of two-way fixed effect model.

	(1)	(2)	(3)	(4)
Variables	COE	OJN	PEG	MPEG
Loss (×10−5)	4.88 *	5.05 *	5.32 *	4.29 *
	(1.728)	(1.786)	(1.996)	(1.660)
ROA (×10−4)	−11.6 ***	−12.2 ***	−5.25	−8.41
	(−2.744)	(−2.689)	(−1.278)	(−1.669)
lnSize (×10−3)	5.62	0.88	2.34	1.28
	(0.579)	(0.091)	(0.313)	(0.151)
Lev (×10−4)	−4.38 *	−3.89	−3.15	−3.53
	(−1.743)	(−1.530)	(−1.244)	(−1.309)
BM (×10−2)	−4.81 **	−3.62 *	−3.81 **	−3.86 *
	(−2.381)	(−1.783)	(−2.028)	(−1.767)
Beta (×10−2)	−2.13 *	−1.70	−1.10	−0.34
	(−1.725)	(−1.312)	(−1.018)	(−0.253)
Constant (×10−2)	5.54	14.7	8.23	11.1
	(0.269)	(0.715)	(0.530)	(0.632)
R-squared	0.306	0.252	0.194	0.305

Notes: This table reports the estimation results of two-way fixed effect model. OJN, MPEG, and PEG represent different equity financing costs from three models. COE is the average cost of equity calculated based on the three models. Loss means the direct economic loss associated with extreme weather-related events (The calculation unit is billion). ROA, Lev, BM, Beta and Insize mean return on total assets, financial leverage, book-to-market ratio, systemic risk, and the logarithm of a firm’s total assets, respectively. *, **, *** indicate significant at the 10%, 5%, and 1% levels, respectively, with t-statistics in parentheses. The meaning of 10-i is scientific notation. For example, 4.88 × 10⁻⁵ is 4.88 × 0.00001.

presents the results. Although the coefficients on Loss decrease, these coefficients are still significantly positive at a level of 10%. Therefore, the conclusion of the two-way fixed effect model is consistent with Table 3.

4.3.3. Severe Meteorological Disaster

As an alternative measure of disaster magnitude, we constructed a dummy variable which captured severe meteorological disasters in the province where a firm is headquartered. The variable HighLoss was set to one when the disaster loss of the province is greater than the national average disaster loss, and zero otherwise. The results are shown in

Table 6. The impact of severe meteorological disaster on equity financing cost.

	(1)	(2)	(3)	(4)
Variables	COE	OJN	PEG	MPEG
HighLoss (×10−2)	1.18 ***	1.19 ***	1.36 ***	1.12 ***
	(2.722)	(2.723)	(3.408)	(2.661)
ROA (×10−4)	−10.5 **	−11.7 **	−5.17	−6.05
	(−2.372)	(−2.500)	(−1.229)	(−1.449)
lnSize (×10−3)	1.31	0.187	2.88	2.91
	(0.285)	(0.0397)	(0.745)	(0.695)
Lev (×10−4)	−4.29 *	−3.90	−3.17	−3.29
	(−1.709)	(−1.576)	(−1.235)	(−1.409)
BM (×10−2)	−7.86 ***	−6.66 ***	−5.30 ***	−7.42 ***
	(−5.871)	(−5.159)	(−4.214)	(−5.679)
Beta (×10−3)	−16.2	−14.6	−8.44	−9.37
	(−1.515)	(−1.356)	(−0.965)	(−0.904)
Constant (×10−2)	17.0 *	19.4 *	8.65	10.9
	(1.679)	(1.872)	(1.024)	(1.160)
R-squared	0.188	0.158	0.126	0.176

Notes: This table reports the impact of severe meteorological disaster on equity financing cost. OJN, MPEG, and PEG represent different equity financing costs from three models. COE is the average cost of equity calculated based on the three models. ROA, Lev, BM, Beta and Insize mean return on total assets, financial leverage, book-to-market ratio, systemic risk, and the logarithm of firm’s total assets, respectively. The variable HighLoss is set to one when the disaster loss of the province is greater than the national average disaster loss, and zero otherwise. *, **, *** indicate significant at the 10%, 5%, and 1% levels, respectively, with t-statistics in parentheses. The meaning of 10-i is scientific notation. For example, 1.18 × 10⁻² is 1.18 × 0.01.

. The coefficients on HighLoss are positive and statistically significant at a level of 1%. Compared with Table 3 and Table 4, the results of two-way fixed effect models in Table 6 clearly show that the impact of severe meteorological disasters on equity capital cost is more significant.

4.3.4. Instrumental Variable Regression

Although meteorological disaster is an exogenous shock, the proxy variables used above may have measurement errors. Additionally, because some factors at the provincial and firm level are difficult to quantify and control, this section uses the instrumental variable method to re-estimate the model as another robustness test. Because population density is significantly related to climate risk, it is unlikely to be related to corporate financial indicators Huang et al. [14]. Therefore, we followed (Huang et al., 2018) to use population density as an instrumental variable of direct economic loss [14]. The results of instrumental variable regressions in

Table 7. Instrumental variable regression results.

Variables	(1)	(2)	(3)	(4)
First-stage regression	Loss	Loss	Loss	Loss
population density	0.040 *** (4.88)	0.041 *** (4.87)	0.039 *** (5.23)	0.039 *** (5.12)
Controls	Yes	Yes	Yes	Yes
Instrumental variable regression	COE	OJN	PEG	MPEG
Loss (×10−4)	1.70 **	1.82 **	1.30	1.61 *
	(1.970)	(2.034)	(1. 572)	(1.908)
ROA (×10−4)	−9.63 **	−12.6 ***	−8.20 **	−9.17 **
	(−2.357)	(−3.001)	(−2.053)	(−2.254)
lnSize (×10−3)	5.73 **	4.82 *	5.61 **	8.18 ***
	(2.130)	(1.745)	(2.289)	(3.179)
Lev (×10−4)	1.59	1.50	1.37	0.49
	(1.020)	(0.931)	(0.996)	(0.332)
BM (×10−3)	−6.34	−6.94	2.87	−7.18
	(−0.389)	(−0.416)	(0.196)	(−0.464)
Beta (×10−2)	−1.11	−1.07	−1.25	−1.14
	(−1.116)	(−1.049)	(−1.318)	(−1.177)
Constant (×10−2)	−1.61	0.99	−3.15	−7.26
	(−0.286)	(0.171)	(−0.625)	(−1.372)
R-squared	0.185	0.140	0.152	0.193

Notes: This table reports the instrumental variable regression results. Loss means the direct economic loss associated with extreme weather-related events (The calculation unit is billion). ROA, Lev, BM, Beta and Insize mean return on total assets, financial leverage, book-to-market ratio, systemic risk, and the logarithm of firm’s total assets, respectively. *, **, *** indicate significant at the 10%, 5%, and 1% levels, respectively, with t-statistics in parentheses. The meaning of 10-i is scientific notation. For example, 1.70 × 10⁻⁴ is 1.70 × 0.0001.

still show that meteorological disaster has a significantly positive impact on equity financing cost.

4.4. The Moderating Effects

The above empirical research has shown that disasters are associated with an increase in the equity financing cost. Another core issue is whether the impact of meteorological disasters on costs can be mitigated. Firms that are highly exposed to disaster risk may take measures to reduce the impact of disaster on equity financing costs. In other words, firms are motivated to mitigate the economic consequences of meteorological disaster. Here, we consider two possible mitigation methods, i.e., cash holdings and managerial ability.

First, cash holdings can be a risk management tool to mitigate the impact of climate risk on equity financing costs. The preventive motivation plays a key role in explaining cash holdings Bates et al. [44]. Sufficient cash holdings have an effective and positive influence on mitigating unhedged risks Huynh et al. [3]. The literature has reached a consistent conclusion, that is, the substantial increase in corporate cash holdings is consistent with the view that cash demand is to mitigate many risks against which enterprises cannot effectively hedge. We therefore anticipate that cash holding is an effective way for firms to reduce the influence on the equity financing cost caused by disaster risks. We tested the moderating effect of cash holding using the following equation.

$$CO{E}_{i,t}={\beta }_0+{\beta }_1\ln Los{s}_{i,t-1}+Cas{h}_{i,t-1}+\ln Loss\times Cas{h}_{i,t-1}+\delta Control{s}_{i,t-1}+Year+{\epsilon }_{i,t-1}$$

(3)

where $Cas{h}_{i,t-1}$ is firm i cash holding at year $t-1$. $\ln Loss$ is the logarithm of $Loss$. The interaction item (i.e., $\ln Loss\times Cas{h}_{i,t-1}$) is designed to capture the moderating effect of cash holdings.

Second, we anticipated managerial ability to be helpful in moderating the impact of disaster on equity financing costs. (1) Managers differ in their ability to manage a firm’s resources and to enhance a firm’s performance. Managerial ability is a firm’s unique resource and hard to imitate. (Holcomb et al., 2009) demonstrated that managerial ability can be a source of value creation [45]. (2) Those with a higher managerial ability have a better knowledge of the firm’s business environment. This understanding makes managerial decisions align better with the business risk faced by the firm Demerjian et al. [46]. (Bonsall IV et al., 2016) showed that a more competent manager is helpful for reducing the volatility of firm’s future performance and stock returns [47]. (Phan et al., 2020) demonstrated that managerial ability plays a decisive role in the negative impact of crude oil price uncertainty on a firm’s performance [48]. Firms with high scores of managerial ability are impacted less than those with low scores of managerial ability. (Yung et al., 2020) indicated that managerial ability enhances a firm’s value in the face of competition. Therefore, we anticipated that managers with higher ability may develop more effective plans to mitigate the adverse effect on a firm’s performance caused by disasters [34].

The following equation was used to test the moderate effect of managerial ability.

$$CO{E}_{i,t}={\beta }_0+{\beta }_1\ln Los{s}_{i,t-1}+M{A}_{i,t-1}+\ln Loss\times M{A}_{i,t-1}+\delta Control{s}_{i,t-1}+Year+{\epsilon }_{i,t-1}$$

(4)

where $M{A}_{i,t-1}$ is dummy variable. The variable MA takes the value of one when the firm i’s managerial ability score is greater than the year average, and zero otherwise. $\ln Loss$ is the logarithm of $Loss$. The interaction item (i.e., $\ln Loss\times M{A}_{i,t-1}$) is designed to capture the moderating effect of managerial ability. We used the method developed by (Demerjian et al., 2012) to measure managerial ability [46]. We distinguished managerial talent and a number of driving factors affecting firms’ efficiency to quantify manager ability. Specifically, we first used data envelopment analysis to measure a firm’s efficiency θ. A firm’s fixed assets investment, acquisition goodwill, and research and development investment were used as input factors and a firm’s operating income was used as the output factor to obtain the firm efficiency. Then, we separated firm efficiency θ into firm level factors and manager level factors to obtain the independent measurement index of manager ability. We performed Tobit regression on the firm’s efficiency θ to eliminate firm-level factors in their efficiency. The residual of the Tobit regression is our estimate of MA. In the Tobit regression equation, firm-level factors (i.e., explanatory variables) include firm size, market shares, listing time, business segment concentration, free cash flow, foreign currency indicators, and annual fixed effect.

The regression results are listed in

Table 8. Mitigation effect.

	(1)	(2)
Variables	COE	COE
lnLoss (×10−3)	4.78 **	8.76 *
	(2.062)	(1.951)
Cash (×10−12)	9.86 **
	(2.318)
Cash_lnLoss (×10−12)	−2.14 *
	(−1.911)
MA (×10−2)		3.52 **
		(2.347)
MA_lnLoss (×10−3)		−0.902 **
		(−2.601)
ROA (×10−3)	−1.87 ***	−1.19 ***
	(−5.237)	(−2.853)
lnSize (×10−3)	−0.479	6.40
	(−0.0767)	(0.665)
Lev (×10−4)	−2.22	−3.80
	(−1.012)	(−1.586)
BM (×10−2)	−1.41	−4.88 **
	(−0.843)	(−2.413)
Beta (×10−2)	−0.549	−1.84
	(−0.572)	(−1.505)
Constant	0.119	0.0046
	(0.921)	(0.022)
R-squared	0.256	0.320

Notes: This table reports mitigation effect. COE is the average cost of equity calculated based on the three models. lnLoss is the logarithm of Loss., Cash represents cash holding, and Cash lnLoss is their interaction item. MA represents managerial ability, which is a dummy variable. MA_lnLoss is the interaction item of MA and lnLoss. ROA, Lev, BM, Beta, and Insize mean return on total assets, financial leverage, book-to-market ratio, systemic risk, and the logarithm of a firm’s total assets, respectively. *, **, *** indicate significant at the 10%, 5%, and 1% levels, respectively, with t-statistics in parentheses. The meaning of 10-i is scientific notation. For example, 4.78 × 10⁻³ is 4.78 × 0.001.

. Column (1) is the results of cash holding and Column (2) is the results of managerial ability. Column (1) shows that the coefficient of $\ln Loss\times Cash$ is −2.23 × 10⁻¹², which is significant at a level of 1%. This result shows that holding more cash can help firms alleviate the rising financing costs caused by meteorological disasters. Similarly, Column (2) shows that the coefficient of $\ln Loss\times M{A}_{i,t-1}$ is −9.50 × 10⁻³, which is significant at a level of 5%. This result shows that higher ability managers can also help firms alleviate the rising financing costs caused by meteorological disasters. In sum, the results of the two-way fixed effect models in Table 8 are consistent with our expectations. Therefore, we can conclude that both cash holdings and managerial ability mitigate the impact of meteorological disasters on the equity financing cost.

5. Conclusions

Meteorological disasters have a great impact on society and the economy. However, there is a lack of studies on the impact of meteorological disasters on firms at the micro level. This paper investigates the impact of meteorological disasters on the equity financing cost. Based on the sample data of mining companies listed in China’s A-shares market from 2004 to 2019, we found that meteorological disasters significantly increase the equity financing cost. This result is robust compared to alternative measurements of equity financing cost, such as the two-way fixed effect model, severe meteorological disaster dummy variables, and instrumental variable regression. Further research shows that cash holdings and managerial ability can mitigate the impact of meteorological disasters on the equity financing cost.

Our contribution is to enrich the literature on the adverse effects of meteorological disasters on firms by using the empirical evidence of the Chinese market. Our study provides significant implications for firms and policymakers. First, firms should carefully evaluate the risk of weather-related disasters. Firms must realize that the occurrence of meteorological disasters has affected the external financing costs of enterprises. Second, in the face of the negative impact of meteorological disasters, firms should take active management strategies, such as maintaining sufficient cash holdings and hiring highly competent managers, to mitigate the negative effect of disasters. Third, when firms are hit by disasters, policymakers should provide favorable policies as far as possible to support enterprises to tide over difficulties.

Our research also has limitations. First, our empirical evidence comes from listed companies in the Chinese market. In the future, this study can be expanded to non-listed companies and other markets. In addition, the dynamic and time-varying impact of meteorological disasters on corporate financing costs is also a direction for future research.