How do climate policy uncertainty and renewable energy and clean technology stock prices co-move? evidence from Canada

In the wake of global warming, global leaders, policymakers, and corporate executives have altered their focus by adapting climate change policies and using preferably renewable energy (RE) sources (Wu et al. 2023; Dam et al. 2023; Bouri et al. 2023), resulting in a huge increase in RE investment from less than $50B in 2004 to around $300B in recent years (Bloomberg New Energy Finance 2019). RE sources are classified as clean energy sources and usually come from natural resources that can often be replenished, and most of them, such as wind and solar, are ecologically friendly and have a positive effect on environmental quality (Chien et al. 2022). Instead, the use of traditional energy sources causes environmental degradation by releasing carbon dioxide, which eventually raises global temperatures.

As a result of this change in the global business environment, green bonds and stocks have experienced exponential growth, starting from $4.2B to $258.9B globally between 2012 and 2019 (Climate Bonds Initiative, 2019).¹ Generally, climate change policies cover the policies aimed at reducing costs and also policies aimed at accelerating research and development investment in energy production technologies or advances in energy efficiency. However, there can be significant uncertainty regarding the accomplishment of climate change strategies in practice.

Furthermore, the transition to using RE sources will need an enormous rise in investments and the availability of funds. As discussed by the International Energy Agency (2018), the global annual investment in RE will need to rise by 97% over current investment levels to ensure environmental sustainability. Given the large financial support needed for investment in RE, having a developed and efficient financial sector could help allocate the necessary funds for investment, especially in RE, at lower financing costs. Policy and other support mechanisms continue to play a significant role in supporting the yields and mitigating risks of RE and clean technology companies.

Due to the importance of using RE toward achieving environmental sustainability goals and diminishing carbon emissions, several empirical works have systematically scrutinized the impact of renewable energy consumption (REC) on the quality of the environment (e.g., Shu et al. 2023). Prior works (e.g., Godil et al. 2021; Chien et al. 2022) revealed the beneficial role of RE in diminishing carbon emissions. Numerous research has mainly focused on exploring the significant determinates that cause increases or decreases in REC in different advanced and emerging countries. Prior studies (e.g., Sadorsky 2009a; Apergis and Payne 2014a) underscored that income, real GDP, carbon dioxide, and trade openness are the significant drivers of REC. Additionally, several works (e.g., Apergis and Payne 2014b; Mukhtarov et al. 2022) showed that a long-run cointegrated nexus exists between real GDP, real coal and oil prices, carbon emissions, and REC. Additionally, several studies (e.g., Mehrara et al. 2015; Athari 2024a) indicated that FDI, financial market development, and institutional quality are other important determinants that impact RE demand.

While extensive studies are examining the important factors of REC, some empirical research has also been accomplished to probe the effect of uncertainties on REC. For instance, prior research (e.g., Shafiullah et al. 2021; Pham and Nguyen 2022) indicated that REC is impacted by economic policy uncertainty (EPU). Their finding also uncovered a negative nexus between EPU and REC, indicating that a higher EPU resulted in a lower REC. Furthermore, several studies (e.g., Shang et al. 2022) stressed that the changes in the Climate Policy Uncertainty Index (CPU) also cause altering the REC, which eventually impacts the efficiency and yields of RE firms (e.g., Liang et al. 2022).

Since there is a linkage between CPU and REC, what is the dynamic co-movement between CPU and stock prices of renewable energy and clean technology companies, especially during the COVID-19 episode? Unlike prior studies, no comprehensive research has been found examining this nexus in particular from both time and frequency perspectives. Since the energy market is linked to economic development and using RE could be a replacement for traditional sources and also contribute to achieving environmental sustainability, it is essential to understand how particularly CPU could impact renewable energy and clean technology firms’ performance and, eventually, their stock prices. Overall, the energy market, theoretically, is more sensitive to CPU than other kinds of uncertainty. Despite other industries, energy companies encounter higher risks due to their large scale and long payback period (Wang and Kong 2022). Furthermore, RE companies are relatively more exposed to CPU than other uncertainty indexes, and changes in CPU lead to impacting RE market returns significantly (Xu et al. 2022).

Besides, few attempts were made to explore the dynamic co-movement between CPU and stock prices of renewable energy and clean technology firms in the context of the developed country of Canada. In 2021, Canada will be a world leader in RE production, and RE sources will constitute nearly 30% of Canada’s energy supply.² Likewise, Canada decided to invest $964M in clean energy technologies in 2021, aiming to lower greenhouse gas emissions and help Canada attain ecological sustainability by reaching net-zero emissions by 2050. In Canada, the share of primary energy from solar, wind, and hydroelectric power was 0.35%, 2.37%, and 25.74% in 2021, respectively. In terms of global ranking, Canada also ranks 3rd in the world based on hydroelectricity production, generating 9% of the world’s hydroelectricity. Moreover, Canada ranks 8th in wind energy production. Consequently, the firms started to invest in RE projects and the number of renewable energy and clean technology companies has increased since the last decade. Based on this knowledge, Canada could be an interesting country to conduct the effect of CPU on the stock prices of renewable energy and clean technology firms.

Hence, the present study aims to shed light and contribute by answering the subsequent questions: i) How do the CPU and renewable energy and clean technology firms’ stock prices dynamically co-move in the short medium and long term? ii) Does COVID-19 impact this co-movement? To achieve these objectives, this study uses novel data for the CPU and renewable energy and clean technology firms’ stock prices during the wide range of 2013–2022 by using monthly data. This work used the Renewable Energy and Clean Technology Index (RECT) which was launched and valued in 2010 and gauges the performance of firms listed on the Toronto Stock Exchange (TSX) whose main business is the improvement of green technologies and sustainable infrastructure solutions. In the meantime, we use the advanced WPS and WC methodologies to observe the dynamic co-movement, which is rarely performed in this field and enables us to observe the causality direction from both time and frequency aspects. In the meantime, it allows us to uncover the causal linkage between the interested time series variables in the short-medium and long-term horizons. To the best of our knowledge, this may be an original work and differs from prior research in numerous aspects. First, the previous works (e.g., Hsiao et al. 2019; Zhao 2020) focused on testing the impacts of oil price shocks and climate change events on the stock price of RE companies. Second, the previous works (e.g., Liang et al. 2022; Sarker et al. 2022) focused on testing the link between CPU and RE volatility forecasting and also probing the asymmetric effects of CPU on the realized volatility of the returns of RE prices. Third, the prior work (e.g., Shang et al. 2022) focused on testing the impact of CPU on renewable and nonrenewable energy demand.

This research yields some consistently notable highlights. First, the results show that CPU seems only volatile in 2019 and 2021 in the short run. Second, the findings uncovered that there was significant volatility in the RECT in the short and long terms (SLT) between 2018 and 2022. Third, the results show that RECT significantly caused the CPU between 2014 and 2018; however, after 2019, the CPU started to cause RECT in the short and medium terms (SMT). The findings implied that due to the oil price shocks between 2014 and 2018, RECT causes CPU through rising transition risk by deferring the implementation of climate change alleviation policies and changing energy demand globally. Nevertheless, after 2019, CPU started to cause RECT, and the performance of green technology firms significantly reacted to high volatility in CPU.

The rest of this work proceeds as follows. Section 2 shows the literature review. Section 3 discusses the data and methodology. Section 4 discusses the results. Section 5 concludes the paper.

Energy has become an interesting field of study and triggers scholars to explore the important determinants of energy consumption. For example, Bartleet and Gounder (2010) and Shahbaz and Lean (2012) found that there is a co-integration nexus between economic development, employment, industrialization, urbanization, and energy consumption. Shahbaz and Lean (2012) also found that both industrialization and urbanization raise energy consumption in the long run. In a global research, Wang et al. (2019) revealed a long-term co-integration nexus between energy prices, urbanization, economic development, and energy consumption. Besides, Yao et al. (2019) showed that human capital lowers energy consumption by 15.36% in OECD countries. Also, Chiu and Lee (2020) documented that capital development impacts energy consumption, though the extent of the impact of development by the banking sector is relatively higher than the stock market development. Also, the authors highlighted the role of country risk, implying that financial development could help decline energy consumption when an environment is less vulnerable to country risk. With increasing global worry about the unfavorable impacts of climate change and the worldwide economic costs of environmental pollution, environmentally friendly energy sources like RE have become more essential and led to policymakers adopting new policies and strategies for raising the share of REs in supplying energy.

This work collected the data for CPU from the policy uncertainty³ website. The CPU gauges uncertainty associated with climate policy, based on news about uncertainty, climate risk, greenhouse, global warming, green energy, and regulation terms from eight leading US newspapers. Also, the data for RECT were obtained from Blomberg.⁴ The RECT was valued in 2010 and computes the performance of listed companies on the Toronto Stock Exchange (TSX) whose main business is the improvement of green technologies and sustainable infrastructure solutions. Notably, the period of the study was selected based on the matching and accessibility of both time series variables between 2013M01 and 2022M08. Table 1 reveals the definitions and descriptive summary of the variables. As presented in Table 1, the mean of CPU and RECT is 152.248 and 146.754, respectively. Furthermore, it shows the CPU with a standard deviation of 75.426 is more volatile than RECT by a value of 41.326. The kurtosis test also reveals that the distribution of CPU and RECT variables has small tails. As shown in Table 1, CPU and RECT variables are also positively skewed. Moreover, based on Table 1 results, the null hypothesis that CPU and RECT variables have a normal distribution can be rejected, denoting that the errors of the CPU and RECT variables are not normally distributed.

Furthermore, the pattern of the CPU and RECT is demonstrated in Fig. 1. As shown in Fig. 1, there is a positive co-movement between CPU and RECT during the period of the study, though CPU is more volatile. This implies that by rising CPU, RECT also rises and vice versa.

Full size image

In this work, the wavelet method is used to scrutinize the time–frequency co-movement of CPU and RECT. The test is initially developed by Goupillaud et al. (1984). One of the advantages of the approach is that the test can be applied to the nonstationary dataset without converting them to the stationary series (Ourir et al. 2023). Since time series variables in economics and finance are mostly not stationary, traditional time domain causality estimates are inaccurate. Another advantage is that financial and economic time series variables are well known to experience structural breaks. Obtaining breaks in the dataset makes the fixed parameters suffer in the traditional causality approaches (Tao et al. 2023). This situation makes the wavelet approach superior to other traditional approaches as well. Contrary to causality from time dimension, “the key problem with a standalone frequency domain approach, more specifically referred to as the Fourier transform, is that by focusing solely on the frequency domain, the information from the time domain is completely omitted” (Pal and Mitra 2017, p. 231). Therefore, a WC approach, which belongs to the Morlet family, is used in the present study to avoid these problems in our estimations. The WC method provides a deeper understanding of the underlying dynamics by capturing nonlinear relationships between signals that traditional linear correlation methods might miss. Moreover, through intuitive and detailed visual representations, WC shows how two signals relate over different time scales.

In this work, a wavelet (\(\psi \)) is based model that is part of the Morlet wavelet family. The equation is as follows: \(\psi \left(t\right)= {\pi }^{-\frac{1}{4}}{e}^{-i{\omega }_{0}t}{e}^{-\frac{1}{2}{t}^{2}}\), p(t), t = 1, 2, 3, T.

Wavelets have two parameters: time (k) and frequency (f). In wavelet theory, the k parameter controls the wavelet’s location in time by exchanging it, while the f parameter controls the distended wavelet to localize various frequencies. \(\psi \) _k,f can initially be created by transforming \(\psi \). In this case, the equation is as follows:

From time series data p(t), the continuous wavelet can be generated as a function of k and f::

In equation three, the regenerated initial times series p(t) is given as follows:

The WPS provides more information about the amplitude of time series. By using the WPS, it is possible to capture both short-duration high-frequency events and long-duration low-frequency events simultaneously. Moreover, the WPS depicts the signal’s characteristics over time. In addition to this, with the WPS, it is easier to detect significant patterns and features in noisy data

In this work, the WC method is applied. The main novelty of WC over the standard correlation is that the method allows the current work to capture any correlation between two-time series p(t) and q(t) in combined time–frequency-based causalities. “To assess the extent to which two-moment series move in sync, the wavelet coherence method is in place. As indicated in Torrence and Compo (1998), let \({W}_{p}(k, f)\) and \({W}_{q}(k, f)\) be the CWT of p(t) and q(t), respectively, where p(t) and q(t) are the respective time series of two moments with the same order” (Ahmed 2022). The cross-wavelet transform (CWT) of the time series is as follows:where \({W}_{q}^{*}\left(k,f\right)\) represents the complex conjugation of q(t) with respect to the CWT. “The cross-WPS, which detects regions in the time–frequency sphere where p(t) and q(t) show high levels of common power (correspondence),” is the modulus of CWT (Ahmed 2022) as follows:

A wavelet squared coherence is made up of the smoothed cross-wavelet power spectrums of p(t) and q(t), normalized by their smoothed WPSs, as follows:

Time and the smoothness of the process over time are denoted by C, with 0 ≤ R²(k,f) ≤ 1. “Whenever R²(k,f) gets close to 1 it indicates that the time series variables are correlated at a particular scale, surrounded by a black line and depicted by a red color. On the other hand, when the value of R²(k,f) approaches 0 it indicates that there is no correlation between the time series variables and is pictured by a blue color.”

Nevertheless, calculating R²(k,f) does not provide any means of distinguishing between positive and negative correlations; thus “Torrence and Compo (1998) postulated a means by which to detect the WC differences through indications of deferrals in the wavering of two-time series” (Pal and Mitra, 2017). In order to construct the equation of the WC difference phase, we use the following formula:

The outcomes of the WPS and WC are generated using MATLAB R2015a software.

To capture the linkage between CPU and RECT, the present study used WPS and WC techniques which belong to the Morlet wavelet family. Using the WPS in the current research allowed us to capture the behavior of the CPU and RECT between 2013M01 and 2022M08. The results of the WPS for the CPU and RECT are reported in Figs. 2 and 3, respectively.

Full size image

Full size image

To identify the vulnerability and behavior of CPU and RECT in Canada, the WPS test is used as an initial test. The outcomes from this test are reported in Figs. 2 and 3. In these figures, this back cone-shaped curve illustrates the cone of influence below an edge where the wavelet power is affected. In these figures, the horizontal axis represents time, while the vertical axis represents frequency. The bands range from low frequency (16–32 months of scale) to high frequency (0–4 months of scale). The intensity of the spectra is color-coded (blue to yellow color; low to high intensity) (Karamti and Belhassine 2022; Ma et al. 2023). Moreover, based on Monte Carlo simulations, the thick black shape indicates a 5% significance level. As shown in Fig. 2, CPU was volatile in 2019 and 2021 in the short term (ranging from 0 to 8 periods of scale) (high frequency). This volatility happened on these dates due to events such as the UN climate action summit in 2019, Trump’s rejection of the new emission rule in 2020, and also the announcement of new greenhouse gas (GHG) emission standards by the Environmental Protection Agency (EPA) in 2021.

In addition, there was significant volatility in the RECT in SLT between 2018 and 2022 in Fig. 3 at different scales, ranging from 4 to 32 periods of scales. This volatility occurred on these dates because of policy uncertainty and oil price shocks. Remarkably, in 2018, Brent oil declined quickly to $58.71, which was the greatest 30-day decline since 2008. On the other hand, the market experienced another shock by rapidly increasing oil prices in 2022.

As prior studies highlighted, oil price shocks significantly impact the RE sector. For instance, Shah et al. (2018) found that there is a strong relationship between the oil and RE sectors in Norway and the USA. Hsiao et al. (2019) revealed that the global oil price has a significant price spillover impact on the stock prices of RE-listed companies in China. The authors also revealed that the variations in global oil prices affect the stock price volatility of RE companies. Zhao (2020) found that oil supply and demand shocks have a positive and negative effect on the returns of clean energy firms, respectively.

Moreover, as a main part of the analysis, the present study employed WC to capture the time and frequency co-movement between CPU and RECT. Notably, in WC analysis, the cold areas away from the significant region imply no interrelation between the factors from both the time and frequency dimensions. However, the state of an arrow in the graphical plots shows the lag and lead phase association between the studied factors. There is a significant negative correlation between variables when the arrows point left and a positive correlation when the arrows point right. In addition, the arrows pointing down, right-down, and left-up imply that the CPU variable causes the RECT variable. Alternatively, the arrows pointing up, left-down, and right-up indicate that the RECT factor causes the CPU variable.

Figure 4 reports the co-movement between CPU and RECT. As shown in Fig. 4, the RECT caused the CPU between 2014 and 2018 at medium frequency (8–16 months of scale), implying that during that period, RECT was an important predictor for the movement of the CPU. In the mid of 2014, there was a sharp plunge in import crude petroleum prices to $44.08 after peaking at $107.95 a barrel on June 20, 2014. This shock triggers a rising global traditional energy demand, which adversely impacts the performance of green technological companies, decelerates R&D investment in green energy production technologies, and, ultimately, causes rising CPU. Likewise, as discussed above, another oil price shock in 2018 drove CPU through rising transition risk by deferring the completion of climate change alleviation policies and changing energy demand globally.

Full size image

However, after 2019, CPU started to cause RECT in the SMT at medium frequency (8–16 months of scale), suggesting that during the COVID-19 period, the direction of the causality changed. As mentioned above, there is high volatility in CPU after 2019, which significantly impacts the performance of clean and RE companies in Canada. Consistently, Pommeret and Schubert (2018) revealed that a rise in CPU leads to an increase in clean capital investment by firms. Antoniuk and Leirvik (2021) also uncovered that events linked to CPU significantly influence firms’ stock returns. Besides, Bouri et al. (2022) showed that CPU is a significant predictor of the performance of green energy stocks. Likewise, Sarker et al. (2022) revealed that a rise and decline in CPU impact clean energy prices’ returns in the long term.

The present study scrutinizes the time and frequency of co-movement between the CPU and RECT. To attain this purpose, this research performs the WPS and WC approaches by using monthly data between 2013 and 2022. The findings of the study reveal that: (i) CPU seems only volatile in 2019 and 2021 in the short run; (ii) there was significant volatility in the RECT in the SLT between 2018 and 2022; (iii) RECT significantly caused the CPU between 2014 and 2018; (iv) after 2019, CPU started to cause RECT in the SMT, implying that during the COVID-19 period, the direction of the causality changed.

The results have important policy implications. First, the findings suggest that oil price volatility is a significant driver impacting traditional global energy demands and the performance of green energy and technology companies. Hence, to control the climate policy risk, the policymakers and governments should be provided strategies and also design specific plans to enable the respected companies to hedge against oil price shocks and also facilitate availably for external financing and reforming tax policies for private and public companies in enhancing the RE schemes and increasing consumption levels. Second, the results recommend that policymakers, governments, and regulatory bodies should alleviate the climate policy uncertainty to reduce the spillover effect of climate risk on the performance of green energy and technology companies. These policies should include the policies targeted at decreasing costs and also policies designed at accelerating R&D investment in energy production technologies or advances in energy efficiency. Further, the governments, by providing incentive plans, should encourage energy and technology companies to spend more on R&D, aiming to achieve green technologies at a faster speed and increase the share of REs in supplying energy. Policy and other support mechanisms continue to play an important role in supporting the returns and lessening risks of RE and clean technology companies. Third, the policymakers should determine the significant drivers that impact the demand and supply of RE. Identifying these determinants (e.g., international remittances, economic openness, financial market development) leads to removing the obstacles that diminish RE from both consumption and supply sides and instead helps to boost the speed of transition from traditional energy to clean and green energy, which ultimately impacts the profitability of RE and clean technology companies and their stock prices, respectively.

It would be interesting to conduct further studies to probe the return and volatility dynamic nexus between the oil price and clean and RE technology companies. Also, it would be beneficial to investigate the impact of other types of risks, such as EPU and geopolitical risk, to observe which type of risk is more important and to test whether the effect is symmetry or asymmetry. Besides, it would be interesting to examine the tested relationship by the present study for other emerging and advanced countries to observe the clean and RE technology companies of which countries are the least and the most impacted by CPU. Moreover, further studies can be applied to the Fourier Engle-Granger Cointegration approach (e.g., Athari et al. 2023) to detect the long-term relationship between CPU and clean and RE technology stock prices by considering the control variables. Further, it would be useful to perform the Markov regime-switching approach to test the CPU-clean and RE technology stock prices nexus in both high- and low-volatile regimes; even quantile regression can be applied to capture this linkage in different quantiles. Moreover, EPU and geopolitical risk can be added to the model, and partial and multiple wavelet coherence can be applied to understand the multivariable linkage.