In this section, we first test the data’s stationarity by the IPS panel unit root test. Then, both the panel-data model and dynamic threshold model will be employed to examine the correlation and Influence transmission between climate policy uncertainty and company investment.
Panel unit root test and multicollinearity test
Before the regression analysis, the stationarity of the variables needs to be tested. Table 3 demonstrates the p-value and statistic of the IPS test, which indicate that all variables are stationarity. Based on these results, we can then employ the panel-data and dynamic threshold models to examine the linkage and influence transmission between climate policy uncertainty and company investment in Chinese energy-related companies. In addition, we performed a multicollinearity test on our variables based on the least square regression (OLS) through the variance inflation factor (VIF) test. Our test results show no significant multicollinearity among the variables in our model.
Based on the outcomes of the panel unit root test and VIF values, the panel models are appropriate for examining the linkage and influence transmission between climate policy uncertainty and company investment. Following the literature involving companies in multiple industries [43, 68], we also examined whether the effect of climate policy fluctuations on the firm’s financial decisions varies by industry.
Impacts of climate policy uncertainty
This part shows the results of the basic panel model. We analyze all companies in the data set and then divide them into two industries according to the Chinese industry classification standard, testing with the panel regression model, respectively. The dynamic threshold panel model will also be employed in the next part to further investigate the asymmetric effects of climate policy uncertainty on firm investment.
Table 4 demonstrates the empirical results of the dynamic panel estimation, which aims to explore the linkage and influence transmission between climate policy uncertainty and company investment in Chinese energy-related companies. Column 1 reports the overall outcome of policy uncertainty on firm investment. Columns 2 and 3 display the empirical results of the industry I and II, respectively. The results indicate that the negative effect of CPU on investment expenditure is statistically significant at 1% significance level overall and significant at 5% level for industry II. These results show that, on the whole, the increase of uncertainty of climate policy will lead to the decrease of investment expenditure of companies in China’s energy-related industries. This shows our hypothesis H1 overall. Judging from the results of different sub-industries, the negative impact of climate policy changes on the mining industry is much more significant. On the other hand, the effects on the industry I are not statistically significant. This proves the correctness of our hypothesis H3.
There are several possibilities that could make the mining industry more susceptible to disruption. China’s coal mining industry occupies a leading position and is one of the main raw materials for power production. The use of mineral resources such as coal directly increases carbon emissions. Therefore, when climate policy changes, these traditional energy companies are the first to be hit. As mentioned earlier, mining companies may face large amounts of stranded assets (resources already mined or equipment used, etc.), leading to a large amount of impairment of the investment funds of these companies. On the other hand, the unpredictability or suddenness of the direction of climate policy changes may disrupt the plans of companies and investors, making companies unable to put their money into specific projects decisively. They are more likely to sit tight and choose their investment strategies more cautiously.
The industry I is mainly engaged in the production and supply of electricity, oil and gas and water. These companies are closely related to the daily life of the people, and the supply-demand relationship of their products will not change significantly under the condition of steady economic development. Therefore, the change of climate policy will not bring precipitate obstacles to them in a short time. On the other hand, even with transformational climate policy changes, these companies will have more coordination space than mining companies. For example, the current trend of developing clean energy is unstoppable, which will substantially impact the main business of mining companies. But these companies of industry I, such as power supply companies, can choose other energy supply companies.
Numerous previous studies on corporate investment have taken GDP as the control variable of the empirical model and shown that high GDP growth rates lead to increased corporate investment [61, 69]. The effects of GDP are positive for both industry I and industry II but only significant for the former, which indicates that economic growth can increase corporate investment, especially for the industry for the production and supply of electricity, heat, gas, and water. From this perspective, hypothesis H2 is valid for energy-related companies as a whole but not for industry II. This may still be related to China’s environmental protection policy and the trend of low-carbon development. Economic growth will promote the overall progress of the energy industry, activate their investment activities and create better investment channels. However, due to the existence of regulatory pressure caused by environmental problems, mining companies are facing a more severe living environment.
The estimated coefficients of leverage are − 0.1072, − 0.0158, − 0.0987, respectively, for the whole sample, firms from industries I and II. The coefficients are negative and highly significant except for industry I. This result means that an increase in leverage decreases investment expenditure, which is roughly consistent with prior studies . It is because that capital structure plays a vital role in each business process. Based on the agent theory, when executives work for the benefit of shareholders, they tend to give up some investment projects with a positive NPV when the company already has too much debt .
The influence of Tobin’s Q is also different by industry. For the companies from industry II, the empirical results indicate that Tobin’s Q has a statistically significant positive effect on firm investment expenditure. They note that higher Tobin’s Q to some extend captures the future profitability of the company’s existing investment projects. However, the empirical result for industry I is not significant. The growth rate of sales also has a very different impact on industries I and II while the growth of assets does not take effect.
Asymmetric impacts of climate policy uncertainty under different market conditions
The panel model has verified the negative linkage between climate policy uncertainty and corporate investment, but this model could only reveal the linear relationship. This section employs the dynamic threshold panel model to examine nonlinearity and possible asymmetry impacts better, and the results are shown in Table 5. Column 1 reports the overall effect of climate policy uncertainty in two industries. Columns 2 and 3 display the empirical results for the industry I and industry II, respectively.
Compared with the static panel model, the dynamic model has a main character that the lag term of the explained variable is added for analysis. From the empirical results, the past investment of the company (L.INV) has a very significant positive effect on the current investment expenditure. The other coefficients depict that the impact of climate policy uncertainty is negative and statistically significant in the low-level range (below the threshold point). Consistent with the panel regression results (Table 4), this negative influence is only evident for the total sample pool and industry II companies. This shows that hypothesis H1 is valid for both industry I and industry II, and at the same time, it once again confirms the robustness of our hypothesis H3. This reflects one of the contributions of our article, as the previous literature often only verified a single correlation between uncertainty and corporate investment. Our research found the different performance of sub-industry companies in the energy industry when dealing with climate policies. What’s more, the advantages of the threshold model are also reflected because we can see that the entire sample has completely different significant results in the upper and lower threshold ranges.
These results show that the increase in climate policy uncertainty will decrease corporate investment expenditures for the whole piece of Chinese listed companies. Still, when this uncertainty exceeds the threshold, this inhibitory effect disappears instead. In China, most of the listed companies in the energy industry are state-owned enterprises . For Chinese state-owned enterprises, management may be inclined to adopt a more conservative investment strategy when climate policy changes, as they are generally facing less competitive pressure with the support of government forces. For industry II, the upper and lower threshold ranges have significant negative impact coefficients, but it can also be seen that the lower range has a greater impact than the upper range. This may also be because the mining industry will be more directly impacted by climate policy changes, and they’re probably own a large number of sunk costs or stranded assets in the future. The dynamic threshold model reflects the nonlinear and asymmetric relationship between climate policy uncertainty and corporate investment under the condition of controlling the past state of the explained variables and demonstrates the relationship between the two in more detail and comprehensively.
In addition, the dynamic threshold regression results of the industry I are quite different from the panel regression results. The second column of Table 5 shows that climate policy uncertainty has a significant positive effect on the investment of companies in Industry I, and it is more obvious in the low-level range. This suggests that industry I companies will increase their investment when climate policy becomes more uncertain. This result is also logical to some extent. As mentioned in Section 4.2.1, the main business of the companies in Industry I is not directly related to activities such as energy mining, but to produce the corresponding products in the form of purchasing energy. So when climate policy changes, the consumer base of their products will not change significantly, but they may look for alternative investment projects, such as clean energy research as a part of long-term development, and so on.
Moreover, the threshold point for the energy-related companies in the mining industry (2.023386) is the same with the whole sample and is larger than that for industry I (1.92843). These results imply that the companies in the industry I was more responsive to the fluctuations in climate policy uncertainty. This outcome may be own to the profitability of the companies in industry I is impacted by both the demand and supply, while the demand for energy mainly impacts the mining industry.
For the GDP term, the results are consistent with those of the panel model, indicating that economic growth positively affects corporate investment and verifying our hypothesis H3 again. First, this may be explained by the investment opportunities that the company has access to. When the country’s GDP growth rate is high, it can be inferred that the country’s economy is undergoing rapid development, the possibility of enterprises capturing investment opportunities will increase .
In Table 5, the leverage displays significantly negative coefficients of-0.0548, − 0.0379 and − 0.0534 in three columns, thus indicating that the companies with higher debts spend less when investing outside than other companies. The opinions of Jensen , Stulz , and Grossman and Hart  also support that excessive financial leverage will reduce the amount of investment. The estimated coefficients of sale growth are 0.0175, − 0.0612 and 0.0564 for all firms, industries I and II, respectively.
The results of the other control variables are basically consistent with the results of the panel model, and the corresponding coefficients are also significantly different due to the existence of corporate heterogeneity. This shows many differences between the two types of energy industries, and it is worth exploring the different influence mechanisms behind them.