2.1.1 Spillover effects in clean energy markets.

In light of the mounting urgency of environmental concerns, green energy has emerged as a pivotal area of inquiry for scholars in recent years. In particular, the spillover linkage between clean power and other financial markets has been the subject of extensive study by numerous scholars [2,14,15]. The identification of the spillover relationship between clean energy and other financial markets facilitates the formation of more efficient portfolios for investors and contributes to the advancement of green financial markets [16]. Pham (2019) found for the first time that there is heterogeneity in the spillover linkage between traditional financial markets and clean power subsectors and argued that policies should take into account the characteristics of different subsectors of the clean energy industry [17]. Zhang et al. (2023) showed that there is a spillover relationship between green energy, electricity and traditional energy markets, with the correlation being strong at extreme quartiles [18]. Spillover shocks to financial markets from clean energy in the context of epidemics have likewise been studied by scholars, concluding that epidemics make contagion between markets stronger [19]. Chen et al. (2022) compared the spillover relationship between clean energy, fossil energy and metals markets and found that the clean energy market changed from a spillover recipient to a spillover exporter after the Paris Agreement was signed [20]. Karkowska and Urjasz (2023) examined the risk spillover relationship between clean energy, dirty energy, and the stock market and concluded that clean energy is one of the few assets that can be used as a safe haven [21].

2.1.2 Spillover effects in industrial stock markets.

Industrial stock markets have also been the main subject of academic research, usually as part of the stock market, analysing spillovers between industrial stock sector and other financial markets [22–24]. Xu et al. (2022) investigated spillovers between the stock market and the carbon market, finding that industrial stocks are risk receivers in most cases [25]. Wu et al. (2022) also explored the spillover relationship between the industrial sector and the carbon emissions market and came to a different conclusion, concluding that the industrial sector was an information transmitter for carbon emissions trading [26]. Ouyang and Shaw (2024) formed a risk spillover network of the top ten sectors of the Chinese share market and found that systemically important sectors such as industrials need more attention [27]. Nekhili et al. (2024) examined the spillover relationship of higher order moments across sectors in the European stock market and found that industry is the largest source of systemic risk [28]. Several lines of evidence suggest that the industrial sector is the largest contributor and recipient of spillovers in the system [29]. However, there is little literature that further disaggregates the industrial sector, especially by the level of carbon emissions. Zhang and Xu (2023) classified Chinese industrial stocks into clean, dirty and ordinary portfolios, and examined the spillover effects of the three portfolios with respect to oil shocks [7].

There are a number of studies in the existing literature on spillovers in clean energy markets and industrial stock markets, and it is generally agreed that these two markets have strong spillover relationships with other financial markets. This paper links the clean energy market and the industrial stock market, examining in particular the risk spillover relationship between their sub-markets.

2.2.1 Impact of economic uncertainty on financial markets.

At the micro level, an increase in economic uncertainty causes confusion among investors as well as financial institutions, which prevents them from making optimal choices, resulting in inefficiencies in the financial markets [32]. Economic uncertainty likewise has a great impact on listed companies. Rising economic uncertainty causes companies to hold a more cautious attitude towards investment, and cash holdings rise as uncertainty rises [33]. Zhou and Chen (2024) found that economic uncertainty growth is associated with a decline in collaborative research and development among firms, which hinders innovation and growth [34]. From the perspective of the financial market as a whole, a rise in economic uncertainty may cause turbulence in the financial market, both in developed and developing countries. Mo et al. (2024) used the TENET methodology to specifically examine the risk spillover of economic uncertainty on different stock markets and found that economic uncertainty in the US plays a central role in stock market returns [35]. Through the study of Kayani et al. (2024), the effect of economic uncertainty on the stock market is significant for the BRICS countries [36]. The ‘policy-driven’ nature of the Chinese stock market makes it particularly affected by economic uncertainty, so many scholars have explored the relationship between economic uncertainty and the Chinese stock market. Wang et al. (2020) emphasised the existence of asymmetric spillover effects between the Chinese stock market and EPU [37]. Xia et al. (2020) highlighted that economic uncertainty has special predictive significance for the Chinese stock market [38]. In addition to exploring the relationship between economic uncertainty and the stock market, many other scholars have discussed the relationship between EPU and financial markets such as energy market, commodity market, carbon market, etc. [39,40].Al-Shboul et al. (2023) studied the relationship between cryptocurrencies and economic policy uncertainty, and find that economic policy uncertainty has a certain negative impact on aggregate spillovers [41].Dai and Zhu (2023) used real volatility, skewness and kurtosis as risk proxies to examine the spillover relationship between different markets and find that economic policy uncertainty is the centre of risk for extreme downside kurtosis shocks [42].

2.2.2 Impact of political uncertainty on financial markets.

Political risks can directly or indirectly have a significant impact on financial markets, and political risks such as wars, terrorism, and interstate conflicts that can lead to geopolitical tensions can have a particularly dramatic impact on financial markets [43]. Mohammed et al. (2023) used the example of the Russian-Ukrainian conflict as a contingent geopolitical event, and shows that the Russian-Ukrainian conflict had a significant impact on the returns of the energy stock market [44]. The occurrence of political events causes greater disruptions to the production and exports of crude oil exporting countries, which in turn leads to production cuts in oil-producing countries as well as a decrease in the scale of crude oil exports and an increase in crude oil prices [45]. The occurrence of political events can cause economic instability, which to a certain extent affects the consumption and demand of crude oil [46]. For investors, the occurrence of political events can cause investors to panic about the increase of crude oil prices, which pushes up the price of crude oil futures out of the consideration of reducing losses [47]. In addition, the occurrence of political events can also enable investors to create hedging demand, which in turn leads to price volatility in the crude oil market [47]. The fact that crude oil is the most important raw material for the industrial sector makes it inherent that fluctuations in the price of crude oil will lead to changes in the cost of production for companies, which in turn will cause volatility in the stock market [48]. Li et al. (2024) particularly emphasised that rising political uncertainty will lead to enhanced risk spillovers from the crude oil market to the energy stock market [32]. Zheng et al. (2023) pointed out that when political events occur, political uncertainty acts as a risk transmitter, channelling risk to other markets [49].

2.2.3 Impact of climate uncertainty on financial markets.

As the climate problem is getting worse, countries are jointly or on their own introducing many policies to protect the natural environment, which have the potential to cause economic or financial shocks of varying degrees [50]. Policies for a low-carbon transition may cause the share prices of fossil fuel companies to fall, and the share prices of companies utilising clean energy are likely to rise. Xu et al. (2023) pointed out that climate uncertainty arising from the response to climate change may lead to equity risk, and investors who are unable to judge what direction the policy will take will reduce their long-term equity investments [51]. In addition, when major climate risks are encountered, executive orders imposed to minimise losses may also affect financial markets [52]. Lu Wei (2023) emphasised the heterogeneity of the impact of climate uncertainty on different sectors of the stock market, while surfacing that climate uncertainty has a good effect in predicting stock market volatility [53]. Ji (2024) incorporated several financial markets including the stock market in the systematic analysis at the same time and found that climate uncertainty has significant spillover effects on different financial markets [54]. Ding et al. (2022) argued that climate change has a long-term causality in the transmission of shocks between the carbon market and the fossil energy market [19]. Wang et al. (2023) discussed the relationship between climate uncertainty and risk spillovers between three financial markets, pointing out that climate uncertainty is a receiver of risk [13].

In summary, uncertainty arising from economic, political, and climate changes can lead to changes in the real economy, as well as investment concerns of investors, which in turn affect financial market volatility. Uncertainty arising from the response to economic and climate policy uncertainty and the occurrence of geopolitical events can easily trigger economic and financial market turbulence in this way. It is therefore necessary for this paper to incorporate all three types of uncertainty into the framework at the same time, and to explore how uncertainty will affect spillovers from clean energy and equity markets.

4.1.1 Static spillovers in the time and frequency domain.

In order to investigate the spillover connectedness between clean power and industrial share markets, we employed the DY and BK approach to gain a table of connectedness between the markets, as showed in Table 3. The value of the TCI index for the network comprising the green energy market and the industrial stock market is 67.60%, which suggests a notable spillover effect between the variables within the network. This result is considerably higher than the empirical findings for crude oil and industrial stock markets, who determined that the total connectedness value was below 50% [7]. This is related to the choice of the Brent crude oil index as an representation of the crude oil market, which does not represent the trend of local crude oil prices in China. Furthermore, the occurrence of this result is also related to the fact that the data we have chosen for both the clean energy and the industrial sector are derived from the stock market. Huang and Liu (2023) demonstrated a greater degree of connectedness between variables within the same market [71]. Finally, it can be observed that the value of the short-term TCI (60.95 per cent) is significantly greater than the value of the long-term TCI (6.65 per cent). This implies that spillovers in the network are driven by the short run, with individual assets rapidly spilling over shocks in the short run.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Static return connectedness.

https://doi.org/10.1371/journal.pone.0316171.t003

The spillover values for the ‘TO’ rows indicate that spillovers in the industrial stocks submarkets categorised by carbon emissions are larger than spillovers in the clean energy submarkets. The low-carbon portfolio has the greatest spillover effect on the network, followed by the high-carbon portfolio and the ordinary portfolio, with values of 85.25%, 82.64% and 79.92%, respectively. In the clean energy submarkets, solar energy and wind energy have the fourth and fifth highest spillovers on the network, with 64.08% and 63.71%, respectively. In comparison, the spillover result for hydropower is much lower at 30.02%. A further examination of the shocks received by the variables from the network, also known as the ‘FROM’ column, reveals that the industrial stock market is also affected to a greater extent than the clean energy market. The low-carbon portfolio receives the most shocks from the network (74.31%), while hydro energy receives the least shocks from the network (47.58%). The low-carbon portfolio, which exerts the strongest influence on the connectedness network, also receives the strongest risk spillovers from the network. This represents a very strong link between the low carbon portfolio and other markets. In contrast, the water energy market that exerts the lowest shocks to the connectedness network also receives the lowest shocks from the network. It can be argued that hydro energy is more independent in the connectedness network. The potential reason for these two results is that hydro is a relatively mature and independent energy technology, whereas many sectors in the low-carbon portfolio need to interact and integrate with network technologies frequently and have more links to other markets. This implies that the low-carbon portfolio is hardly a hedging asset, whereas hydro energy is a good hedging asset.

We have divided the table into four sections, with the top left corner showing mutual spillovers between clean energy submarkets, the bottom right corner representing mutual spillovers between submarkets in the industrial stock market, the top right corner showing spillovers from the industrial stock market to the clean energy market, and the bottom left corner representing spillovers from the clean energy market to the industrial stock market. Comparing the top left corner to the bottom right corner, it is clear that the bottom right corner has the better value. This represents that the connectedness between the submarkets of the industrial stock market is stronger than the connectedness between the clean energy submarkets. The low carbon portfolio, the high carbon portfolio and the ordinary portfolio are more strongly connected in the stock market than hydro, wind and solar. A comparison of the top right and bottom left corners of the table shows that the influence exerted by the industrial stock market on the clean energy market is stronger than the influence exerted by the clean energy market on the industrial stock market. This represents the fact that the industrial stock market dominates the entire network. This is in line with our expectations, as China’s industrial stock market not only has more participants than the clean energy market, but also has deeper linkages between internal sectors. Clean energy is also relatively independent in terms of stock market performance due to its relatively mature technology. As a result, the industrial stock market can be used as a risk indicator, and when the industrial stock market is in turmoil, a financial storm sweeping through the industrial and clean energy sectors will follow.

As shown in the ‘NET’ row of Table 3, the submarkets in the industrial equity market have positive net spillovers, while the submarkets in the clean power market have net pass-through effects of less than zero. The Low Carbon Portfolio, High Carbon Portfolio, and General Portfolio are net senders of risk premiums, while Hydro, Wind, and Solar are net receivers of risk premiums. The highest net spillover is in the low carbon portfolio (10.94%) and the lowest net spillover is in hydro (-17.57%). This is an interesting finding, with the low carbon portfolio being the largest net sender to the connectedness network and hydro being the largest net receiver. The turbulence that occurs in the low carbon portfolio will have the greatest impact on the other markets, while hydro energy passively absorbs the impact from the other markets. Zhang and Xu (2023) argue that China’s stock market is policy-driven, and as low-carbon policies continue to be implemented, low-carbon industries are receiving more attention and are more likely to generate returns for investors, making low-carbon portfolios the main sender of connectedness networks [7]. Investor attention also means that low carbon portfolios have more volume; other sectors may struggle to drive movement in low carbon portfolios, but changes in the low carbon sector will have a significant impact on other sectors. Although hydro energy is the most important clean energy source in China, its technology is relatively mature, making it difficult to enjoy changes in share prices due to technological change, and its market share is relatively stable. These characteristics determine that the main investors in hydro energy are long-term investors who do not easily sell their shares and invest in other markets. But relatively, when there are abnormal changes in other markets, some investors will flock into or out of the water energy market, causing turbulence in the water energy market.

Considering the frequency bands of connectedness, the industrial stock market is the net sender and the clean market is the receiver in both the high (short-term) and low (long-term) frequency bands. In Table 3, The normal font is for the full band spillover values, and the fonts for the high and low band values are bold and italic, respectively. In the industrial stock market, the low-carbon portfolio is the primary net sender of risk in the high frequency band and the secondary net sender of risk in the low frequency band. This suggests that the low carbon portfolio exerts a major impact on the network from the short term, and in the long term, the low carbon portfolio exerts a lower impact on the network compared to the high carbon portfolio as well as the normal portfolio. Low-carbon portfolio, as the most powerful emitter of risk, will always be able to quickly change the structure of the rest of the market in a short period of time, completing the transfer of risk. The high carbon portfolio as well as the portfolio are not only weaker than the low carbon portfolio, but are also impacted by the low carbon portfolio, so the risk spillover is more in the long term. The most visible of the clean energy markets is still the hydro energy market, which is the strongest risk receiver in both the short and long term. The hydro energy market is the strongest risk-receiving pool, and the risk of spillovers from all other markets is largely absorbed by the hydro energy market, so it is subject to major shocks from other markets in both the short and long term. In addition, it is a key finding that the impact of any market on the network and the shocks received from the network are mainly from the short term. This means that in the network, information is processed quickly and then quickly spills out of each variable and into all others. According to the fully efficient market hypothesis, changes in share prices reflect all valid information, but China’s stock market takes five days to process the information to a greater extent, which in challenges the efficient market hypothesis and supports the behavioural finance explanation of the market.

4.1.2 Dynamic total spillovers in the time and frequency domain.

We have analysed the static connectedness between the clean power and the industrial stock market above, but since the static connectedness is only an average over the sample period, it does not reflect the relationship between the markets over time. Experience has shown that spillovers between markets change as a result of changes in the economic environment, the political environment and even the public health environment, so we believe that a discussion of dynamic connectedness is warranted. In order to measure the dynamic spillovers between our chosen variables and to further uncover the relationships between the variables, we use a rolling window and split frequency bands. The rolling window is 200 days and the forecasting step is 10 days. The frequency bands are divided into high and low frequencies, corresponding to short term (1-5 days) and long term (more than 5 days). The results are presented in Fig 1.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Dynamic total spillover.

https://doi.org/10.1371/journal.pone.0316171.g001

The spillover index rises rapidly after the Chinese stock market crash, peaking at 82% throughout the sample period. And the impact of the stock market crash lasted for a long time until early 2017, when the spillover index started to decline at an accelerated rate. This suggests that the Chinese share market crash has resulted in a notable intensification of the connectedness between the clean energy market and the industrial stock market. As might be expected, the variables selected are associated with the Chinese stock market. It is therefore unsurprising that a rise in the spillover index is accompanied by a corresponding increase in the volatility of the share market. A stock market crash represents a significant setback for investor confidence, which is likely to result in a prolonged period of caution among investors with regard to further investment in the share market.

It can be observed that the connectedness index demonstrates two distinct periods of accelerated growth during the period of the US-China trade conflict. The first of these occurred in January and early February 2018, following the announcement by the US of global safeguard tariffs on imported solar panels. The China Photovoltaic Industry Association (2018) released a report indicating that China accounts for 67 per cent of the total global cell production and 71 per cent of the total global module production. Consequently, the imposition of tariffs on solar panels by the US will have a detrimental impact on the interests of China’s PV-related industries. In response, China implemented tariffs on a range of US goods, initiating a trade conflict between the two major economies. This trade conflict has resulted in an increase in the spillover index to 65 per cent. The second phase of the trade conflict commenced in September 2018, when the United States announced tariffs on approximately $200 billion of Chinese imports, and China responded by increasing tariffs on approximately $60 billion of US goods. This intensified the impact of the trade conflict, resulting in a peak in the connectedness index for the period (approximately 72%).

In the wake of the global outbreak of the novel coronavirus, a substantial corpus of literature has appeared, underscoring the profound economic impact of the pandemic [72,73]. It is evident that the outbreak led to a surge in the spillover index between markets [74,75], a finding that aligns with our own observations. In the early months of 2020, when the outbreak first emerged, the total spillover index exhibited a notable acceleration, rising from 65% to 70% in a relatively short period. This implies that the outbreak of the COVID-19 had a profound influence on the clean energy and industrial stock markets, with a concomitant increase in market risk. Subsequently, the connectedness index declined as the outbreak was effectively contained and the Chinese economy rebounded rapidly. In the aftermath of the global pandemic, the global economy experienced a decline, with financial markets in all countries experiencing turbulence. The risk spillover index increased, and market risk premiums intensified. This is evidenced by the spillover correlation between clean energy market and the industrial stock market.

At the end of 2021, the spillover index begins to rise after a period of minimal activity (42%). From this point onwards, clashes begin to emerge on the Russian-Ukrainian border, and war becomes increasingly likely. In particular, following the outbreak of the Russia-Ukraine conflict, the spillover index reached a new peak of 58%. It is evident that the Russian-Ukrainian conflict has an influence on the spillovers between the selected markets, with an increase of 10 per cent, which is the highest among the four major emergencies. It is evident that the occurrence of large wars in neighbouring countries has the most significant impact on the spillovers between markets. As a result, market participants must be concerned about possible wars in China’s neighbouring countries, which could have a huge impact on the market.

The four key periods we study represent the impact of financial crises, real economic crises, trade crises and geopolitical crises on markets. Among the four key periods, the level of total connectedness is highest during the Chinese stock market crash, followed by the pandemic period, and then by the US-China trade conflict and the Russia-Ukraine conflict. This implies that this financial and economic crisis had the greatest impact on the markets we study. The market we study is actually part of the stock market, which is an important part of the financial market, so the financial crisis will have the most direct impact on the stock market, as evidenced by the extremely high level of connectedness of the market network. Financial markets serve the real economy, and the outbreak of COVID-19 greatly disrupted China’s real economy, and the economic turmoil was transmitted to the financial sector, resulting in turmoil in the financial markets and a rise in the connectedness between stock markets. The U.S.-China trade conflict has had a significant impact on China’s commodity exports, and the decline in exports has affected China’s real economy, leading to an increase in connectedness. Moreover, the trade conflict between the U.S. and China has a tendency to intensify, which could easily lead to investors’ disillusionment with the Chinese market, and the lowering of expectations can also affect the financial market. The Russian-Ukrainian conflict has the lowest level of connectedness because there is not much direct connection between the Russian-Ukrainian conflict and China. The main reason for the increase in the level of market connectedness due to the Russian-Ukrainian conflict may be attributed to investors’ fear of World War III.

Examination of the frequency bands shows that the level of total spillover connectedness is consistently and significantly higher in the high-frequency bands (blue line) than in the low-frequency bands (green line). The proportion of high-frequency connectedness fluctuates between 40% and 70% over the entire sample period, while the proportion of low-frequency spillovers fluctuates between 5% and 10%. The high volatility of high-frequency connectedness is suggestive of the vulnerability of the Chinese stock market and the risk of causing short-term turbulence in the Chinese stock market on a regular basis. In addition, the study also finds that short-term aggregate spillover connectedness is highly correlated with the aggregate spillover index, especially in the case of sudden large-scale events. This proves that short-term aggregate spillovers dominate the entire market, especially in the case of unexpected major events, where risks spill over quickly and cause unimaginably heavy damage to the entire network. In contrast, the long-term connectedness index shows greater stability. This suggests that major events and minor surprises mainly affect short-term market trajectories, with minimal impact on the long term. In terms of frequency band performance over the four key periods, the highest total short-term spillover effect is observed during the Chinese crash, representing a period of great risk, cumbersome information, and constant shock to the stock market. Moreover, the performance of the Chinese stock market crash in terms of connectedness in the low frequency bands is also noteworthy, with spillover effects also being the highest in the four periods. The occurrence of a stock market crash not only causes market turmoil in the short term, but its effects continue to be felt for a long period of time before the market fully digests them. The Russian-Ukrainian conflict had the lowest impact on the market in both the short and long term. This is because the Russian-Ukrainian conflict took place on the border between Russia and Ukraine, and China is not connected, the fire did not spread to the Chinese border. As for why the Russia-Ukraine conflict could affect China’s stock market, we suspect it’s because people are worried that the Russia-Ukraine conflict could be a trigger for World War III. And the impact of war on financial markets is extremely scary [21].

4.1.3 Dynamic net spillovers in the time and frequency domain.

The dynamic total spillover connectedness index represents the overall average of the network of selected variables. However, this index does not permit the observation of the role played by each submarket separately. Consequently, it is necessary to analyse the net connectedness index. The role played by each submarket will also be examined in both the time and frequency domains, in a manner analogous to the analysis of the dynamic total spillover index. Fig 2 presents the dynamic net spillover index for each market, offering a comprehensive visual representation of the direction of spillovers in each market.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Net return spillover.

https://doi.org/10.1371/journal.pone.0316171.g002

The clean energy submarket is a net recipient throughout the sample period, and even if occasionally the spillover value is greater than zero, it immediately turns around and returns to a value less than zero. This is consistent with our conclusion from analysing static spillovers that the clean energy market is a recipient of risky spillovers in the network. In addition, our results show that hydropower is the most volatile, with fluctuations of up to -30% and zero. This highlights the high sensitivity and volatility of the hydropower market to changes in the overall network. Of the four key periods, hydro’s performance during the crash is noteworthy. Hydro Energy’s net risk acceptance declined in the wake of the crash. The onset of a financial crisis can trigger panic among investors. While at other times hydroenergy has less of an impact on other markets, as panic spreads, hydroenergy in turn increases its impact on other markets. This then leads to a decrease in the net risk acceptance of the water energy market. In the other three periods, the hydro energy market played its full role as a risk absorber as it was not directly linked to the crises represented in these three periods. Wind energy’s performance during the U.S.-China trade conflict is the most noteworthy, with its initial performance as a strong net risk taker followed by a rapid decline in capacity. Both China and the US are leaders in wind energy, and the trade conflict between the US and China has partly centred around wind products, which has directly impacted the Chinese wind market, causing it to also increase its impact on the network, which in turn has resulted in a decline in the net risk-receiving capacity of wind energy. Similarly to hydro, the wind market’s net risk-absorbing capacity weakened during the stock market crash, acting as a risk receiver during the pandemic and during the Russian-Ukrainian conflict. Solar’s performance during the four pandemics is also noteworthy, as it demonstrated its strong risk-absorbing capacity during both the U.S.-China trade conflict and the Russian-Ukrainian conflict, both of which saw a sudden increase in risk absorption followed by a decline. This is related to solar’s role as a risk receiver, where clean energy is forced to absorb risk from the market network when it is hit. And as events unfold and the market stabilises, the risk absorbed by clean energy declines, as does the risk absorbed by solar. The Russian-Ukrainian conflict strained the raw materials associated with the solar industry, and the solar market suddenly increased its risk spillover to the network, causing a sudden decline in its role as a risk-absorbing pool.

In the industrial stock market, the ordinary and high-carbon portfolios act as net senders of risk for the vast majority of the period, and the low-carbon portfolio is a net sender of risk for the entire sample period. Consistent with Zhang and Xu’s study (2023) [7], as the Chinese government increases its support for low-carbon emitting industries, high-carbon portfolios continue to be subject to risk spillovers from both low-carbon and ordinary portfolios, forcing them to become net receivers of network. Whereas the shift to high-carbon sectors takes a significant amount of time, as the emotional impact of the policy wanes, high-carbon portfolios gradually increase their influence and turn back into net risk spillovers. The ordinary portfolio begins to turn into a net recipient of risk in 2021, a time period in which the COVID-19 pandemic has been effectively contained, the Chinese economy is experiencing a rebound, and the ordinary portfolio rapidly reduces its risk spillover to the network while continuing to receive risk spillover from other markets. At the onset of the crisis, industrial stock market performance was similar across carbon emissions. All three portfolios experienced a decline in net spillover capacity during the crash, due in large part to the diminished ability of clean energy to act as a risk absorber and to spill over greater risk to industrial equity markets. In the other three important periods, all three portfolios acted as strong risk emitters. As investor sentiment stabilised, the industrial equity submarkets exerted less influence on the network and the net spillover effect diminished.

An analysis of the performance of the spillover indices across the frequency bands indicates that the long-term net spillover index remains at approximately zero in all six sub-markets, while the short-run net spillover connectedness index demonstrates considerable fluctuation, essentially aligning with the performance of the total net spillover connectedness index. This suggests that short-term spillovers play a dominant role in each market, consistent with the previous analysis. Similarly, we can observe the changes in the short-term net spillover index and the long-term net spillover index over the four crisis periods. It is evident that the short-term net connectedness indices of the six markets behave broadly in line with the total net spillover indices over the four periods, i.e., when a crisis occurs, markets react quickly to spillover risk to the network and to receive spillover risk from other markets. The behaviour of each market’s long-term spillover index in times of crisis is more complex. Some have stable long-trem spillover indices while short-term and total net spillover indices spike, such as the performance of industrial stock submarkets during the Russia-Ukraine conflict; some have long-term spillover indices that move in the opposite direction of total net spillover indices, such as the hydropower during the COVID-19 pandemic; and some have long-run connectedness indices that behave in line with long-term and total net spillover indices, such as clean power markets during the trade conflict. clean energy submarket during the trade conflict. Furthermore, it can be observed that industrial stock submarkets experience a significant increase in short-term net spillovers in times of crisis, while long-term net spillovers are stable, implying that industrial stock submarkets rapidly spill risk into the network in the short term, with spillovers outweighing the risk received from the network. The long-run net spillover connectedness index of the clean power submarkets basically moves together with the short-term net spillover index, implying that the clean energy submarkets are more permanently affected by the crisis shocks and have more uncertainty about the future than the industrial stock submarkets.

4.2.1 Asymmetric static spillovers in the time and frequency domain.

In order to study the asymmetric spillover effects of the network, we further obtain the spillover tables for different frequency bands, both for positive and negative series. Table 4 shows the asymmetric spillovers between markets for the full frequency band, while Tables 5 and 6 show the asymmetric spillover matrices for networks in the high frequency band (short term) and low frequency band (long term), respectively. Values without parentheses are positive spillover and those with parentheses are negative spillover. Table 4 shows that the overall bad spillovers are higher than the overall good spillovers for the full frequency bands, their values are 68.84% and 61.87% respectively. This result shows that the overall connectedness of the network is asymmetric and that the bad news is a bigger shock to the network. Notably, the spillovers that each market receives (transmits) from the network are heterogeneous and exhibit asymmetry under good news shocks and bad news shocks. Spillovers from all six markets to the network exhibit stronger bad than good returns, while spillovers from the network to each market, except for LC, also exhibit stronger bad than good spillovers. That is, all markets except the low-carbon portfolio are more sensitive to bad news, which is in line with the research by Zhang and Xu (2023) [7]. The result that markets are more drastically affected by bad news supports the idea of loss aversion in behavioural finance. Investors are more averse to losing gains of equal value than they are to gaining gains of equal value. For the low-carbon portfolio, good news receives more spillovers from the whole network compared to bad news, which is an interesting finding. In terms of net spillovers, the industrial stock submarkets are net transmitters and the clean energy submarkets are net receivers, both under positive and negative shocks. That is, the industrial stock submarkets transmit more risk to the network, regardless of whether it is good message or bad message, while the clean energy submarkets act as risk absorbers for the network.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Asymmetric static connectedness in the full frequency band.

https://doi.org/10.1371/journal.pone.0316171.t004

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Asymmetric static connectedness in the high frequency band.

https://doi.org/10.1371/journal.pone.0316171.t005

Download:

• PNG
  larger image
• TIFF
  original image

Table 6. Asymmetric static connectedness in the low frequency band.

https://doi.org/10.1371/journal.pone.0316171.t006

Similarly, the spillover effects in different frequency bands are asymmetric. Table 5 shows that the total spillovers in the higher frequency bands are 52.71 per cent and 60.41 per cent for positive and negative shocks, respectively. This indicates that spillovers are more susceptible to unfavourable information in the short run. Bad news brings panic to investors, which quickly spreads across markets and causes market turmoil. Table 6 shows that the total spillovers from positive and negative shocks in the low frequency band are 9.16% and 8.43%, respectively. In the long run, good news can generate sustained positive sentiment in the market. The persistence of this positive sentiment is stronger than the negative sentiment from bad news. It is worth noting that this difference between the high and low frequency bands is also reflected in the TO and FROM columns. In the high frequency band, the impact of bad news has a greater impact, both in terms of transmitting risk to the network and receiving risk from the network. In the low frequency band, on the other hand, good news dominates. In terms of net spillovers, the industrial equities submarket is a net sender and the clean energy submarket is a receiver of risk, regardless of which frequency band it falls into.

4.2.2 Asymmetric total spillovers in the time and frequency domain.

Fig 3 provides a comparison of the influence of positive and negative shocks on total connectedness over time at different frequencies. It is clear that both the full frequency band and the short-term aggregate spillovers are dominated by negative returns over the sample period. Spillovers in the long-term are different, with spillovers from positive returns being stronger than spillovers from negative returns. Regardless of the frequency band they belong to, the spillover indices show asymmetry over time. The same can be observed for positive and negative returns in times of crisis, and it can be observed that the full frequency band and the short-run spillovers are similar in that there is a sudden rise in times of crisis. The long-run spillovers, on the other hand, are relatively moderate in the crisis period compared to the short-run spillovers. That is, the shocks from the crisis were absorbed in the short term and had little impact in the long term, a finding consistent with the previous section. Moreover, in terms of long-term spillovers, it can be found that good news brings more long-term shocks than bad news, because the spillover connectedness of positive returns is larger than that of negative returns.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Asymmetric dynamic total spillover in different frequency bands.

https://doi.org/10.1371/journal.pone.0316171.g003

Good news and bad news perform essentially similarly in the four key periods across frequency bands. Consider first the performance of the four periods in terms of total spillovers. During a stock market crash, good news and bad news cause about the same shocks, with the bad news perhaps a little ahead. In this period, the financial catastrophe tied all financial markets together so that good news and bad news had almost the same impact on markets. In the other three periods, shocks caused by bad news are in the lead, due to the fact that markets react more violently to bad news. The spillovers in the high and low frequency periods behave in the same way as the total spillovers, with bad news leading the way, but the difference is that the difference between the spillovers from shocks to the market caused by bad and good news is greater in the high frequency period than in the low frequency period. This is mainly due to the fact that spillovers from good and bad news are dominated by high-frequency spillovers.

4.2.3 Asymmetric net spillovers in the time and frequency domain.

The net spillovers over time for different markets under positive and negative shocks are shown in Fig 4, while Figs 5 and 6 represent the positive and negative net connectedness for each variable under different frequency bands. Fig 5 shows the findings in the high frequency band and Fig 6 shows the findings in the low frequency band. It is worthwhile to identify the dominant spillover effect in each market over time, so that we can find out which market is most affected by spillovers in response to good news (bad news) shocks. The larger the absolute value of the connectedness index, the greater the impact on the market of the positive (negative) shock represented by that spillover index. Among the green energy submarkets, it is clear that the hydropower is dominated by negative spillovers for long periods of time, and that the hydro market is a receiver of risk from the system network. This suggests that when bad news hits the hydro market, the hydro market will be hit harder by the network. The wind and solar markets, on the other hand, are dominated by positive spillovers. That is to say that good news shocks the wind and solar markets, causing these two markets to absorb more risk from the network. For the industrial stock submarkets, the low-carbon portfolio is found to be overwhelmingly dominated by positive spillovers most of the time, while it is difficult to tell who is in the lead for the ordinary and high-carbon portfolios. Therefore, when there is good news for the low-carbon portfolio, more attention needs to be paid to the impact it has on the systemic network.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Asymmetric net return spillovers in the full frequency band.

https://doi.org/10.1371/journal.pone.0316171.g004

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Asymmetric dynamic return spillover in high frequency bands.

https://doi.org/10.1371/journal.pone.0316171.g005

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Asymmetric dynamic return spillover in low frequency bands.

https://doi.org/10.1371/journal.pone.0316171.g006

Examination of the net positive and negative spillover indices for each market over different frequency bands shows that the short-term indices trend in line with the full-frequency indices. Hydropower markets are dominated by negative spillovers, while wind and solar markets and low-carbon portfolios are driven by positive spillovers in most cases. In the lower frequency band, i.e., the long-term spillovers, we note a moderate trend in the green line (bad news) and a significantly larger magnitude in the blue line (good news). This suggests that spillovers between the clean power submarket and the industrial equities submarket, driven by good news, are more noteworthy in the long run. In addition, the behaviour of positive and negative spillovers over four key periods in different frequency bands is also of interest. Hydropower is more risk-absorbent in good news shocks during the crash, but the effect is less pronounced. In the other three periods, it shows a clear dominance of negative spillovers. Water energy is more vulnerable to bad news shocks during crisis periods, with the exception of the financial crisis. The financial crisis made the financial markets so interconnected that both good and bad news hit the hydro market very hard. Wind is dominated by positive spillovers in all four periods, which is an interesting finding, as good news strengthens the risk-absorbing capacity of wind. Solar is more risk-absorbent in the presence of good news in the first three periods, while risk-absorption is stronger in the presence of bad news during the Russia-Ukraine conflict. It may be that bad news during the Russia-Ukraine conflict had a greater impact on other markets, leading to enhanced risk spillovers from other markets to the solar market. Similarly, the low-carbon market is dominated by good news shocks in the first three periods, acting as a risk transmitter and exerting influence on other markets. During the Russian-Ukrainian conflict, it was dominated by shocks caused by bad news, which acted as an emitter of bad news causing enhanced risk spillovers from the low carbon market to other markets. The performance of the high and average carbon markets is indistinguishable across the four periods, with both showing an equalisation of the spillover effects of good and bad news on market shocks. The high frequency period is the same as the total spillover effect, while the low frequency period is an equilibrium of net spillovers due to good news and bad news during the stock market crash, during the pandemic, and during the Russian-Ukrainian conflict, with risk spillovers due to good news stronger than to bad news during the U.S.-China trade conflict. The potential reason for this may be that the arrival of good news can enhance the risk spillover from clean energy to the industrial stock market.

4.3.1 Selection of variables in the regression equation.

To ascertain the impact of uncertainties on spillovers between these markets, this paper incorporates uncertainties and total spillovers into the regression equation for an OLS regression. We select three variables, global economic policy uncertainty (EPU), global geopolitical uncertainty (GPR), and climate policy uncertainty (CPU), as explanatory variables, the Arca Tech 100 index (PSE), the CSI 300 index (CSI), the US dollar exchange rate against the renminbi (FX), and a dummy variable (DR) as control variables, and the different frequency bands (full frequency, high frequency, and low frequency) of total connectedness as dependent variables for the regression of the equation. The dummy variables are those assigned a value of 1 for time of crisis and 0 for non-crisis periods. Of course, the processing of the raw data is necessary, and we take the monthly means of all the data and treat them logarithmically except for the dummy variables, which are unchanged. Considering the availability of data, we select the monthly data from January 2014 to August 2023 for the study. The EPU index was developed by Baker et al. (2016) and there is evidence that it has an impact on financial markets and that economic policy uncertainty has a positive effect on connctedness [76]. The GPR index is derived from Caldara and Iacoviello (2022) and is based on their count of newspaper articles on geopolitical tensions [77]. Feng et al. (2023) showed that geopolitical risk triggers an upward trend in the level of market spillovers, due to the fact that risky events can exacerbate market price volatility [78]. Finally, the CPU index constructed by Gavriilidis (2021) affects price volatility in financial markets by influencing investors’ expectations [79]. The choice of control variables refers to the studies in these articles [6,18].

4.3.2 How uncertainties affect the spillover effect.

Table 7 shows the regression results for uncertainty on aggregate and asymmetric spillovers, and it is notable that not all uncertainty has a significant effect on spillovers, and that different aggregate spillovers behave differently. A review of the regression results for total spillovers reveals that, with the exception of GPR, the remaining variables exert a notable influence on total spillovers. Among the factors under consideration, EPU has been identified as exerting a considerable positive influence on the total spillover effect. This implies that an increase in EPU is associated with a corresponding rise in the interconnectivity between the green energy market and the industrial share market. The finding of this result supports the idea of the herd effect in behavioural finance, where investors may tend to mimic the trading behaviour of others in order to reduce their own decision-making pressure when market uncertainty increases, and this herd effect leads to a rise in linkages between different sectors. CPU also significantly affects aggregate spillovers, but in the opposite direction. That is, concerns about climate issues can lead to a strengthening of the market’s spillover to itself. This is an interesting finding, as some of the literature indicates that CPU causes spillovers between markets to rise [54,79]. Among the control variables, PSE affects aggregate spillovers in the reverse direction, while the other variables CSI, FX and DR all have a positive contribution to aggregate spillovers. All the significant variables hold at 1% level of significance except DR, where the original hypothesis is rejected at 10% confidence interval.

Download:

• PNG
  larger image
• TIFF
  original image

Table 7. Regression results of uncertainties on spillover.

https://doi.org/10.1371/journal.pone.0316171.t007

We also verify the effect of uncertainties on asymmetric spillovers, that is, on positive and negative spillovers. The regression results for positive spillovers show that in the ‘positive’ column, among the various types of uncertainty, climate policy uncertainty has a remarkable inverse effect on positive spillovers, whereas economic policy uncertainty and global geopolitical uncertainty do not have a significant effect on positive spillovers. The variables (PSE, CSI, FX) give a significant effect on positive total connectedness, with PSE having a negative effect and CSI and FX having a positive effect. DR, being a dummy variable for the crisis period, does not give a significant effect on positive spillovers. An examination of the negative spillovers, as reflected in the “negative” column, reveals that the significance of the variables is basically the same as the total spillovers, i.e., EPU, CSI, FX, and DR positively affect negative spillovers, and CPU and PSE have a negative correlation on negative spillovers. In general, among the various uncertainties, only climate policy uncertainty has a notable impact on all three types of spillovers. Economic policy uncertainty has a significant effect on both total and negative spillovers, while geopolitical risk does not have a significant effect on any of the three spillovers. That is, both investors and policymakers should pay attention to changes in climate policy uncertainty as well as economic policy uncertainty because they can cause some degree of shock to the market. Moreover, during bad news shocks, economic uncertainty and climate policy uncertainty are sensitive to spillovers, while during good news shocks, only climate policy uncertainty is sensitive to spillovers. Therefore, the attitude towards these two types of uncertainty should be different under different shocks.