Green bonds’ liquidity in COVID-19 and low carbon investments in China: A stochastic trend analysis

Abstract

Central banks and regulators increasingly consider climate-related financial risks (CRFR) relevant to their responsibilities for maintaining financial stability and using daily data from 2016 to 2021 for China. Specifically, we used the S&P Green Bond Price Index, the Solactive Global Solar Price Index, the Solactive Global Wind Price Index, and the S&P Global Clean Energy and Carbon Price Index as our data set. We use the TVP-VAR method to probe return spillovers and interconnectedness. We test several portfolio strategies, including the minimum variance portfolio, the minimum correlation portfolio, and the more recent minimum connectedness portfolio. However, the evolving policy structure for dealing with CRFR has generally focused on market-based solutions that attempt to address perceived data gaps that preclude the appropriate pricing of CRFR, even though CRFR is thought to have certain distinctive features. Disclosure and openness fall within this category. We propose limiting the approach’s influence since CRFR is characterized by extreme attainability. A ‘precautionary’ financial policy option is presented as an alternative, providing a conceptual foundation for justifying more aggressive financial policy intervention in the present to better cope with these long-term dangers.

Introduction

Green financing is one of the most well-liked and widely recognized approaches to reconciling the goals of ecology and business (Sun et al. 2019a). The system is working, as the booming green bond market attests. Despite being a relative apparatus, the green bond industry has provided significance through reduced financing expenses and increased long-term liquidity, as well as the participation of investors who are partial to green bonds and who are interested in environmental protection and portfolio diversification (Mohsin et al. 2020). Since the European Investment Bank issued the first ‘green bond’ in 2008, the market for these types of bonds has expanded exponentially, demonstrating widespread interest in them. The Climate Bond Initiative predicts that by 2019, the total amount of green bonds issued will have surpassed $1 trillion. Despite the unusual and continuing detrimental effects of the COVID-19 outbreak, the achievement of sustainable development goal shows an increasing trend (Sun et al. 2019b). The market for green bonds will unquestionably grow rapidly shortly.

According to research, companies may reap financial rewards from socially responsible projects, including climate change and ecological degradation mitigation (Iram et al. 2020). Coincidentally, these non-monetary factors distinguish green resources from conventional financial resources and are the primary draw for shareholders. Because of this shift, scientists are now questioning whether or not businesses can benefit from engaging in environmentally responsible practices (Iqbal et al. 2021). For example, would green financial products’ dynamics differ from conventional fixed-income equities and bonds? Specifically, would climate-focused shareholders encounter the same uncertainty when purchasing environmentally friendly financial instruments? In this study, we analyse the second part of this question the dynamics of unexpected occurrences and their consequences for investors using tools borrowed from control theory and stochastic dynamic macroeconomic models (Anh Tu et al. 2021). When making predictions, rational investors consider all the data at their disposal. Therefore, any unexpected events must be accounted for as data volatility. For the same reason that ‘correctly expected share prices vary arbitrarily’, as Mohsin et al. (2022a) phrased it, this idea is central to the efficient market hypothesis (EMH) made famous by the pioneering work of Zerbib (2019). Formal frameworks benefit from being theoretically coherent and mathematically treatable, even if the efficient market hypothesis is hotly contested in the financial sector. In addition, this structure allows for a theoretical separation between the intended (and plannable) changes and the unanticipated (and non-playable) implications of new information. We suggest preventive, necessary for light of this, and the well-accepted reality that delays in taking action would enhance the severity of CRFR. The ‘precautionary principle’, which advocates preventive actions that safeguard. The shortcomings of macro prudential policy in addressing have gained traction in the aftermath of the crisis (Li et al. 2021).

Our study contributes to the literature in the following ways. Using a TVP-VAR strategy to address the threats to financial stability caused by climate change in China, we were using daily data beginning in 2016–2021. Specifically, we used the S&P Green Bond Price Index, the Solactive Global Solar Price Index, the Solactive Global Wind Price Index, and the S&P Global Clean Energy and Carbon Price Index as our data set. To probe return spillovers and interconnectedness, we use the TVP-VAR method, and we first analyse how current policy may change with a cautious strategy structure. The financial hazards associated with climate change are the primary emphasis of this research rather than wider climatic issues. We measures an economic volatility; having New Green Finance Standard (NGFS) and the growing importance of central banks in the macro economy since 2005–2006.

Here is how the rest of the paper is laid out. In the ‘Literature review’ section, we survey the research on economic risk and uncertainty from a scholarly and policy perspective before analysing how these concepts pertain to CRFR. Our rationale for adopting a cautious stance toward CRFR is laid forth in the ‘Methodology’ section. The ‘Empirical results and discussion’ section explores the empirical results and discussion. Finally, conclusion and policy recommendations for further study are presented in the ‘Conclusion and policy recommendations’ section.

Literature review

The green bond market

There are two main lines of inquiry. In the first, which is focused on administration, efforts are made to improve market openness and efficiency via regulatory and advisory frameworks. The term ‘green bond’ is challenging to pin down because of its vague meaning. If the green bond market fails because of worries about greenwashing, it would be a major environmental setback which argues that the greenwashing loophole is one of the biggest problems facing authorities (Mohsin et al. 2022b). Even though several standards have been created outlining broad requirements for issuing a green bond, optional procedures, worldwide harmonization has improved. To advance a viable and sustainable green bond market, it is crucial to improve the uniformity of procedures and criteria in identifying the existing qualification (Mohsin et al. 2020). While this is happening, the financial sector is pushing for more information on how green initiatives that qualify for funding might help achieve environmental goals. The second line of thought is market-oriented and is concerned with recording market pricing dynamics to aid players in comprehending market behaviour and making more sound expenditure choices (Zhang et al. 2022). The bulk of articles in this section investigates the profitability of the green bond market, specifically the presence of green bond premiums compared to more traditional bond alternatives (Abbas et al. 2022).

As financial markets become more interconnected, data may move freely across them (Yang et al. 2022). Because of this trend, researchers have begun investigating interactions between the green bond trade and other areas of the banking industry (Sun et al. 2022). According to the data gathered, green bonds and other fixed-income securities have very little in common (Huang et al. 2021). However, as highlighted, the green bond industry is connected to the government and corporate bond markets (Zhang et al. 2021). However, there is not much of a connection between the green bond industry, the stock market, and the energy industry. The bond industry for ecological-friendly securities is linked to the foreign currency industry, as shown by the structural vector autoregression (VAR) model (Ullah et al. 2020). Based on TVP-VAR analysis, it is argued that a negative market environment is associated with a strong link between oil prices and the green bond indicator. However, this relationship fades when the industry enters higher quantiles (Xia et al. 2020). Likewise, smaller quantile fluctuations in the Green Bond Index may be predicted with high accuracy by levels of geopolitical uncertainty. The findings demonstrated a favourable and statistically important relationship between investor attitude and green bond market performance. When compared, green bond sector returns are unfavourably impacted by the performance of the share and non-renewable energy sectors (Shah et al. 2019).

Related studies involving green bonds

Researchers have shown that green bonds, as opposed to traditional bonds, carry a substantial ‘green bonds and financial markets’. Green bonds have been the subject of several studies in recent years, many of which have looked at the interplay between these instruments and the financial markets to conclude portfolio and hedging capabilities (Mohsin et al. 2020). Even though climate bonds have become a popular expenditure option among those concerned about the future of our planet, there is still room for expansion in the growing but still sparse body of scholarly work that examines the interplay between these instruments and the global financial system (Zhao et al. 2022). Using copula functions and conditional diversification metrics, Iqbal et al. (2019) demonstrate that green bonds are highly correlated with state and business bonds but very weakly correlated with the share and energy sectors. In addition, they show that green bonds and state bonds may be combined to diversify risk, but only marginally so for the share and energy sectors. Green bond prices are affected by movements in the fixed-income and currency markets, as reported by using a variety of techniques based on the structural vector autoregressive (VAR) model. However, there is little interaction between green bonds and the equities or energy sectors. The effects of green bonds on the financial markets of the USA and the European Union are studied by Iqbal et al. (2022). After decomposing the series using a wavelet-based technique, they used a VAR methodology to analyse connectivity measures, discovering a relationship between green bonds, treasury bonds, and business bonds across time horizons in the China. They also reveal that green bonds have only a weak relationship with the higher business bond, equity, and energy markets across all time horizons (Wang et al. 2022). Green bonds are analysed in connection to the 10-year China and the treasury bond index, the clean energy share price index, and CO2 Emission Allowance pricing. Their findings reveal that green bonds cannot forecast the price fluctuations of the three asset indexes under consideration since there is an inconsistent indication of causation going from each asset to green bonds across different periods. Return spillovers among green and non-green funding are greater during stressful times and are driven by socioeconomic circumstances, according to a quantile-based methodology used by Zheng et al. (2022); nobody has studied the dynamics of green bond market volatility. The findings demonstrated that the green bond market generally experiences a domino effect of volatility from the traditional bond market.

Meanwhile, displayed idiosyncratic features typical of financial markets show that financial market uncertainty significantly impacts green investments. Agyekum et al. (2021) found that investors’ focus may have a time-varying effect on the volatility of green bonds. Green assets’ reaction to oil price fluctuations was studied by Xiuzhen et al. (2022). The results imply that environmentally friendly investments are more exposed to oil market volatility than pricing. Based on these results, greater research into the correlation between volatility sectors is required to comprehend the green bond market’s dynamics fully (Liu et al. 2022c).

The present study’s findings fit neatly into the expanding body of important administration green economic sector, specifically the green bond industry. This is a first study of its kind to predict R.V. in the green bond market by elucidating the difference in cyclicality R.V. in the green bond market and R.V. in other markets.

Methodology

Time-varying parameter vector autoregression

By estimating a TVP-VAR model with homoscedasticity variance–covariance, we investigated the dynamic relationships between renewable energy, green bonds (Huang et al. 2022), wind, solar, and carbon dividends. The TVP-VAR (1) model is chosen according to the Bayesian estimated model (BIC), which may be expressed quantitatively as,

$$y_t=B_t{y}_{t-1}+{\epsilon }_t{\epsilon }_t\sim N(0,{\mathrm{\Sigma }}_t)$$ 1

$$\mathrm{vec}(B_t)=\mathrm{vec}(B_{t-1})+v_t{v}_t\sim N(0,S_t)$$ 2

$${\phi }_{\mathrm{ij},t}^{\mathrm{gen}}(H)=\frac{{\sum }_{h=0}^{H-1}{(e_i^{{}^{\prime }}{A}_{\mathrm{ht}}{\mathrm{\Sigma }}_t{e}_j)}^2}{(e_j^{{}^{\prime }}{\mathrm{\Sigma }}_t{e}_j){\sum }_{h=0}^{H-1}(e_i^{{}^{\prime }}{A}_{\mathrm{ht}}{\mathrm{\Sigma }}_t{A}_{\mathrm{ht}}^{{}^{\prime }}{e}_i)}$$ 3

$${\mathrm{gSOT}}_{\mathrm{ij},t}=\frac{{\phi }_{\mathrm{ij},t}^{\mathrm{gen}}(H)}{{\sum }_{k=1}^K{\phi }_{\mathrm{ik},t}^{\mathrm{gen}}(H)}$$ 4

where ei is a zero-vector in K dimensions, and the position is set to unity is the unscaled (j = 1) K. Normalizing it by splitting by the row sums yields the scaled bond index, ${\mathrm{gSOT}}_{\mathrm{ij},t}$, as recommended by Dilanchiev and Doctor (2021).

The basic idea behind the interconnectedness method is the scaled GFEVD, which makes it possible to calculate the total unidirectional interconnectedness to (from) all series from (to) series. While, the influence of series I on all other series is represented by the TO total directional connectivity, the influence of all series on series I is shown from the total bidirectional connections. These measurements of connectivity are computed via,

$$S_{i\to \bullet ,t}^{\mathrm{gen},\mathrm{to}}={\sum }_{j=1,i\ne j}^K{\mathrm{gSOT}}_{\mathrm{ji},t}$$ 5

$$S_{i\to \bullet ,t}^{\mathrm{gen},\mathrm{to}}={\sum }_{j=1,i\ne j}^K{\mathrm{gSOT}}_{\mathrm{ji},t}$$ 6

The net total directed connectivity of series I is calculated by subtracting the TO total interconnectedness from the FROM total interconnectedness.

$$S_{i,t}^{\mathrm{gen},\mathrm{net}}=S_{i\to \bullet ,t}^{\mathrm{gen},\mathrm{to}}-S_{i\leftarrow \bullet ,t}^{\mathrm{gen},\mathrm{from}}.$$ 7

If $S_{i,t}^{\mathrm{gen},\mathrm{net}}$ >0 ($S_{i,t}^{\mathrm{gen},\mathrm{to}}$<0), series i series equation (7) drives the connectivity because it affects all other series.

There is also extended information available (Ren et al. 2022) via the connectivity method. Bilateral net conduction of shocks within series I and j may be seen in the net paired directed interconnectivity.

$$S_{\mathrm{ij},t}^{\mathrm{gen},\mathrm{net}}={\mathrm{gSOT}}_{\mathrm{ji},t}-{\mathrm{gSOT}}_{\mathrm{ij},t}.$$ 8

f $S_{\mathrm{ij},t}^{\mathrm{gen},\mathrm{net}}$ >0 ($S_{\mathrm{ij},t}^{\mathrm{gen},\mathrm{net}}$ <0), series i stating that series I has a stronger impact on series j than the reverse is true.

Another useful indicator of price volatility and network interconnection is the total interconnectedness index (TCI). Considering that the TCI may be computed as the average total directional connectivity to (from) others, it is equal to the average quantity of spillovers one series sends to (receives from) all others (Nhuong and Quang 2022; Maithya et al. 2022). As own variance portions are always greater than or equal to all cross volatility shares, Pincus and Winters (2019) have proved that the TCI is within the plausible range 0. To obtain a TCI which is within (Downturn 2019), we have to adjust the TCI slightly:

$${\mathrm{gSOI}}_t=\frac{1}{K-1}{\sum }_{i=1}^K{S}_{i\leftarrow \bullet ,t}^{\mathrm{gen},\mathrm{from}}=\frac{1}{K-1}{\sum }_{i=1}^K{S}_{i\to \bullet ,t}^{\mathrm{gen},\mathrm{to}},$$ 9

When the value is high (low), market risk is high (low).

We conclude by computing its pairwise wholeness index (PCI), which can be considered the TCI on the bilateral level, exemplifying the degree of interconnection among both series I and j. A possible formulation of this is:

$${\mathrm{PCI}}_{\mathrm{ij},t}=2\left(\frac{{\mathrm{gSOT}}_{\mathrm{ij},t}+{\mathrm{gSOT}}_{\mathrm{ji},t}}{{\mathrm{gSOT}}_{\mathrm{ii},t}+{\mathrm{gSOT}}_{\mathrm{ij},t}+{\mathrm{gSOT}}_{\mathrm{ji},t}+{\mathrm{gSOT}}_{\mathrm{jj},t}}\right),0\le {\mathrm{PCI}}_{\mathrm{ij},t}\le 1.$$ 10

Portfolio back-testing models

Using a portfolio back-testing technique, we analyse the earnings quality of the securities under study to acquire new insights into the insuring capacity of bond issuance over sustainable energy stocks and their macroeconomic importance (Diniz et al. 2022). We use many estimating methods throughout inventory creation, including time-tested and more modern methods focused on connectivity. Our portfolio analysis is based on a set of assumptions. The index is available for direct purchase by investors, and there is a development to green finance through both green bonds and sustainable energy equities among buyers and sellers (Darling et al. 2022). These are restrictive assumptions. The following is a quick synopsis of the many portfolio estimate methods used.

Bilateral hedge ratios and portfolio weights

Jinzhou (2011) formulate the following,

$${\beta }_{\mathrm{ij},t}={\mathrm{\Sigma }}_{\mathrm{ij},t}/{\mathrm{\Sigma }}_{\mathrm{jj},t},$$ 11

where Σij,t is the indirect effect of serial is the contractual covariance I at iteration t.

Bilateral array weights of series I and j, as shown by Hafner et al. (2020), may be determined as:

$$w_{\mathrm{ij},t}=\frac{{\mathrm{\Sigma }}_{\mathrm{ii},t}-{\mathrm{\Sigma }}_{\mathrm{ij},t}}{{\mathrm{\Sigma }}_{\mathrm{ii},t}-2{\mathrm{\Sigma }}_{\mathrm{ij},t}+{\mathrm{\Sigma }}_{\mathrm{jj},t}},$$ 12

$$w_{\mathrm{ij},t}=\left\{\begin{array}{cc}0,& \mathrm{if}{w}_{\mathrm{ij},t}<0\\ w_{\mathrm{ij},t},& \mathrm{if}0\le w_{\mathrm{ij},t}\le 1\\ 1,& \mathrm{if}{w}_{\mathrm{ij},t}>1\end{array}\right)$$ 13

The Markovitz Variance Parity (MVP) methodology is widely used in equity valuation; it seeks to construct portfolios with the lowest volatility based on various assets (Brunner and Norouzi 2021). For a rough approximation of the portfolio weights, we may use the following equation:

$$w_{{\mathrm{\Sigma }}_t}=\frac{{\mathrm{\Sigma }}_t^{-1}I}{I{\mathrm{\Sigma }}_t^{-1}I}$$ 14

Modern’s Portfolio Creation (MCP) method is a relatively new development in portfolio construction (2014). Although this method is quite similar to the MVP, it uses conditional correlation minimization rather than conditional covariance minimization to determine portfolio weights. Specifically, this may be broken down as follows,

$$R_t=\mathrm{diag}{({\mathrm{\Sigma }}_t)}^{-0.5}{H}_t\mathrm{diag}{({\mathrm{\Sigma }}_t)}^{-0.5}$$ 15

$$w_{R_t}=\frac{R_t^{-1}I}{I{R}_t^{-1}I}$$ 16

Minimum connectedness portfolio (MCoP)

After constructing the Most Valuable Player and Most Committed Player portfolio strategies, build relying on the connection indices of relations (Ahmadian-Yazdi et al. 2022). Reducing the number of connections between two assets in a portfolio may insulate it from the effects of disruptions in the underlying network. As a result, the created portfolio gives more weight to assets that other factors influence neither impact and the expression of the equation is as follows:

$$w_{C_t}=\frac{{\mathrm{PCI}}_t^{-1}I}{I{\mathrm{PCI}}_t^{-1}I}$$ 17

Portfolio evaluation

We use the Index and the success of our hedges to evaluate portfolio performance (Erumban et al. 2019). Firstly is the formula for determining the Sharpe ratio (SR), commonly known as the benefit to volatility ratio:

$$\mathrm{SR}=\frac{{\overline{r}}_p}{\sqrt{\mathrm{var}(r_p)}}$$ 18

where rp stands for the portfolio returns presuming a risk-free rate of 0%. Since a greater SR number indicates a larger return compared to the portfolio’s risk, the SR may be used to evaluate different assets by revealing which one has (Hotel et al. 2013) the best return for a given amount of volatility.

As for the second statistic, hetero effectiveness (HE) tells us how much less risky the portfolio is compared to investing in just one asset. We use the proposed by Khan et al. (2021a) to determine whether the decrease is statistically significant (Ghorbanpour et al. 2022). It is possible to calculate the HE using,

$${\mathrm{HE}}_i=1-\frac{\mathrm{var}(r_p)}{\mathrm{var}(r_i)}$$ 19

Data

In light of the Chinese green bond issue on August 5th, 2016, our sample period extends until April 28th, 20,221. We do not include green bonds issued by banking sectors or policy banks in our sample since their credit ratings are identical to those of government bonds. Only green bonds issued by corporations and businesses were considered for inclusion in this article. From the China Stock Market & Accounting Research Database (CSMAR), after excluding bonds issued by banks and financial institutions and those with missing data, we followed 124 designated green bonds; data on the bonds’ underwriters and macroeconomic factors came from the Wind database. For the final statistics, we undoubtedly use more filtering processes. First, we do not include observations with maturities of less than 1 month; second, we use Winsor’s 1% and 99% quartiles for the primary explanatory and dependent variables, respectively. As per Ren and Dong (2018), the macroeconomic variables of all items under review had significant volatility besides the green bond, additional volatility clusters (Banerjee et al. 2021).

Empirical results and discussion

The descriptive statistics are summarized in Table 1. When news of the worldwide epidemic broke, all sectors saw sharply declining anomalous yields as seen in Table 1; the high-yield bond industry saw the highest cumulative anomalous yields throughout all periods examined. This was followed by the corporate bond sector, the green bond trade, and the treasury bond industry. The fixed-income industry was hard by the COVID-19 pandemic (Lin 2022). However, government bonds are suggested for those looking to diversify their fixed-income holdings. The green financial instrument’s non-pecuniary feature cannot give additional information to mitigate the effects of significant external shocks (Liu et al. 2022a). This is to the green bond industry reaction is very similar to that of its analogues.

Table 1.

Summary statistics

	GBI	CURI	STI	CEI	OILI
Mean	0.65	0.066	0.643	0.066	0.006
	10.789	3.213	1.223	1.778	0.087
Skewness	− 0.453***	− 0.767***	− 0.887***	− 0.886***	− 1.265***
	(0.000)	(0.000)	(0.007)	(0.000)	(0.000)
Kurtosis	11.665***	15.998***	4.889***	13.8765***	13.887***
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
JB	16,548.660***	20,343.099***	1767.998***	634,678.997***	76,451.886***
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
ERS	− 4.065***	− 6.887***	− 2.8770**	− 3.8876***	− 9.776***
	(0.000)	(0.000)	(0.098)	(0.000)	(0.000)
Q(20)	36.345***	78.993***	75.213***	85.887***	83.7765***
	(0.000)	(0.000)	(0.001)	(0.000)	(0.000)
Q2(20)	180.667***	26.778***	667.889***	2893.665***	2234.889***
	(0.000)	(0.006)	(0.000)	(0.000)	(0.000)
	Unconditional correlations
	GBI	CURI	STI	CEI	OILI
GBI	1.231	0.055	0.066	0.085	− 0.0007
CURI	0.077	1.324	0.987	0.667	− 0.053
STI	0.066	0.768	1.324	0.987	0.008
CEI	0.086	0.889	0.778	1.234	− 0.065
OILI	− 0.0007	− 0.056	0.006	− 0.045	1.043

Investors should use green economic instruments (such as green bonds) as a buffer against market uncertainty (GBI, CEI, STI). In this situation, investors seeking security in the green finance market may be disappointed since neither green nor non-pecuniary properties reduce risk and average dynamic connectedness in Table 2. The Error Correction Model (E.C.M.) exhibits more significant size distortion than procedures, although, for small samples appropriate of macroeconomic time series analysis, the real size of the Co integration treatment regimen testing is steady. Because severe size distortion may not be acceptable, the Granger causality technique is more desirable for its tiny size distortion in finite samples. As a result, the Granger causality test is suited for use in underdeveloped economies, where a time series data for more extended periods is generally lacking the connectedness.

Table 2.

Averaged dynamic connectedness

	GBI	CURI	STI	CEI	OILI
GBI	92.765 (92.564)	2.778 (2.066)	2.765 (2.887)	2.767 (2.687)	0.886 (0.880)
CURI	1.778 (1.4887)	59.770 (59.879)	10.665 (10.234)	27.554 (28.087)	0.887 (0.644)
STI	1.768 (1.786)	11.665 (11.787)	61.987 (62.667)	24.778 (24.876)	0.654 (0.345)
CEI	1.665 (1.987)	24.667 (25.987)	20.987 (20.743)	53.776 (52.98)	0.598 (0.344)
OILI	1.066 (0.775)	1.654 (1.786)	1.875 (1.776)	1.866 (1.776)	94.874 (95.765)

The in-sample prediction was used until about halfway through the trial, and then, an out-of-sample prediction with a lead time was used for the rest of the period (Si et al. 2021). To keep the duration of the estimating window consistent from one cycle of predicting to the next, CURI, OILI the starting data was excluded (Liu et al. 2022b). In the classical theory of economic development, which has been around for more than two centuries, population growth is assumed to lead to increased economic output. However, there is disagreement as to whether population growth has a positive or negative impact on economic growth. This research seeks to examine the nexus by utilizing the whole population devoid of distinctiveness between rural and urban regions that are still unemployed. Even though the dependency ratio in China has been progressively decreasing, many studies do not address the consequences of demographic transition following capital neoclassical growth theory. Evaluate the performance of a TVP-VAR model with four loss functions for forecasting the COVID-19 averaged dynamic connectedness in Table 3.

Table 3.

COVID-19 averaged dynamic connectedness

	CO2	Solar	Wind	Clean energy	Green	From others
GBI	70.654 (69.887)	10.554 (9.778)	6.880 (7.654)	11.778 (9.943)	2.776 (3.887)	31.887 (30.234)
CURI	6.998 (6.054)	42.887 (44.1657)	17.865 (17.432)	31.554 (32.076)	0.554 (0.887)	57.088 (55.808)
STI	4.677 (3.880)	19.876 (18.980)	47.998 (48.900)	27.213 (28.098)	0.76 (0.887)	52.876 (51.166)
CEI	6.087 (5.654)	29.776 (29.778)	23.776 (23.877)	40.887 (40.009)	0.877 (0.556)	59.778 (59.445)
OILI	0.987 (1.778)	2.5876 (1.687)	2.787 (1.998)	2.887 (1.665)	91.987 (93.876)	8.887 (6.787)

Given the sensitivity of volatility forecasting to assessment criteria, this study chose as the primary causes of green bond market volatility those predictors that outperformed a weekly TVP-VAR model without a predictor (the benchmark) across all criteria. Table 4 shows that the volatility prediction effectiveness employing the currency, share, and fixed-income sectors and the Green Building Index as forecasters outperformed the GBI, STI, CURI, and CEI benchmarks across all loss criteria. This result held regardless of the predicting period (Qiao et al. 2022). To a large extent, the above indicators push the green bond market forward. The fixed-income market had the highest median position. The Green Building Index measures the level of activity in the global green construction sector, which influences the volatility of the green bond market. If this is the case, then it follows that green expenditure activities in a country’s economy are connected to the green bond market.

Table 4.

Bilateral hedge ratios

	Mean	Std.Dev	5%	95%	HE	p-value
GBI	0.07	0.22	− 0.08	0.87	0.06	0.87
CURI	0.08	0.06	− 0.08	0.87	0.09	0.87
STI	0.08	0.12	0.00	0.87	0.07	0.54
CEI	0.07	0.00	− 0.08	0.08	0.08	0.98
OILI	0.05	0.76	− 0.77	0.77	0.08	0.33
NGI	0.44	0.98	0.76	0.98	0.76	0.00
ENVI	0.77	0.65	0.65	1.11	0.87	0.00
BUILI	0.00	0.07	− 0.09	0.87	0.08	0.88
TREAI	0.87	0.87	− 0.07	0.87	0.07	0.87
CORI	0.88	0.87	0.87	0.66	0.87	0.00
HYI	0.66	0.76	0.87	1.42	0.53	0.00
VIX	0.00	0.06	− 0.09	0.08	0.08	0.54

A 90 percentage level of confidence was used in the MCS test, and 4000 bootstrap replicates were used for the analysis (Jiang and Yoon 2020). Statistics for just the same outcomes is reported by the other statistics in Table 5. The fixed-income market accepts SSMs more frequently than any other market. The Green Building Index outperforms other volatility prediction indices when the predicting horizon is stretched to green bond, concerning the frequency with which the model emerges in an SBI. There was insufficient data to conclude that any other options were more reliable than those without forecasters regarding weekly R.V. predictions (Huang and Liu 2021). The MCS analysis confirms that the fixed-income sector has had exceptional success, with the Green Building Index coming in second.

Table 5.

Bilateral portfolio weights

	Mean	Std.Dev	5%	95%	HE	p-value
GBI	0.76	0.87	0.18	0.87	0.66	0.00
CURI	0.98	0.87	0.08	0.43	0.76	0.00
STI	0.886	0.22	0.04	0.87	0.32	0.00
CEI	0.00	0.11	0.00	0.07	1.11	0.00
OILI	0.66	0.76	0.55	0.87	0.87	0.00
NGI	0.87	0.45	0.87	0.65	0.65	0.00
ENVI	0.22	0.98	0.00	1.11	0.98	0.00
BUILI	0.08	0.07	0.00	0.07	0.88	0.00
TREAI	0.97	0.65	0.55	0.98	0.87	0.00
CORI	0.67	0.98	0.53	0.65	0.76	0.00
HYI	0.76	0.87	0.07	1.11	0.87	0.00
VIX	0.02	0.08	0.00	0.07	0.77	0.00

The accuracy of the results was increased by using the all-encompassing forecasting test developed by Freyre et al. (2020). According to investigators, whether or not a predicting model includes alternatives based on the importance of a corresponding variable when using a regression test for predicting in Table 6. Previously, per-capita figures were used in other investigations. However, empirical evidence is still few, particularly in green economies. Those who employ an expanded production function framework to analyse the relationship between energy consumption, international commerce, and economic growth in China are the only prominent and similar study in the South Asian countries. According to a unidirectional Granger causation that ties electricity consumption to economic growth, China might increase economic growth by expanding hydropower investments. The study begin by contrasting traditional finance and green economic growth, promoting that green economic growth safeguards the environment while exploring economic progress. Green economic growth, in contrast to traditional finance, emphasizes the benefits of human society living environment, emphasizing the strengthening of natural resource conservation, and thus the economies and human society long-term (Alemzero et al. 2021).

Table 6.

Multivariate portfolio weights

	Minimum variance portfolio
	Mean	Std.Dev	5%	95%	HE	p-value
GBI	0.00	0.00	0.00	0.87	1.11	0.00
CURI	0.02	0.02	0.00	0.06	0.88	0.00
STI	0.02	0.03	0.00	0.07	0.88	0.00
CEI	0.02	0.06	0.00	0.07	0.66	0.00
OILI	0.54	0.04	0.54	0.88	0.08	0.98
	Minimum correlation portfolio
	Mean	Std.Dev	5%	95%	HE	p-value
GBI	0.87	0.07	0.87	0.87	0.76	0.00
CURI	0.33	0.08	0.76	054	0.98	0.00
STI	0.98	0.09	0.76	0.98	0.76	0.00
CEI	0.07	0.07	0.00	0.65	0.98	0.00
OILI	0.98	0.08	0.29	0.87	− 44.77	0.00
	Minimum connectedness portfolio
	Mean	Std.Dev	5%	95%	HE	p-value
GBI	0.88	0.03	0.26	0.32	0.77	0.00
CURI	0.54	0.03	0.16	0.22	0.77	0.00
STI	0.87	0.03	0.17	0.23	0.44	0.00
CEI	0.07	0.03	0.00	0.12	0.88	0.01
OILI	0.87	0.03	0.27	0.33	− 0.59	0.00

The outcomes of the encompassing test for various horizon lengths and sharp ratio are summarized in Table 7. Investment in environmentally friendly buildings and the performance of the currency, treasury bond, and business bond sectors all had statistically important factors independent of the prediction horizon length. While the fixed-income market absorbed the data offered by the remaining forecasters about the forecasts’ lack of importance (Lokhandwala and Gautam 2020), the equities market. According to this situation, there is data in the green bond prediction that cannot be captured by other forecasters, such as green building expenditure and the currency, treasury bond, and business bond markets (Emrouznejad and Yang 2016). The excellent effect of the fixed-income market in anticipating the volatility of green bonds is supported by the findings of the encompassing test.

Table 7.

Sharpe ratio

	MVP	MCP	MCoP
Mean	0.007	0.087	0.088
Std.Dev	0.776	1.076	1.066
SR	0.087	0.087	0.087

Dynamic analysis reveals that key events at certain epochs likely influence connectivity changes throughout time. Indeed, we find that connectivity takes on levels between about 50 percentage points and 84%. This insight may be gained since it suggests that connectivity, which in this case realistically represents the level (De Oliveira-De Jesus 2019) of uncertainty contagion across energy sources, is on the rise due to certain occurrences. For example, connectivity increased at the start of the Great Financial Crisis (GFC) in 2005–2006, continued to climb following the oil price collapse of 2014, and reached new, unprecedented peak levels near the conclusion of our sample period that may be attributed to the COVID-19 outbreak (Kong et al. 2019), who study the frequency connection between COVID-19, crude oil, and the financial market, also highlight the severe effect of the outbreak on the price of crude oil. It is shown that COVID-19 has had a greater effect on crude oil market volatility than the GFC of 2005–2006. Because volatility is a measure of risk, it is clear that the epidemic has significantly increased people’s awareness of the dangers inherent in the energy markets. Taghizadeh-Hesary et al. (2021) also detail the detrimental effects of COVID-19 on the energy industry, providing evidence that the outbreak has had a significant negative influence on both the demand and supply of crude oil. Since there is a tight relationship between the crude oil market and the refined petroleum product market, it may be claimed that the harmful effect of the outbreak on the level of volatility of refined petroleum products should also be large. For example, Petrović-Ranđelović et al. (2020) indicate that GBI contributed significantly to the prediction error variance of both heating oil and gasoline throughout the COVID-19 outbreak period, providing influence of economic structure on carbon neutrality in Fig. 1. The results of the dynamic assessment confirm those of the static assessment and provide new insight into the significance of the dynamic research (Dalheimer et al. 2021) of the interconnections within the particular network. The dynamic TCI effectively captures the extent to which market volatility spreads across the studied network over time.

Fig. 1.

Influence of economic structure on carbon neutrality

Looking at the panels of Yu et al. (2021), it becomes clear that the picture is relatively simple concerning the refined petroleum products’ group; in other words, these marketplaces do not seem to play a new function inside the network as time goes on. GBI was a major net transmitter of market uncertainty up until early 2009, whereas CEI was primarily a net transmitter of market uncertainty in the time that followed. This makes the image of the energy markets of GBI and CEI less clear.

are significant net transmitters in the system because they both have low liquidity, which makes trading in these assets more difficult and spreads increased exposure to losses from those commodities more widely (Barroco and Herrera 2019; Goh and Ang 2019). This is connected to the findings of Chica-Olmo et al. (2020), who highlight that hedgers and investors often seek greater premiums alongside illiquid products. Last but not least, scholars likely have argued that kerosene-reliant enterprises, like airlines, incur substantial illiquidity premia when dealing in derivative markets to check the impact of green bond which is presented in Fig. 2. Therefore, it stands to reason that products with limited liquidity would operate as significant drivers of risk contagion during market turmoil since liquidity seems to be a crucial component shaping attitude toward risk in derivative markets.

Fig. 2.

Impact of green bond and stock return on carbon neutrality approach

Regarding the two crude oil indicators, it seems that the years 2009–2010 were a watershed in the history of the crude oil market. The period following 2007 and up until at least 2016 has been established in the existing research on the WTI-Brent price difference as one in which WTI was trading at a considerable discount on Brent crude oil (Chen et al. 2021). Consequently, significant events in the crude oil market (such as increased oil production in the China, due to the broad use of fracking and horizontal drilling) might account for the shift in the relative importance of the two crude oil indicators around 2008. Some unknowns might lead to a larger risk of loss if either form of crude oil becomes a widespread source of the virus. US export restrictions on WTI were eased until 2016, and even then, there were still logistics to consider, such as transportation limitations and the country’s preexisting pipeline network (International Energy Agency 2019). Concerns about the oil reserves being depleted in recent years are likely a key factor affecting Brent’s price (Khan et al. 2021b).

Now that we have a better idea of what could be driving our findings, we can dig into the dynamics among different pairings of variables in our network (Ang 2015). We examine the dynamics among the network variables to research their connections at greater length. The findings of our investigation are net pairwise connectivity.

Conclusion and policy recommendations

In this study, we suggest using a TVP-VAR strategy to address the threats to financial stability caused by climate change in China. We used daily data beginning in 2016–2021. Specifically, we used the S&P Green Bond Price Index, the Solactive Global Solar Price Index, the Solactive Global Wind Price Index, and the S&P Global Clean Energy and Carbon Price Index as our data set. To probe return spillovers and interconnectedness, we use the TVP-VAR method. We employ several different portfolio techniques to evaluate portfolio performance, including the minimum variance portfolio, the minimum correlation portfolio, and the recently developed minimum connectedness portfolio.

Although situation assessment and stress testing are better than simple voluntary disclosure strategies, they still fall under the purview of a market-correcting structure that sees effective price discovery as the best way to guarantee a smooth transformation and necessitate heroism when compared to the ceteris paribus assumptions that underpin such a structure. Also, these ensures test frameworks have been used up to this point with the hope that market players would take action without any further intervention from policymakers. In order to prevent long-term, possibly disastrous financial and monetary losses caused by physical climate change, this strategy established a bias against the short-term market disruption and heightened transition risk that is required.

A PFP strategy, on the other hand, should make the case for early preventive action and point financial markets in the direction of a net-zero carbon future that is favoured. Our proposal for implementing CRFR includes incorporating it into quantitative credit controls and credit guidelines, capital adequacy requirements, economic policy operations (including asset purchases and collateral standards), and strengthening the financial system’s resilience. In addition to bolstering new dispositions like risk disclosures, benchmarks, and non/sustainable taxonomies, a TVP-VAR structure should help to justify making existing dispositions like these mandatory and standardized in place of the existing voluntary frameworks that a growing number of stakeholders want to make mandatory.

Regulators must be necessary of the potential short-term trade-off between efficiency and resilience and the potential pushback from market players with shorter-term time horizons while taking a cautious approach. Just as policymakers gain insight from the successes and setbacks of macro prudential policy interventions over the previous several decades, so must those operating in this new setting ‘learn by doing’. However, not all preventative measures work. However, we argue that non-interventionist analysis, modelling, and forecasting miss out on more useful information than learning from interventions and examining the (endogenous) responses that follow a specific intervention.

Reorienting the thought processes of monetary policymakers is neither a simple nor a quick endeavour, given that some central bankers and regulators are questioning their mandates and what more can do in the face of obvious market failures that threaten to slow down the progress toward decarbonization objectives.

Rather than providing a one-size-fits-all approach to financial regulation, this study serves as an investigation that aims to provide a new strategy architecture for dealing with CRFR. In this study, we have only touched on the issue superficially; future scholarship might benefit from more in-depth studies of the available tools and strategies. While this paper has focused primarily on central lenders and economic managers, other state agencies, such as the ministries of finance, commercial strategy, and other public investment firms, will also have a role to play in coordinating their policies with financial supervisors, particularly during times of crisis like the COVID-19 pandemic.

The timetable of the policy choices that this study is meant to inform must be compatible with the practicality of the study, especially when it comes. This is particularly true when working with circumstances characterized by extreme uncertainty, which severely restricts the applicability of findings from TVP-VAR methods.

We expect that the TVP-VAR approach presented in this research may be used to address additional complex ecological concerns that are marked by extreme uncertainty, such as the loss of biodiversity, the contamination of water and air, and the depletion of natural resources, in addition to climate change. Almost all of these spheres have substantial linkages with global warming and should be included in CRFR strategy frameworks whenever practicable.

Author contribution

Yi Wang: conceptualization, methodology, software, data curation, writing — original draft. Lei Chang: writing — review and editing, visualization, investigation.

Data availability

The data that support the findings of this study are openly available on request.

Declarations

Ethical approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

All authors reviewed and approved the manuscript for publication.

Preprint service

Our manuscript is not posted at a preprint server prior to submission.

Conflict of interest

The authors declare no competing interests.