Climate risk and digital transformation of enterprises – evidence from China Open Access

Global temperatures have continued to increase steadily over the past few decades (Chen and Yang, 2019), and the planet is rapidly becoming hotter than ever (Lovelock and Rapley, 2007). Simultaneously, the incidence of extreme cold and warm temperatures has increased (Stott, 2016), making it important to study the economic consequences of temperature volatility.

Substantial research has been conducted on the economic consequences of climate risk. It has been found that high temperatures reduce the supply of labor (Graff Zivin and Neidell, 2014), leading to labor losses (Parsons et al., 2021), reduce the stability of the financial system (Campiglio et al., 2018) and the efficiency of the stock market (Hong et al., 2019) and negatively affect global supply chain networks (Pankratz and Schiller, 2024). However, how climate risk affects the business sector’s behavior remains poorly understood (Addoum et al., 2020). We seek to provide direct evidence of the relationship between climate risk and the digital transformation of enterprises to inform the impact of climate risk on the business sector.

Further research is required to determine whether climate risk has a negative or positive impact on enterprises’ digital transformation. Digital transformation is the use of digital technologies by firms to improve their business, enhance customer experience and streamline operational processes to meet market demands (Fitzgerald et al., 2014). Climate risk can hinder the digital transformation of enterprises in different ways. First, it can reduce firms’ profitability, make it more difficult and costly for them to obtain bank loans, and increase the cost of equity financing, thereby increasing their financial constraints and hindering their digital transformation. Second, climate risk can reduce the level of corporate risk-taking, thus hindering firms’ digital transformation.

Climate risk may also facilitate enterprises’ digital transformation. It can harm employees’ physical and mental health, thereby reducing their productivity. As a result, companies will attempt to mitigate the impact of climate risk by driving digital transformation in a way that reduces the use of human capital.

To test the competing hypotheses above, we use firm-level data to examine how firms are responding to climate risk. Climate risks can be categorized into physical and transitional risks: physical risk is the risk of direct damage to economic activity from climate change, whereas transitional risk is the impact on economic activity from changes in policies, technologies and preferences during the transition to a low-carbon economy (Carney, 2015). It is necessary to understand the impact of physical risk on the business sector (Gasparini and Tufano, 2023). This study focuses on climate physical risks; specifically, we examine the impact of temperature change, which is a physical risk, on enterprises’ digital transformation. It uses day-to-day temperature volatility in the province where the firm is registered to measure temperature changes using daily average temperature data from the China Meteorological Administration (CMA) Meteorological Information Center.

This study adopts textual analysis to construct proxy variables for digital transformation of enterprises and uses 32,192 samples from 4,120 listed companies in China’s A-share market from 2010 to 2022 to study the relationship between climate risk and firms’ digital transformation. The fixed-effects model regression results show that day-to-day temperature volatility has a significant negative effect on the digital transformation of enterprises.

To address the potential endogeneity problem in the baseline regression, we construct instrumental variables based on the forest cover rate of the province where the firm is registered and obtain the same results as in the baseline regression. In addition, we conduct several robustness tests, including changing proxy variables for day-to-day temperature volatility. The results of these robustness tests further support the hypothesis that climate risk hinders firms’ digital transformation.

Next, we explore the mechanisms by which climate risk affects firms’ digital transformation. First, it is demonstrated that climate risk can increase financial constraints, thus hindering firms’ digital transformation. Second, we find that risk-taking is not the mechanism of climate risk affects digital transformation.

We also examine the effects of political connections and institutional investors on the relationship between climate risk and firms’ digital transformation. First, firms with political connections are more likely to obtain low-cost loans and access lower-cost equity capital. Thus, climate risk is more likely to affect firms’ digital transformation when they are not politically connected. Second, institutional investors are important in corporate financial decisions. Similarly, firms with a higher percentage of institutional investor ownership are more likely to obtain bank loans. The results suggest that the negative relationship between climate risk and firms’ digital transformation is more significant for those with lower institutional investor shareholdings.

This study contributes to the literature in three ways. First, it adds to the literature on the economic consequences of climate risk. Climate finance is at the forefront of research in the area of financial risk as climate risk will have an impact on almost all sectors of the economy (Gasparini and Tufano, 2023). A large body of literature has focused on the adverse effects of climate risk on exports, agriculture and the economy (Li et al., 2021; Mendelsohn, 2014; Kotz et al., 2021). Yet firms’ behavior under climate risk shocks has received less attention. This study fills this gap by examining the perspective of firms’ digital transformation.

Second, this study adds to the evolving research on the factors influencing enterprises’ digital transformation. Previous research has attributed it to social factors such as digital universal financial and economic policy uncertainty (Cheng and Masron, 2023; Guo et al., 2023). In contrast, we focus our study on temperature change as climate risk, expanding research on the determinants of firms’ digital transformation from a new perspective.

Third, this study empirically tests potential mechanisms for firms’ digital transformation. Our findings emphasize the substantial role of financial constraints in the impact of climate risk on firms’ digital transformation. They have important implications for policymakers concerned with climate change and corporate digital transformation.

Most relevant to our research is Chen and Zhang (2025), who found that climate risk for heavy polluters in China drives digital transformation of companies. Our study differs from theirs in the following ways: first, their study focuses on heavily polluting firms, while our study examines non-financial listed companies. Second, they use annual report data to construct climate risk indicators, while we use objective day-to-day temperatures to construct climate risk indicators. Finally, we use forest cover rate as an instrumental variable to mitigate endogeneity.

The rest of this study is structured as follows: Section 2 is the literature review; Section 3 presents the hypotheses; Section 4 describes the regression data; Section 5 analyzes the regression results, and Section 6 is the conclusion.

Over the past few years, a growing number of scholars have begun to study climate-related financial issues. However, although it is undeniable that climate risk can have serious long-term impacts on firms’ behavior (Hossain et al., 2023; Painter, 2020), the literature on related topics remains relatively scarce in the field of corporate finance (Gasparini and Tufano, 2023).

First, climate risk can have an impact on firm value (Berkman et al., 2023). Using the Global Climate Risk Index compiled by Germanwatch, Huang et al. (2018) found that a rise in climate risk can cause physical damage to firms’ tangible assets, thereby reducing their value. Extreme heat can lead to a decline in a company’s operating income (Pankratz et al., 2023), and a lack of climate risk disclosure (the KeyedVectors model was used to construct this indicator) can lead to an increase in its stock price crash risk (Lin and Wu, 2023). Moreover, a firm’s bankruptcy risk increases with climate exposure risk (Feng et al., 2022). Global warming can also cause a decline in corporate reputation (Pan et al., 2024).

Therefore, in response to increasingly severe climate risk, companies take various actions to maintain their market value or gain more market share. Drawing from the earnings calls of more than 10,000 companies worldwide, Sautner et al. (2023) used machine learning to construct company-level climate risk exposures and found that companies with higher climate exposure create more green technology-related jobs and generate more green patents. Through a study of companies in different countries, Javadi et al. (2023) and Ma et al. (2024) found that firms would hold more cash in response to costly external financing due to climate risk. Climate change also makes it more difficult and expensive for firms to obtain loans from banks (Huang et al., 2022; Nguyen et al., 2022), leading to reduced and less efficient business investments (Agoraki et al., 2024). In addition, companies at greater risk of sea level rise will invest less in R&D, resulting in fewer patents being granted (Du et al., 2024). Because of climate risk aversion, firms are more cautious and thus reduce M&A activity (Lodh et al., 2024).

Firms can improve their organizational resilience to better anticipate, absorb and adapt to, and recover from negative shocks from physical risks (Linnenluecke and Griffiths, 2010). Corporate social responsibility helps reduce the cost of climate risk for firms (Hossain and Masum, 2022).

Evidence also shows that climate risk may not significantly impact corporate investment behavior. An empirical study by Gu and Hale (2023) found that physical risk does not have a significant impact on foreign direct investment. Deng et al. (2024) found no significant effect of extremely high or low temperatures on firms’ financialization. Through a case study, Weinhofer and Busch (2013) found that executives of firms have limited knowledge of climate change, resulting in these firms’ climate risk management processes not being significantly different from other risk management processes.

Firms’ digital transformation has become a core strategy for many enterprises to improve business efficiency (Luo et al., 2023); thus, it is of great theoretical and practical significance to study the factors influencing firms’ digital transformation. From a macro point of view, digital universal financial and a favorable business environment can promote enterprises’ digital transformation (Guo et al., 2023; Luo et al., 2023), while climate policy uncertainty and economic policy uncertainty will force enterprises to implement digital transformation (Mo and Liu, 2023; Cheng and Masron, 2023). From the perspective of enterprises themselves, the maturity mismatch will exacerbate the enterprises’ financial difficulties and thus hinder their digital transformation (Hu et al., 2023), and digital self-efficacy, the presence of a chief digital officer, and high-quality employees are positively correlated with digital transformation (Eller et al., 2020; Firk et al., 2021; Malodia et al., 2023). In addition, younger CEOs are more inclined to digital transformation than older CEOs (Zou et al., 2024).

The above literature suggests that research on the impact of physical climate risk on firms’ behavior is already underway; however, further research is required. Scholars have examined the factors influencing firms’ digital transformation from both macro- and micro-perspectives, but they have not yet examined the impact of physical climate risk on firms’ digital transformation. In the context of the global climate crisis, studying the impact of physical climate risk on enterprises’ digital transformation is of great theoretical and practical significance.

In the face of climate risk shocks, firms are likely to reduce their digital transformation due to the intensification of financial constraints and reduced levels of risk-taking. However, they may also increase digital transformation to hedge against declines in employee productivity.

On one hand, we propose that climate risk hinders firms’ digital transformation by increasing corporate financial constraints. Corporate finance can be raised either through endogenous or exogenous financing, and studies have found that climate risk negatively affects both types of financing.

Climate change can adversely affect firms’ revenues and profitability, and a decline in profitability can lead to a reduction in the sources of endogenous financing for firms. Somanathan et al. (2021) suggested that high temperatures can make workers less productive as they will be more susceptible to fatigue and cognitive impairment. High temperatures will also affect the machines’ performance and reduce the productivity of capital (Zhang et al., 2018); this will reduce the firm’s total factor productivity and lead to a decline in output (Chen and Yang, 2019). Moreover, when firms face high temperatures, the cost of goods sold, selling expenses and wages increase (Pankratz et al., 2023). As a result, the decline in revenue and increase in expenses of the firm lead to a decline in its net profit. Furthermore, a decline in net profit can lead to a decline in the firm’s ability to raise endogenous finance.

Climate risk can also lead to increased financial constraints by making it more difficult for firms to borrow from banks and increasing interest rates (Huynh et al., 2020; Javadi and Masum, 2021). Bank loans are an important source of financing for Chinese companies. While most bank loans require collateral, real estate is the collateral of choice for many companies. Real estate values are subject to uncertainty due to climate change (Bernstein et al., 2019), and as a result, lenders will charge higher loan rates for loans that face greater climate risk (Nguyen et al., 2022). At the same time, extreme weather can adversely affect a company’s revenue and profitability, which can make banks skeptical of a company’s ability to repay its loans, thus reducing creditors’ approval of loans as well as the amount of the loan (Duan and Li, 2024). Finally, banks themselves are also highly concerned about climate risk, and as a result, they are more cautious about granting loans when faced with climate risk (Zhou et al., 2023).

Climate risk also exposes firms to a higher cost of equity capital. Investors are highly sensitive to climate risk and therefore require a higher risk premium when investing in capital in the face of climate change (Brown et al., 2018). Huynh et al. (2020) found a significant positive effect of drought on the cost of equity capital, with the cost of equity for firms affected by severe drought being 92 basis points higher than that of equity for unaffected firms.

In summary, climate risk can negatively affect firms’ endogenous and exogenous financing, increasing their financial constraints; the greater the financial constraints, the lower the likelihood of realizing the enterprise’s potential investment (Almeida and Campello, 2007; Fazzari et al., 1987). Digital transformation is a major shift in a firm’s own strategy and requires a large investment; thus, an increase in financial constraints due to climate risk reduces a firm’s digital transformation.

Second, climate risk can hinder enterprises’ digital transformation by reducing their level of risk-taking. Previous research has found that the executives of firms that have experienced severe natural disasters are more risk averse (Guiso et al., 2018); therefore, climate risk can change executives’ risk appetite and favor conservative decisions about the firms’ digital transformation. Javadi et al. (2023) found that firms will be forced to hold more precautionary cash to cope with adverse shocks from climate risk. Zhou et al. (2022) found that temperature change can significantly affect the risk-taking level of non-financial listed firms in China; in other words, temperature change significantly discourages firms’ risk-taking behavior. For companies, digital transformation is a major strategic decision with a great deal of uncertainty about whether it will be successful. Therefore, climate risk can hinder firms’ digital transformation by reducing their risk-taking level.

Then, we propose the following hypothesis:

On the other hand, we propose that climate risk promotes firms’ digital transformation. Climate risk can have a negative impact on human capital. First, climate change can take a physical toll on employees, leading to a decrease in their willingness to work and productivity (Graff Zivin and Neidell, 2014; Groen et al., 2020). Ebi et al. (2021) found that high temperatures can negatively affect physical labor capacity and athleticism, increase occupational health-related risks, and lead to a decrease in employee productivity. Second, climate change can damage employees’ mental health, causing negative emotions such as anxiety and irritability, and weakening employees’ work enthusiasm and motivation. Wang et al. (2014) found that climate risk increased the incidence of mental illness in Toronto, Canada.

The International Energy Agency (2018) found that when faced with extreme heat, firms can adapt to the heat by installing air conditioning, but this is costly. Firms’ digital transformation is the use of machines or computers to replace a portion of human labor (Brynjolfsson and McAfee, 2014); in other words, companies can use the increased level of digital transformation to reduce the cost of labor (Verhoef et al., 2021). Thus, in the face of climate risks, firms have an incentive to increase enterprise digital transformation to reduce the impact of climate shocks on productivity.

Based on the above analysis, we propose the competing hypothesis:

The data used in this research include China’s climate-risk data and the financial data of A-share listed companies. Among them, the climate-risk data are obtained from the CMA Meteorological Information Center, and the macro data of the regional level and the financial data of listed companies are from the CSMAR database. The sample interval of this study is 2010–2022. We also exclude samples with missing indicators, samples of listed companies that are insolvent, samples of listed companies in the financial industry and samples of listed companies in ST and *ST. To avoid the impact of extreme values on the findings of this study, we also winsorize all continuous variables by 1% up and down. The final sample contains 32,192 firm-year observations for 4,120 listed companies.

Intra-monthly standard deviations of daily temperature anomalies provide a better measure of climate change than absolute changes in temperature (Moberg et al., 2000). Two regions with the same mean annual temperature have large differences in daily temperature change (Kotz et al., 2021). Therefore, referring to Kotz et al. (2021), we use day-to-day temperature volatility as a proxy variable for climate risk. This approach eliminates the effect of seasonal temperature cycles on the regression results. The specific measurements are as follows:

In equation (1), $CRp,y$ is our independent variable, the day-to-day temperature variability in a province(p) for a given year (y), where p is the province where the firm is registered. Higher values of $CRp,y$ mean higher climate risk. $Tx,d,y$ is the daily(d) temperatures for a given year(y) and city (x).$T¯x,m,y$ is the average temperature of the city for a given year(y) and month (m). $Dm,y$ denotes the number of days in each month(m) of a given year(y). $Np$ is the number of cities governed by each province(p). $wp,x$ is the ratio of the administrative area of the city(x) to that of the province(y) in which the city is located. In the robustness test, we will also use the ratio of the number of city population to that in the province where the city is located to calculate the day-to-day temperature variability.

Referring to Wu et al. (2021) and Zhou and Li (2023), we adopt text analysis to construct firms’ digital transformation indicators in the following way: first, we construct a library of statements including digital transformation keywords; second, we use the Python tool to crawl the data of annual reports of listed companies, extract all text contents, and match the text contents with digital transformation keywords. Again, we clean the acquired content to eliminate the text with negative words before the keywords and the text that is not of the company. Finally, we count the number of occurrences of digital transformation keywords in the annual report of each listed company per year. It is reasonable to use the word frequency of digital transformation in the annual reports of listed companies to measure the digital transformation of enterprises because annual reports have high credibility (Unerman, 2000). And digital transformation, as a strategic behavior of enterprises, will be reflected in annual reports (Huang et al., 2023). We measure enterprises’ digital transformation in two ways, $Digital1$ is the degree of digital transformation, measured using the natural logarithm of the number of times of digital transformation keywords plus 1; the larger the value, the higher the degree of digital transformation of the enterprise. $Digital2$ is the dummy variable of enterprises’ digital transformation, which takes the value of 1 when the number of times of firms’ digital transformation in the year is greater than 0, and 0 otherwise.

According to the previous theoretical analysis, climate risks may either drive or hinder enterprises’ digital transformation. To test the above hypothesis, we take day-to-day temperature variability as an independent variable and the degree of enterprises’ digital transformation as a dependent variable and construct the following model:

In equation (2), we use the independent variable $CRp,t$ as a proxy for climate risk and the explanatory variable $Digitali,t+1$ as a proxy for firms’ digital transformation. In this study, the degree of digital transformation (⁠$Digital1i,t+1$⁠) and whether digital transformation is implemented (⁠$Digital2i,t+1$⁠) are used to measure the level of enterprises’ digital transformation. Referring to the study of Porfírio et al. (2021), Zhang et al. (2023) and Nie et al. (2024), we use variables that significantly affect enterprises’ digital transformation as control variables: $sizei,t$ is firm size, measured as the natural logarithm of the firm’s total assets; $leveragei,t$ is firm leverage, measured as the ratio of total liabilities to total assets; $roei,t$ is the return on net assets, measured as the ratio of net income to net assets; $firsti,t$ is the ratio of the largest shareholder’s shareholding, measured as the ratio of the number of shares held by the largest shareholder to the total number of shares in the listed company; $boardi,t$ is the independent director ratio, measured as the ratio of the number of independent directors to the total number of directors on the board; $agei,t$ is the age of listing; $growthi,t$ is the growth rate, measured by the growth rate of the current year’s sales revenue over the previous year. $dualityi,t$ is a dummy variable for duality, taking the value of 1 if the chairman and the CEO are the same person, and 0 otherwise. $soei,t$ is a dummy variable for the nature of the firm, which takes the value of 1 when the nature of the firm is state-owned, and 0 otherwise. $GDPp,t$ is the growth rate of GDP per capita in province p. We also include year- and industry-fixed effects to control for un-observables that vary over time and with industry. $εi,t+1$ is the random error term. $α1$ is the coefficient of the independent variable, which is the most important concern of this study. If $α1$ is greater than 0, it implies that climate risk can facilitate firms’ digital transformation; reproduction, if $α1$is less than 0, implies that climate risk can hinder firms’ digital transformation. We use the lead (t + 1) explanatory variables in the baseline regression to alleviate the endogeneity problem and because firms’ digital transformation is a major decision, and it takes a longer time for managers to formulate a strategy and implement.

Table 1 reports descriptive statistics for the main variables of the baseline regression in this study. The mean value of $CRp,t$ is 2.767, and the maximum value is 4.554, while the minimum value is 1.414, indicating a large gap in climate risk among Chinese provinces. The maximum value of $Digital1i,t+1$ is 6.301, and the minimum value is 0, which indicates a large gap in the level of digital transformation of listed companies. The mean value of $Digital2i,t+1$ is 0.662, which implies that 62.2% of listed companies have undergone digital transformation during the sample period. The distribution of control variables is basically consistent with the existing literature (Shang et al., 2023; Zou et al., 2024).

Table 2 reports the baseline regression results of this study, with the dependent variable in Columns 1 and 2 is the level of firms’ digital transformation, and the dependent variable in Columns 3 and 4 is the dummy variable of whether firms have undergone digital transformation, all controlling for industry- and year-fixed effects. The results in Columns 1 and 3 show that climate risk and firms’ digital transformation show a significant negative correlation. After adding the control variables, the results show that the coefficient on $CRp,t$ in column 2 is −0.120 and significant at the 1% level, meaning that for every unit change in day-to-day temperature volatility, a firm’s digital transformation for the following year will be reduced by 12%. The coefficient on $CRp,t$ in Column 4 is −0.032 significant at the 1% level. The results of these four columns imply that climate change hinders firms’ digital transformation, which supports H1a.

In addition, the results of the control variables are consistent with those of previous research. The coefficient of firm size is significantly positive, which indicates that larger firms are more likely to undergo digital transformation. The coefficient of $leveragei,t$ is negative and significant at the 1% level, which implies that the higher the firm’s gearing ratio, the lower the digital transformation. The coefficient of $roei,t$ is significantly positive, which indicates that firms with higher ROE are more likely to undergo digital transformation. The coefficient of $soei,t$ is significantly negative, suggesting that non-state-owned firms are more inclined to undergo digital transformation compared to state-owned firms. Overall, the results in Table 2 show that after controlling for other factors, listed firms’ digital transformation decreases when the climate risk in the province where they are registered becomes higher.

This study may present an endogeneity problem, such as omitted variables that have an effect on both dependent and independent variables; therefore, we use instrumental variables to mitigate it. Specifically, we use the forest cover rate in each province during the sample period as an instrumental variable. Forests mitigate climate change (Bonan, 2008) and buffer against temperature extremes (De Frenne et al., 2021). Therefore, the instrumental variable of forest cover in each province fulfill the correlation condition. Forest cover rate does not directly affect a firm’s digital transformation, thus satisfying the exogeneity condition.

Table 3 presents the regression results. Column 1 presents the results of the first-stage regression, where the coefficients of the instrumental variable are significantly negative at the 1% level, in line with theoretical expectations. Columns 2 and 3 show the results of the second-stage regression, in which the regression coefficients for climate change are significantly negative at the 1% level. The regression results using the instrumental variables are consistent with the baseline regression, indicating that the findings of this study are robust. An F-value of 3,066.31, which is greater than 10, indicates that the weak instrumental variables in this study are less problematic.

We use the propensity score matching (PSM) method to address sample selection bias. First, based on the mean value of climate risk, samples above the mean are categorized into treatment groups and samples below the mean are categorized into control groups. Second, the control variable is selected as the matching variable and the nearest neighbor matching method is used to match the observations in the treatment and control groups. The regression results in Table 4 indicate that climate risk is still significantly and negatively correlated with firms’ digital transformation.

We conduct other robustness tests using alternative measures of climate risk, with different subsamples and exclude the impact of transitional climate risks.

First, in the baseline regression, we use the administrative areas of cities to construct a proxy variable for climate risk. Here, we use city population to construct a proxy variable for climate risk in the regression, and the results in Columns 1 and 2 of Table 5 show that the results remain robust after the proxies for the independent variables are replaced.

Second, we exclude the impact of COVID-19. At the end of year 2019, an outbreak of COVID-19 occurred in China, and subsequently, employees in many organizations could only work remotely, which indirectly had an impact on the organizations’ digital transformation. Therefore, to exclude the impact of COVID-19, we conduct a regression using a subsample of 2010–2019, and the results are presented in Columns 3 and 4 of Table 5; they show that the coefficient of climate risk is still significantly negative at 1% level, which may indicate the robustness of the benchmark regression results.

Third, we exclude the effect of climate transition risk on the regression results. Climate risk can be categorized into physical and transitional risk, and in this research, we focus on physical risk. But in regions with higher climate risk, governments may implement more climate policies and market demand may change, exposing companies to higher transition risks (Li et al., 2025). In this case, the digital transformation behavior of the enterprise may be attributed to transformation risk rather than physical risk. By referencing Mo and Liu’s (2023) transformational risk using climate-policy uncertainty, we include both transitional risk (CCPU) and physical risk (⁠$CRp,t$⁠) in the regression, and the regression results are displayed in Columns 5 and 6 of Table 5. We find that the effect of climate risk hindering firms’ digital transformation is independent of transitional risk, which again supports the robustness of the baseline regression results.

Through empirical analysis and robustness tests, in the previous section, we demonstrated that climate risk hinders firms’ digital transformation. On this basis, we further examine the economic mechanism of climate risk hindering enterprises’ digital transformation and intend to validate it in terms of financial constraints and enterprise risk-taking.

According to the hypotheses analysis, climate risk reduces corporate earnings, increases the difficulty and cost of corporate borrowing and increases the cost of equity financing, which leads to greater corporate capital constraints and hinders corporations’ digital transformation. Therefore, we first test the mediating effect of financial constraints.

Referring to Baker et al. (2003) and Meng et al. (2020), we use KZ index as a proxy variable for financial constraints because the KZ index is more sensitive to firms’ short-term financing pressures (Lamont et al., 2001). KZ index is calculated as follows:

In equation (3), CF is free cash flow; DIV is the standard deviation of total assets; CASH is cash, and LEV is the ratio of liabilities to assets. Larger values of KZ imply a higher degree of financial constraints. Table 6 presents the regression results.

The independent variable in column 1 of Table 6 is climate change, and the dependent variable is financial constraints. The coefficient of climate change is 0.116 and significant at the 1% level, which indicates that climate risk significantly increases firms’ financial constraints. The dependent variable in Column 2 is the level of firms’ digital transformation, and the dependent variable in Column 3 is whether firms have undergone digital transformation. The coefficients of climate risk and financial constraints are significant in both Columns 2 and 3, which implies that financial constraints bear part of the mediating effect; thus, we argue that financial constraints are a mechanism by which climate risk hinders firms’ digital transformation.

Following Li et al. (2013), we use return on assets (ROA) volatility (Risk-taking) as proxy variables to explore the risk-taking mechanism of climate risk impeding firms’ digital transformation. Risk-taking is measured in two ways:

where $Adj_Roa$ is industry-adjusted ROA to mitigate industry and cyclical effects and is calculated using the following formula:

Due to the large number of manufacturing firms in China, manufacturing firms are subdivided into two-digit codes, and the sample where the industry contains only one firm is deleted. The larger $Risk-taking1i,t$ and $Risk-taking2i,t$ means a higher level of enterprise risk-taking. Table 7 presents the regression results. The dependent variable in Column 1 is $Risk-taking1i,t$⁠, and the dependent variable in Column 2 is $Risk-taking2i,t$⁠. The regression results show that climate risk can significantly reduce enterprise risk-taking, which suggests that climate risk hinders the risk-taking mechanism of enterprises’ digital transformation. However, from the results in Columns (3) to (6), it can be found that there is no significant correlation between corporate risk-taking and firms’ digital transformation. Therefore, we conclude that corporate risk-taking is not a mediator of the impact of climate risk on enterprises’ digital transformation.

The previous section has argued that climate risk can exacerbate financial constraints and thus hinder corporations’ digital transformation. Political connections alleviate corporate financial constraints. Politically connected firms are more likely than nonpolitically connected firms to obtain loans from banks (Khwaja and Mian, 2005), enjoy a lower cost of equity capital (Boubakri et al., 2012) and be bailed out by governments in case of financial distress (Faccio et al., 2006). Thus, we argue that the negative effects of climate change on digital transformation are more pronounced among firms that are not politically connected.

Referring to Boubakri et al. (2012), we construct a dummy variable for firms’ political connection (PC), where we consider a firm to be politically affiliated if the chairman or CEO of the firm has governmental work experience, assigning a value of 1, and 0 otherwise. We conduct subgroup regressions, and the results are presented in Table 8. Columns 1 and 2 are the regressions for the sub-sample of firms that are politically connected; the results show that, in the sub-sample of firms with political connection, no significant correlation exists between climate risk and firms’ digital transformation. Columns 3 and 4 present the results of the regression for the subsample without political connection, which shows that climate risk and business digital transformation are significantly and negatively correlated.

Institutional investors play an active role in corporate governance (Lewellen and Lewellen, 2022). It has been found that institutional investor shareholding can similarly reduce the cost of corporate loans (Roberts and Yuan, 2010) and the cost of equity (Attig et al., 2013), thereby easing corporate financial constraints. Therefore, we argue that the negative relationship between climate risk and digital transformation is more pronounced in firms with lower institutional investor ownership relative to firms with higher institutional investor ownership.

We construct the institutional investor dummy variable (IS), which is assigned a value of 1 when the listed company’s institutional investor shareholding is greater than the average level of institutional investor shareholding in the same industry, in the same year and in the same province, and 0 otherwise. Columns 1 and 2 of Table 9 report the results of the regression on the impact of climate risk on firms’ digital transformation when institutional investor holdings are relatively high. Columns 3 and 4 report the results of regressing climate risk on firms’ digital transformation when institutional investor shareholding is relatively low. It is found that climate risk has a significant negative effect on firms’ digital transformation in the case of relatively low institutional investor ownership compared with the results in the case of relatively high institutional investor ownership.

This study analyzed the impact of climate risk on corporate behavior. Using a sample of listed companies in China from 2010 to 2022, we examined whether day-to-day temperature volatility hinders or promote firms’ digital transformation. The empirical results show that climate risk hinders firms’ digital transformation, and the conclusion remains robust after a series of endogeneity and robustness tests. In the mechanism test, the mechanisms through which climate change hinders digital transformation is increased financial constraints. Therefore, the negative effect of climate risk on firms’ digital transformation is not significant for politically connected firms and higher institutional investors shareholding firms with lesser financial constraints.

This study not only enriches the literature on climate economics and enterprises’ digital transformation but also provides new insights for governments to promote strategic innovation in firms. Enterprises’ digital transformation has a resource effect under climate risk. This implies that firms with higher financial constraints under climate risk struggle more to achieve digital transformation as climate risk increases their financing difficulty.

Our findings can provide a theoretical basis for policymakers. On the one hand, the government can alleviate corporate financing constraints by means of financial subsidies or encouraging financial institutions to provide more special loans for digital transformation, so as to reduce the negative impact of climate risk on corporate digital transformation. On the other hand, regulators can encourage firms to introduce more institutional investors to alleviate information asymmetry in the credit market. There are still shortcomings in our study, for example, we did not use spatial measurement to conduct empirical tests, and we will conduct in-depth research on this in the future.