1 s2.0 S0040162524005134 Main

Technological Forecasting & Social Change 208 (2024) 123715
Contents lists available at ScienceDirect
Technological Forecasting & Social Change

journal homepage: www.elsevier.com/locate/techfore
Analyzing the interplay between eco-friendly and Islamic digital currencies

and green investments
Mahdi Ghaemi Asl a,* , Sami Ben Jabeur b,c , Younes Ben Zaied d
a
Kharazmi University, Tehran, Iran
b
UCLy (Lyon Catholic University), ESDES, Lyon, France
c
UCLy (Lyon Catholic University), UR CONFLUENCE: Sciences et Humanités (EA1598), Lyon, France
d
OCRE, EDC Paris Business School, Paris, France
A R T I C L E I N F O A B S T R A C T
Keywords: This study aimed to provide a comprehensive understanding of the interplay between digital innovations and
Data analytics sustainable investments, specifically regarding their roles in promoting economic resilience and sustainability in
Eco-friendly the energy sector. To this end, the present study examined the possible consequences of the widespread use of
Green investment
energy-intensive crypto-asset technologies, focusing on those that rely heavily on non-renewable energy sources.
Islamic digital currency
The analytical results demonstrated that the interconnectedness between cryptocurrency-based digital in
novations and sustainable investments was most pronounced in bull markets characterized by high returns and
upper tails, except for the Dash Green market. Bitcoin and Stellar exhibited the strongest connections with eco-
friendly financial ventures across various market states, indicating their robust interplay with sustainable
portfolios. The total connectedness index using an Elman neural network revealed that higher non-linear dy
namic information was associated with green-synthesized cryptocurrency-driven innovations, compared with
other digital innovations. This study contributes to scholarly understanding of the interconnectedness between
cryptocurrency-based digital innovations and sustainable investments by elucidating their dynamics under
different market conditions. These findings underscore the need for further analysis to fully comprehend the
intricate relationship between these assets, particularly across different quantile levels, as well as to inform sound
investment decisions in sustainable finance.
1. Introduction the carbon impact of blockchain technology. Many of these elements

may also be applied to current and increasingly polluting crypto
Environmental sustainability refers to the preservation of natural currencies in the quest for sustainable alternatives. Although it is diffi
resources, which is essential for supporting social and economic well- cult to add sustainable quality to the delicate equilibrium among
being (Wang and Xu, 2024). It lays a solid foundation for effective decentralization, security, and scalability that cryptocurrencies strive to
natural resource management, pollution control, and the conservation achieve, experts have identified some essential factors for greening
of critical elements such as water, land, air, and biodiversity (Goodland, cryptocurrencies, such as the transition to renewable energy.
1995; Moldan et al., 2012). In recent years, governments worldwide Achieving sustainable economic progress while simultaneously pre
have paid increasing attention to the environmental implications and serving the environment requires the widespread adoption of eco-
energy consumption of cryptocurrencies, such as Bitcoin. According to friendly and clean technologies. Sustainable growth requires improved
De Vries (2023), the electric power consumption of the Bitcoin network understanding, technological advancement, and adaptability in the
may approach 13 GW, with a carbon footprint of 65 megatons of CO2 transition to renewable energy sources (Yu et al., 2022). According to
annually. Considering the ongoing climate emergency and worldwide Cao et al. (2021), the current global financial system’s limitations have
energy crisis, regulatory bodies have considered measures to restrict the heightened emphasis on the digital revolution within the financial
power consumption of cryptocurrency networks. Meanwhile, emerging sector. The advent of technologies such as blockchain technology, social
green cryptocurrencies have been built on novel processes that reduce networking, big data analytics, and artificial intelligence has enriched
* Corresponding author at: Faculty of Economics, Kharazmi University, No. 43, Mofatteh Ave., Tehran, Iran.
E-mail addresses: [email protected] (M. Ghaemi Asl), [email protected] (S. Ben Jabeur), [email protected] (Y. Ben Zaied).
https://doi.org/10.1016/j.techfore.2024.123715
Received 9 June 2023; Received in revised form 11 August 2024; Accepted 26 August 2024
Available online 5 September 2024
0040-1625/© 2024 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
M. Ghaemi Asl et al. Technological Forecasting & Social Change 208 (2024) 123715
the financial industry. These advancements offer lower costs, superior in the current global climate and technological paradigm, the present
services, reduced information asymmetry, increased transparency, and study aims to clarify the intricate connections between various
more diverse and resilient systems. Considering these advancements in cryptocurrency-driven innovations and sustainable investment. We
the financial sector, digital finance can offer solutions to the funding explore this persistent memory of connectedness to understand how it
challenges encountered by the new energy sector. Specifically, the dig influences persistent memory and learning of long-term dependencies
ital revolution, including the rise of third-party financing, could present (Asl and Jabeur, 2024). To this end, we employ the Elman network
innovative and effective strategies to address the financial hurdles faced methodology, which is well-suited for modeling such dependencies due
by solar power businesses and the new energy sector (Li et al., 2023; Sun to its recurrent structure. As documented by Elman, neural networks
et al., 2023). possess the capability of short-term memory and become a good option
Moreover, it is imperative to consider the emerging growth patterns for improving overall accuracy.
of Islamic finance. According to a 2015 comprehensive report by The Compared to previous research, this study makes the following
Banker, the Islamic finance sector experienced a remarkable surge, with contributions: First, it explores the impact of green, green-synthesized,
an annual growth rate of 12.7 %. Notably, several European Union- Islamic, and conventional digital cryptocurrencies on clean energy.
member countries were among the top 20 nations that demonstrated Our study provides valuable insights for the burgeoning literature on
adherence to Islamic Shariah principles in offering Islamic financial eco-innovation (Cainelli et al., 2020; Albitar et al., 2023; Xu et al., 2023;
services. Furthermore, the Islamic Financial Services Board’s 2019 Chai et al., 2022). According to Valero-Gil et al. (2023), eco-innovation
report highlighted the substantial value of the Islamic finance and differs from general innovation by prioritizing sustainable development
banking industry, which amounted to $2.19 trillion in the second and being motivated by environmental considerations. The primary
quarter of 2018, compared with $2.05 trillion in 2017 (Hassanein and objective of eco-innovation is to enhance environmental sustainability,
Mostafa, 2023). Notably, certain alternative cryptocurrencies, such as which encompasses reducing environmental burdens and achieving
Stellar, have obtained Sharia compliance certificates (Taghdiri, 2020). sustainability goals, as defined by environmental parameters. The
Furthermore, extensive scholarly research has been conducted on the adoption of green technological innovation is an intrinsic factor that
Islamic X8X Token, as evidenced by studies conducted by Ali et al. drives sustainable economic growth and plays a pivotal role in shaping
(2022), Mnif et al. (2022), Yousaf and Yarovaya (2022), Wasiuzzaman the quality of economic development (Ghisetti and Quatraro, 2017;
et al. (2023), and Ali et al. (2023). These studies highlight the signifi Zhang, 2023). Our research expands the existing body of knowledge by
cance of X8X Token in the context of Islamic finance, emphasizing its enhancing the connection between digital finance and green innovation
potential role as a prominent instrument in the rapidly evolving cryp through the incorporation of clean energy performance into the con
tocurrency market. Therefore, Islamic digital assets, as burgeoning and ceptual framework. Second, our findings enrich existing scholarly
rapidly expanding financial instruments, have substantial potential in knowledge by adding further evidence of the link between digital
the cryptocurrency market. This potential is further underscored by the finance and green innovation. This was achieved by integrating clean-
integration of Islam’s environmental ethics into contemporary Islamic energy performance into a conceptual framework. Third, we use the
finance (Moghul and Safar-Aly, 2014), as well as its pivotal role in novel quantile vector autoregressive (QVAR) extended joint connect
fostering circular economy practices, promoting green economy initia edness framework proposed by Cunado et al. (2023), which captures the
tives, and contributing to sustainable development (Rashid and Siddi connectedness of volatility by categorizing it into low- and high-
que, 2021). volatility stages. Cunado et al. (2023) innovatively merge the cor
This study aims to provide some insight for the interplay between rected quantile connectedness approach proposed by Chatziantoniou
digital currencies and green investments. To this end, the present study et al. (2021) with the expanded joint connectedness framework intro
examines the possible consequences of the widespread use of energy- duced by Balcilar et al. (2021). This framework incorporates a refined
intensive crypto-asset technologies, focusing on those that rely heavily connectedness normalization technique, leading to more precise results.
on non-renewable energy sources. Such technologies might significantly Moreover, we use an Elman neural network (Elman, 1990) to analyze
thwart a nation’s efforts to fulfill its nationally determined contribu the non-linear information contained in the connectedness of green,
tions, as outlined in the Paris Agreement. This is because of the potential green-synthesized, Islamic, and conventional cryptocurrencies with
for these technologies to exacerbate the carbon footprint and hinder green investments. The Elman neural network, which is a variant of
progress toward reaching ecological equilibrium. According to Huang recurrent neural networks, identifies sequential dependencies, making it
et al. (2022), eco-innovation boosts economic growth and expedites the particularly suitable for time-series analyses (Ruiz et al., 2018). This
transition from non-renewable to clean energy, thereby serving as a method has been applied in multiple fields, including finance and eco
beneficial and effective policy tool. Consequently, eco-innovation fos nomics, and has produced successful prediction outcomes (e.g, (Xu
ters the conversion of non-renewable energy sources into cleaner al et al., 2024; Jiang et al., 2022)). In fact, we propose a new hybrid
ternatives. For example, the expansion of energy-intensive crypto-asset artificial intelligence (which we describe as an efficient fusion of
technologies, particularly when they rely on non-renewable energy computational frameworks) that outperforms or is more competitive
sources, could potentially limit a country’s ability to meet its nationally than straightforward conventional techniques. The popularity of hybrid
determined contributions under the Paris Agreement. Despite significant intelligent systems has increased in recent years due to their ability to
advances in sustainability research (Pham et al., 2023), a critical gap effectively address complex real-world problems (Medsker, 2012;
remains in exploring the relationships and intersections between cryp Jabeur and Serret, 2023). As each method has its benefits and draw
tocurrency technologies and sustainable investments, which is crucial backs, it is necessary to develop hybrid models that integrate the ad
because any potential incongruence between them could pose chal vantages of many approaches while simultaneously addressing the
lenges in effectively mitigating climate change. As digital currencies limitations of each approach and leveraging its strengths. Therefore, this
continue to evolve, what approach should investors use when building a study aims to leverage the unique capability of the QVAR method to
portfolio of these innovative assets? Furthermore, what is the correla model the connectedness between variables (Ando et al., 2022) with the
tion between environmentally friendly financial investments and the goal of enhancing the performance of the Elman neural network. This
success of digital currencies? Does including environmentally sustain integration is expected to create a synergy that significantly boosts this
able assets in a digital currency portfolio provide any advantages in network’s capacity to make accurate and robust predictions across our
terms of diversification? What are the consequences for policymakers in database. This network can be trained to delineate the relationships
their efforts to allocate more financial resources toward sustainable between various groups of cryptocurrencies and green investments over
economic activities? To fill this gap in the literature and contribute to a time. It can detect and quantify non-linear information or patterns
larger discussion on the sustainable trajectory of blockchain technology within data. Based on our rigorous scientific methodology, which
2
discerns the interconnectedness of volatility under varying market speed of innovative ideas’ and technology’s propagation within a social
conditions, we offer precise insights into the relationship between dig framework. Based on this theory, our study extends previous research by
ital assets and sustainable investment. This constitutes a significant analyzing the nuanced factors that influence the adoption process across
enhancement to the existing body of knowledge, contributing both different industries. Corbet et al. (2021) analyze how the volatility of
methodologically and substantively to the empirical evidence. Bitcoin’s price and the underlying dynamics of cryptocurrency mining
Furthermore, the findings of this study provide important insights. characteristics impact energy markets and utility companies. The au
First, the analysis demonstrates that the interconnectedness between thors underscore the need for further assessment of the environmental
cryptocurrency-based digital innovations and sustainable investments is impacts of cryptocurrency growth. Further, they demonstrate the sus
most pronounced in bull markets characterized by high returns and tained and significant influence of Bitcoin’s energy usage on the per
upper tails, except for the dash-green market. This finding suggests that formance of specific companies in the energy sector, differentiated by
extreme market conditions intensify the relationship between digital jurisdiction. Bouteska et al. (2023) examine the volatility patterns and
assets and green investments. Second, Bitcoin and Stellar exhibited the interconnections among cryptocurrencies, US equity, and bond markets,
strongest connections with eco-friendly financial ventures across as well as the impact of COVID-19. They reveal return volatility spill
various market states, indicating their robust interplay with sustainable overs and emphasize the influence of major global events on crypto
portfolios. The type of technology employed, whether proof-of-work or currency prices. Their findings can guide fund managers and
proof-stake, did not significantly alter the strength of cryptocurrency- policymakers in implementing timely interventions during unforeseen
driven technologies or green investment opportunities. This finding crises. Moreover, Li et al. (2019) reveal that blockchain applications
implies that market capitalization and unique attributes play a more have the potential to revolutionize the renewable energy industry.
influential role in linking cryptocurrencies to climate-focused strategies, Meanwhile, according to Dogan et al. (2022), the nature of the energy
compared with their inherent climate-aware characteristics. Addition sources used in cryptocurrency mining plays a crucial role in assessing
ally, this study highlights the limited association between Bitcoin Green its environmental impact. The use of renewable energy in this process is
(BITG) and Dash Green (DASHG) (green-synthesized cryptocurrency- significant, as it aids in the transition toward an energy supply that is not
driven innovations) and eco-friendly investment approaches, possibly only cost-effective and secure, but also environmentally friendly. This
due to their perceptions and reputations within the investment com transition is essential for mitigating environmental concerns associated
munity. Moreover, this study contributes novel findings regarding the with cryptocurrency mining, thereby promoting sustainable practices in
Total Connectedness Index (TCI) using an Elman neural network. The the burgeoning industry. Additionally, Ye et al. (2023a, 2023b) explore
results demonstrate a higher non-linear dynamic information associated the impact of cryptocurrencies and blockchain technology on green in
with green-synthesized cryptocurrency-driven innovations, compared vestment and environmental sustainability in the USA from 2011 to
with other digital innovations. This suggests that the eco-conscious or 2020 using the non-linear autoregressive distributive lag technique.
Shariah-compliant nature of cryptocurrency-driven innovations does They identify an asymmetric relationship between cryptocurrencies and
not significantly alter the nature and structure of their linkages with biofuel and renewable energy usage, thus indicating that the traditional
green investment initiatives. assumption of a symmetric impact is misleading. Sharif et al. (2023)
Importantly, our findings diverge from those of recent studies by investigate the relationship between green economy indices and cryp
highlighting the complexity and variability of the relationship between tocurrencies, distinguishing between “black” (energy-intensive) and
cryptocurrency assets and sustainable investments. This study provides “clean” (energy-efficient) cryptocurrencies in the US, European, and
further evidence supporting the notion that cryptocurrencies, as a Asian markets using the quantile spillover index method. Their findings
nascent asset class, have significant potential as valuable diversification show strong connections between green economy indices and clean
tools for sustainable portfolios. These findings align with previous cryptocurrencies, with a notable increase in spillover effects since the
research that emphasizes the low correlation and relative isolation be onset of the COVID-19 pandemic, particularly in Asia, thus highlighting
tween cryptocurrencies and other financial assets. This study contrib clean cryptocurrencies’ diversification benefits for green economies.
utes to the understanding of the interconnectedness between Moreover, Wang et al. (2023a, 2023b) examine consumer behavior to
cryptocurrency-based digital innovations and sustainable investments ward central bank digital currencies using data from 1308 Chinese re
by shedding light on their dynamics under different market conditions. spondents to assess their willingness to pay and the factors influencing
These findings underscore the need for further exploration and analysis it. The authors find that willingness to pay is influenced by positive
to fully comprehend the intricate relationship between these assets, perceptions, consumer traits, and the value placed on social benefits, but
particularly across different quantile levels, and to inform sound in not infrastructure views, thereby highlighting the role of consumer data
vestment decisions in sustainable finance. in developing central bank digital currency policies.
The remainder of this paper is organized as follows: Section 2 pre In this context, the escalating environmental issues associated with
sents an extensive review of the existing literature to contextualize the Bitcoin mining have led to intense debates among investors (Naeem and
current study within a broader research landscape. Section 3 describes Karim, 2021). Under these circumstances, they must decide whether to
the methodologies employed and elucidates the materials used in this capitalize on the financial advantages of Bitcoin investments or pursue
study. Building on this foundation, Section 4 presents the empirical eco-friendly investment options for risk diversification. Despite the
findings and offers an in-depth analysis and discussion. Section 5 checks significant growth in the latter market in recent years, there has been
the robustness of the results, while Section 6 presents the concluding limited focus on the relationship between cryptocurrencies and green
remarks, encapsulates the key insights derived from the study, and markets. This growth is particularly evident in the adoption of sustain
provides valuable policy implications. able alternatives to traditional carbon-intensive energy sources (Arfaoui
et al., 2023). For instance, Symitsi and Chalvatzis (2018) investigate the
2. Literature review shock spillovers between Bitcoin and the stock indices of clean energy,
fossil fuel energy, and technology companies. Their findings indicate
Given the significant energy consumption associated with crypto that both energy and technology stocks significantly influence Bitcoin’s
currency markets, it is crucial to examine their interplay with clean returns. In the short term, volatility in technology companies tends to
energy and carbon emission permits. This is crucial because the affect Bitcoin; conversely, in the long term, Bitcoin’s volatility appears
expansion of cryptocurrency markets is projected to persist (Dogan to affect energy companies.
et al., 2022; Bouchmel et al., 2024; Tian et al., 2024; Schilling et al., Moreover, Ren and Lucey (2022a) examine the hedging and safe-
2024). The goal of innovation diffusion theory is to comprehend the haven properties of numerous clean energy indices against two
mechanisms (Rogers et al., 2014; Wani and Ali, 2015), reasons, and distinct types of cryptocurrencies, categorized as “dirty” and “clean,”
3
based on their respective energy-consumption levels. “Dirty crypto emissions. This emphasizes the crucial role of creative financing
currencies” primarily utilize energy-intensive proof-of-work consensus methods in improving the sustainability of mineral extraction. (Hung,
mechanisms that rely on non-renewable energy sources, thus contrib 2023) investigates the promotion of economic sustainability in Vietnam
uting significantly to carbon emissions and environmental degradation through green investment, digitalization, and financial development by
(Ren and Lucey, 2022a). Conversely, “clean cryptocurrencies” reduce employing a novel three-stage methodology. He finds that these factors
their impact on the environment by using more energy-efficient strongly impact Vietnam’s economic sustainability, thus indicating their
consensus processes (e.g., proof of stake and delegated proof of stake) crucial role in supporting the country’s shift toward sustainable devel
or running on platforms that primarily utilize renewable energy sources. opment. These results offer a roadmap for developing countries to
Their statistical analysis reveals that none of these groups benefit leverage technological innovation, green investment, and financial
directly from renewable energy sources. However, in extremely adverse growth to achieve development goals.
markets, both types offer a certain degree of protection, albeit minimal. However, there is a notable lack of studies exploring the major dig
Furthermore, during periods of heightened uncertainty, “dirty” crypto ital currency markets, green and Islamic digital currencies, and clean
currencies are more likely to act as safe havens, compared with their technology markets within a unified framework. Therefore, the present
“clean” counterparts. Moreover, Sharif et al. (2023) explored the in study offers insights into the relationship between cryptocurrencies and
terconnections and spillover effects between green economy indices, environmentally sustainable assets by focusing specifically on green and
five “dirty” cryptocurrencies, and five “clean” cryptocurrencies within Islamic cryptocurrencies and clean energy technologies.
the US, European, and Asian markets. They found that the connection is In light of the reviewed literature, the following hypotheses will be
stronger between green economy indices and clean cryptocurrencies, rigorously tested:
compared with their “dirty” counterparts. Furthermore, green economy
indices generally display net receiving behavior, whereas crypto 1. There is a significant interconnection between eco-friendly digital
currencies fluctuate based on various factors such as variables, quan cryptocurrencies (such as Bitcoin green and Dash green) and green
tiles, and time periods. Patel et al. (2023) investigate the spillover effects investments.
that occur between green “dirty” cryptocurrencies and socially respon 2. Islamic digital cryptocurrencies exhibit a distinct pattern of inter
sible investments in the context of the Russo-Ukrainian conflict. Their action with green investments compared to conventional and eco-
results show that the level of connectedness exhibits significant shifts friendly digital cryptocurrencies.
between the pre- and post-war scenarios. Throughout the cycle, the 3. The market capitalization and established position in the FinTech
magnitude of the spillovers and the role of each cryptocurrency and industry of certain cryptocurrencies (e.g., Bitcoin and Stellar)
socially responsible investment change. Pham (2019) reveals that correlate with their high interconnectivity with green investments.
incorporating green investments into cryptocurrency portfolios offers 4. Nonlinear analysis methods, such as the Elman neural network, can
diversification benefits. This is attributed to the minimal correlation effectively capture the time-varying dynamics and accurate in
observed between cryptocurrencies such as Bitcoin and Ethereum and terconnections between digital cryptocurrencies and green
green assets during non-crisis periods. investments.
However, progress in financial technology can bolster the integration 5. Green digital innovations show stronger associations with indices
of sustainable energy sources into comprehensive energy portfolios tracking renewable energy and clean technology compared to their
(Zhao et al., 2024). This underscores the need for new incentives to conventional counterparts.
produce and use renewable energy (Anwer et al., 2023). Furthermore, it 6. Cryptocurrencies employing energy-efficient consensus mechanisms
facilitates the development of tools to manage complex power grids such as proof-of-stake are more closely aligned with environmental
consisting of various small-scale consumers, while introducing innova performance metrics than those using more energy-intensive
tive risk-management strategies. Numerous studies explore the impli algorithms
cations of sustainable assets in the energy industry (Tiwari et al., 2022; 7. Implementing environmental synthesizing processes within crypto
Tiwari et al., 2023). For example, Chatziantoniou et al. (2022) examine currency platforms can enhance their eco-friendly attributes and
the dynamic integration and return transmission of four established promote sustainable digital asset ecosystems.
environmental financial indices. They find that the MSCI Global Envi
ronment is a significant shock transmitter. Consequently, strategies 3. Methods and materials
aimed at promoting the expansion of green finance markets should
consider the considerable influence of eco-friendly assets. Moreover, The ENN-QVAR model is an innovative and powerful approach
given the minor correlation between green bonds and clean energy designed specifically for analyzing multivariate time series data. This
stocks, as well as their vulnerability to external shocks, it is prudent for model combines the strengths of two main components. The procedure
policymakers to treat these markets separately when devising policy to construct this model involves several steps. First, the ENN-QVAR
strategies. However, there has been growing interest in understanding model starts by estimating the QVAR model. This step involves
how these Islamic assets are connected to clean energy investments extracting non-linear dynamic information coefficients from the multi
(Abakah et al., 2023a, 2023b; Rizvi et al., 2022; Anwer et al., 2023). variate time series data. These coefficients capture the relationships and
This is likely because of the increasing importance of sustainable and dependencies among the variables in the data. Second, the estimated
ethical investments in global financial markets. QVAR model is decomposed into the joint total connectedness index.
Several studies investigate digital innovation and sustainable in The joint total connectedness index represents the sum of the joint time-
vestment (Dabbous et al., 2024; Yang et al., 2024; Wang et al., 2023a, varying connectedness between each pair of variables in the data. This
2023b). Ye et al. (2023a, 2023b) examine the impact of digital invest index provides insights into the overall interconnectedness and in
ment on environmental performance in 273 Chinese pollution-intensive terdependencies among the variables. Third, the QVAR outputs are used
firms from 2016 to 2020 and find that digital investment enhances to prepare input sequences for an Elman Neural Network (ENN). The
environmental outcomes through improved production efficiency and ENN is a type of recurrent neural network that can effectively capture
green innovation. These benefits are significant in state-owned and temporal dependencies in sequential data. In this step, the QVAR out
heavily industrialized firms, with efficiency and innovation working as puts serve as the basis for constructing input sequences that will be used
essential mediators. He et al. (2024) explore the influence of financial to train the ENN model. Fourth, the ENN model is trained on the pre
innovation and green investment on sustainable mineral extraction in pared input sequences. By leveraging the connectedness information
China from 2010 to 2021. They argue that implementing green inno extracted from the QVAR model, the ENN model learns the dynamic
vation and investment leads to a considerable decrease in carbon information and patterns present in the data. The training process allows
4
the ENN to adapt and optimize its parameters to accurately capture the Subsequently, to illustrate the effect of a change in variable j on
underlying dynamics of the time series. Fifth, the trained ENN model is variable i, we computed the h-step ahead Generalized Forecast Error
used to make predictions. By utilizing the learned dynamic information, Variance Decomposition (GFEVD), as proposed by (Koop et al., 1996)
the ENN can forecast future values or behaviors of the variables in the and (Pesaran and Shin, 1998).
multivariate time series data (Xu et al., 2024). Finally, the performance ∑ ( )2
of the ENN-QVAR model is evaluated using appropriate metrics such as σ (τ)− 1 H− 1 eʹ Aj (τ)σ(τ)ej
ωgij (H) = ∑iiH− 1 (h=0 i ) (4)
RMSE (Root Mean Square Error) and MSE (Mean Squared Error). These h=0 ei Aj (τ)σ (τ)Aj ei
ʹ ʹ
metrics provide a quantitative measure of the accuracy and effectiveness
of the model’s predictions compared to the actual values. At horizon H, ωgij (H) represents the jth contribution share of the
In summary, the ENN-QVAR model as a method which has two level variance of the ith variable’s prediction error. σ is the variance matrix of
of analysis, combines the strengths of the QVAR model and the ENN to the vector of errors, and ei is a vector where the ith element has a value
extract non-linear dynamic information and capture temporal de of one and the remaining elements have a value of zero.
pendencies in time series data. The main distinctiveness and superiority The variance decomposition matrix was then normalized as follows
of QVAR over its competitor approaches is that it could measure the for each entry:
dynamic connectedness for different market conditions, from extreme
bear markets to the most bull market states. The key advantages and ωg (H)
̃ gij (H) = ∑k ij g
ω (5)
unique features of the ENN approach lie in its ability to model and j=1 ωij (H)
capture temporal dynamics and sequential patterns in data through its
The GFEVD was subsequently employed to generate four connect
recurrent connections from the hidden layer to a context layer. This
edness measures at each quantile. Cross-variances, which represent the
architecture enables ENNs to remember past inputs, handle non-linear
percentages of h-step-ahead error variances in forecasting yi due to
information, and dynamically adapt to changes in input sequences,
shocks in yj, are summed to form the Total Risk Spillover Index (TRSI).
making them highly effective for time series prediction, speech recog
nition, and real-time data analysis. Their non-linear processing capa ∑k ∑k
̃ gij (τ)
i=1 =j ω
j=1,i∕
bilities and memory of past inputs lead to more accurate and reliable TRSI = ∑k ∑k × 100 (6)
̃ gij (τ)
j=1, ω
predictions compared to traditional feedforward neural networks, i=1
making ENNs a versatile and powerful tool for various applications The Total Directional Connectedness (TO) was calculated to assess
involving sequential and temporal data. the comprehensive impact that variable i exerted on all other variables j:
∑k
̃ gji (τ)
=j ω
3.1. QVAR method TO = ∑k
j=1,i∕
× 100 (7)
̃ gji (τ)
j=1, ω
In our study, we used the quantile connectedness approach proposed
The impact of startling all other variables j on variable i represented
by Ando et al. (2022) as well as by Chatziantoniou et al. (2021) to
by “FROM” at quantile was then evaluated using the total directional
examine the quantile propagation mechanism across green, green-
connection, as shown below.
synthesized, Islamic, conventional cryptos and green investments. To
∑k
calculate all connectedness measures, we first estimate the Quantile ̃ gij (τ)
=j ω
j=1,i∕
Vector Autoregression (QVAR(p)), which can be articulated in the FROM = ∑k × 100 (8)
̃ gij (τ)
j=1, ω
following manner:
p
∑ The Net Total Directional Connectedness (NSI), which can be
yt = ϴ(τ) + Փj (τ)yt− j + νt (τ) (1) perceived as the net effect variable i has on the examined network, is
j=1 derived from the difference between Total Directional Connectedness
(TO) and Total Directional Connectedness (FROM). The subsequent
where τ ∈ (0,1) symbolizes quantile index, ϴ(τ) and νt (τ) denote n- formula was utilized to obtain the Net Total Directional Spillover Index
vector of constants and residuals at quantile τ, respectively. Meanwhile, (NSI) at a given quantile:
Փj (τ) represents the matrix consisting of coefficients that are lagged, at
quantile τ. The τth conditional quantile of the dependent variable y in NSIi (τ) = TOi (τ) − FROMi (τ) (9)
the population is formulated as follows:
3.2. QVAR extended joint connectedness approach
p
∑
Qτ (yt ) = γ(τ) + ̂
Փj (τ)yt− j (2)
j=1 The novel extended joint connectedness technique and QVAR models
are used to create a unique connectedness framework in this research. In
The mean-based metrics proposed by (Diebold and Yilmaz, 2012) particular, the revolutionary change that distinguishes the joint from the
were extended by Ando et al. (2022) to compute multiple measures of original connectedness is the fact that the normalizing approach is taken
return connectivity for each quantile. According to Chatziantoniou et al. from the widely used R2 goodness-of-fit measure. FROMjnt represents
(2021), Eq. (2) can be expressed as an infinite order Vector Moving the influence that each variable within the network exerts on series i.
Average (MA) process, based on Wold’s theorem: This may be expressed numerically as follows:
∑ ∑H− 1 ( ʹ ∑ ( ʹ ∑ ʹ )− 1 ʹ ∑ ʹ )
∞
yt = c(τ) + Aj (τ)νt− i (3) h=0 ei Aj Mi Mi Mi Mi Aj ei
jnt
i=1 FROM = ∑H− 1 ( ʹ ∑ ʹ ) (10)
h=0 ei Aj Aj ei
( )− 1
where c(τ) = In + Փ1 (τ) + … + Փp (τ) ϴ(τ),
⎧ where Mi is K × K-1 matrix that has the same values as the identity
⎨ 0, j < 0
matrix, except for the ith column, Σ ∀ ∕ = i denotes the K-1 dimensional
Aj (τ) = In , j = 0
⎩
Փ1 (τ)Aj− 1 + … + ՓP (τ)AP− 1 , j > 0 vector of shocks at time t + 1 for all series, excluding series i.
The joint total connectedness index is then calculated in the next
step:
5
1∑ k Elman networks are constructed using the backpropagation (BP)

JTO = FROMjnt (11) method. However, the storage and delay characteristics of the context
K i=1
layer inherently connect the output of the hidden layer to its input (Jia
As highlighted by Chatziantoniou et al. (2021), this value ranges and Taheri, 2021). This method of connection is attuned to the neural
between zero and one, contrasting with the TCI of the originally sug network’s historical data. As a result, the internal feedback mechanism
gested method. enhances the network’s ability to process dynamic and complex infor
An essential extension of Balcilar et al. (2021) is that multiple scaling mation more effectively.
factors are used to link TRSI to JTRSI: Recurrent Neural Networks (RNNs) have demonstrated notable ef
ficacy in evaluating the intrinsic nonlinear patterns present in time se
FROMjnt ries data, particularly in tasks involving sequential data. This
φi (H) = (12)
FROM effectiveness stems from their unique capacity to retain historical input
JTRSI = φi (H)TRSI (13) information through recurrent connections. Elman Neural Networks
(Elman NNs), originally proposed by Elman (1990), stand out for their
The total directional connectedness from variable i to all others, the ability to capture and utilize temporal dependencies within sequential
net total directional connectedness, and even the net pairwise direc data, making them a well-suited choice for analyzing time series data. A
tional connectedness may all be calculated by using this equivalence as notable advantage of Elman NNs is their utilization of nonlinear acti
follows: vation functions, such as the sigmoid or hyperbolic tangent, within their
k
concealed layers. These nonlinear activation functions enable the
∑
JTO = JTRSIij (14) effective modeling of intricate relationships intrinsic to the time series
j=1,i∕
=j (Rumelhart et al., 1986). The recurrent architecture inherent in Elman
NNs plays a pivotal role in preserving historical observations, which is
JNSIi (τ) = JTOi (τ) − JFROMi (τ) (15) crucial for apprehending and assessing nonlinear temporal motifs
embedded within the time series (Jordan, 1986). During the training
3.3. Elman neural network (ENN) based QVAR model phase, Elman NNs dynamically adapt their internal weights and biases
through optimization processes aimed at minimizing prediction dis
The Elman Neural Network (ENN), introduced by Elman in 1990, is a crepancies. This adaptability allows them to acquire and articulate the
partially recurrent network consisting of four layers: an input layer, a multifaceted nonlinear attributes and motifs that underlie the time se
hidden layer, an intermediate layer, and an output layer, as depicted in ries data (Werbos, 1988). Additionally, Elman NNs employ a dynamic
Fig. 1. Like a feedforward network, there is a connection between the adjustment of model intricacy throughout training, thereby ensuring the
input layer, hidden layer, and output layer. The input layer’s sole capture of pertinent nonlinear motifs and the optimization of their
function is signal transmission. The output layer acts as the linear predictive potentialities (Hochreiter and Schmidhuber, 1997). To
weighted component. The hidden layer, situated between the input and enhance their generalization capacity while preventing overfitting,
output layers, serves as a relationship map. The intermediate layer is Elman NNs can assimilate regularization methodologies such as dropout
utilized to retain the previous output of the hidden layer. Consequently, (Srivastava et al., 2014) or L2 regularization (Hastie et al., 2009). These
the Elman network can manage dynamic information (Yongchun, 2010). techniques effectively preserve the proficiency of Elman NNs in utilizing
The ENN with n inputs, h hidden neurons, and δ outputs are defined by and assimilating nonlinear information while ensuring optimal perfor
the following equations: mance in capturing the complexities of the time series data. It should be
h ∑
h n ∑
h
mention that linear information is characterized by a direct and pro
∑ ∑
Sj (t) = f γjr Sr (t − 1) + Vji Xi (t) (16) portional relationship between variables, where changes in one variable
r=1 j=1 i=1 j=1 produce consistent and proportional adjustments in another variable,
( ) whereas nonlinear information exhibits complex and non-proportional
h ∑
∑ δ connections between variables, leading to intricate and non-linear al
δk (t) = g Wkj Sj (t) (17) terations in one variable in response to changes in another (Witten and
j=1 k=1
James, 2013; Raudys and Jain, 1991; Haykin, 1998; Drucker et al.,
where Sj (t) represents the output of hidden neuron j, Xi (t) designs the 1999).
Overall, the procedure to construct an ENN-QVAR model can be
input data to neuron i. The matrices γ, V, and W represent the weights of
described as follows:
the network.
Algorithm: ENN-QVAR algorithm
Input: Multivariate time series data: X;
Setup 1: Estimate QVAR model → Extract non-linear dynamic in
formation coefficients.
Setup 2: Decompose into joint total connectedness index: JTO =
∑k
=j JTRSI ij
j=1,i∕
Setup 3: Prepare ENN input sequences using QVAR outputs.
Setup 4: Train ENN model on these sequences.
Setup 5: Use the ENN model to make predictions. The model will
leverage the dynamic relationships learned from the QVAR outputs.
Fig. 1. Architecture of the Elman Neural Network. Note: The Elman Neural
Setup 6: Evaluate the model using appropriate metrics like RMSE
Network (ENN) is commonly structured with four layers. One of these layers is
the state unit, which plays a crucial role in incorporating historical states into (Root Mean Square Error) and MSE (Mean Squared Error)
the network’s computations. This capability makes the ENN particularly well- Moreover, a simple description of the concept of the algorithm ENN-
suited for addressing time-varying problems. By utilizing the information QVAR is given in Appendix A.
from previous states, the ENN can capture temporal dependencies and effec
tively model dynamic systems. The inclusion of the state unit enhances the
network’s ability to handle sequences of data and adapt to changing patterns 3.4. Data
over time. Consequently, the ENN has found extensive applications in various
domains where time-varying information is prevalent (Xie et al., 2020). The primary objective of this study is to examine the
6
Table 1
Descriptive statistics of variables.
Bitcoin Bitcoin Green Dash Dash Green Evergreencoin Greencoin Stellar X8x RECT
Mean 0.183 − 0.304 − 0.048 − 0.620 − 0.018 − 0.001 − 0.018 − 0.128 0.009
Std. Dev. 4.493 15.214 6.646 36.623 7.773 10.302 6.391 11.237 1.633
Skewness − 1.097 1.294 − 0.328 − 0.009 − 0.458 0.417 0.669 0.627 − 0.966
Kurtosis 16.332 21.646 11.021 21.933 17.069 26.404 14.672 16.670 15.524
JB 8275.9 16,065.4 2935.8 16,249.4 9011.6 24,862.8 6257.5 8542.9 7280.4
ADF − 31.764 − 28.864 − 32.879 − 22.671 − 35.822 − 39.623 − 33.159 − 35.536 − 11.595
PP − 33.818 − 45.193 − 32.888 − 55.430 − 35.789 − 40.210 − 33.159 − 35.884 − 34.523
KPSS 0.085 0.064 0.040 0.075 0.034 0.031 0.093 0.035 0.047
interconnections among different classes of digital crypto-based in 3. Green-synthesized digital innovations:
novations and their association with heterogeneous blockchains based
on renewable energy and clean technologies. To achieve this goal, we • DASHG: An eco-friendly digital money system that creates coins
collected a comprehensive dataset that included the following crypto- through a process called “minting,” which has negligible environ
based innovations and their corresponding blockchain technologies: mental impact.
• BITG: An open and permissionless blockchain built to meet the needs
1. Conventional digital innovations: of non-governmental organizations; environmental, social, and
corporate governance groups; and purpose-driven innovation on
• Dash: A well-established digital currency known for its secure and Web3.
fast transactions.
• Bitcoin: This pioneering cryptocurrency revolutionized the financial 4. Islamic digital innovations:
landscape with its decentralized nature and widespread adoption.
• X8X Token: A value preservation payment token approved by Islamic
2. Green digital innovations: scholars to be marketed as a Sharia-compliant digital currency. This
certification enabled the expansion of X8 AG, a land-based block
• EverGreenCoin: A peer-to-peer digital asset that aims to raise funds chain technology company in the Middle East.
for green environmental projects and promote sustainability. • Stellar (XRP Ledger): Leveraging XRP Ledger, Stellar offers perfor
• Greencoin: A cryptographically secure digital asset generated by a mance, scalability, and inherently green attributes. XRP Ledger is
distributed network of miners integrated with a compulsory sharing one of the first major carbon-neutral blockchains to use a federated
mechanism with certified renewable commerce providers. consensus algorithm for energy-efficient and low-cost transactions.
Fig. 2. Plot of returns.
7
Table 2
Average of dynamic connectedness in each bi-variate model.
Quantile 25th 50th 75th 25th 50th 75th 25th 50th 75th
Bitcoin Green-RECT Bitcoin Green RECT FROM

Bitcoin Green 93.70 97.85 92.55 6.30 2.15 7.45 6.30 2.15 7.45
RECT 6.94 2.93 8.36 93.06 97.07 91.64 6.94 2.93 8.36
TO 6.94 2.93 8.36 6.30 2.15 7.45 TCI
NET 0.64 0.78 0.90 − 0.64 − 0.78 − 0.90 6.62 2.54 7.90
Bitcoin-RECT Bitcoin RECT FROM
Bitcoin 77.78 86.86 76.41 22.22 13.14 23.59 22.22 13.14 23.59
RECT 22.53 14.14 24.64 77.47 85.86 75.36 22.53 14.14 24.64
TO 22.53 14.14 24.64 22.22 13.14 23.59 TCI
NET 0.31 1.00 1.05 − 0.31 − 1.00 − 1.05 22.37 13.64 24.12
Dash Green-RECT Dash Green RECT FROM
Dash Green 97.08 99.28 97.70 2.92 0.72 2.30 2.92 0.72 2.30
RECT 3.51 1.39 2.67 96.49 98.61 97.33 3.51 1.39 2.67
TO 3.51 1.39 2.67 2.92 0.72 2.30 TCI
NET 0.59 0.67 0.37 − 0.59 − 0.67 − 0.37 3.21 1.06 2.49
Dash-RECT Dash RECT FROM
Dash 84.36 93.30 82.83 15.64 6.70 17.17 15.64 6.70 17.17
RECT 16.72 7.45 17.58 83.28 92.55 82.42 16.72 7.45 17.58
TO 16.72 7.45 17.58 15.64 6.70 17.17 TCI
NET 1.08 0.75 0.41 − 1.08 − 0.75 − 0.41 16.18 7.08 17.38
EverGreenCoin-RECT EverGreenCoin RECT FROM
EverGreenCoin 83.24 91.08 81.20 16.76 8.92 18.80 16.76 8.92 18.80
RECT 17.02 9.53 18.99 82.98 90.47 81.01 17.02 9.53 18.99
TO 17.02 9.53 18.99 16.76 8.92 18.80 TCI
NET 0.26 0.61 0.19 − 0.26 − 0.61 − 0.19 16.89 9.23 18.90
Greencoin-RECT Greencoin RECT FROM
Greencoin 89.87 93.06 87.41 10.13 6.94 12.59 10.13 6.94 12.59
RECT 11.15 8.29 14.04 88.85 91.71 85.96 11.15 8.29 14.04
TO 11.15 8.29 14.04 10.13 6.94 12.59 TCI
NET 1.02 1.35 1.45 − 1.02 − 1.35 − 1.45 10.64 7.61 13.31
Stellar-RECT Stellar RECT FROM
Stellar 80.65 90.10 80.13 19.35 9.90 19.87 19.35 9.90 19.87
RECT 19.53 10.49 20.76 80.47 89.51 79.24 19.53 10.49 20.76
TO 19.53 10.49 20.76 19.35 9.90 19.87 TCI
NET 0.18 0.59 0.90 − 0.18 − 0.59 − 0.90 19.44 10.20 20.32
X8X-RECT X8X RECT FROM
X8X 90.13 95.16 88.88 9.87 4.84 11.12 9.87 4.84 11.12
RECT 10.30 6.01 11.72 89.70 93.99 88.28 10.30 6.01 11.72
TO 10.30 6.01 11.72 9.87 4.84 11.12 TCI
NET 0.43 1.17 0.61 − 0.43 − 1.17 − 0.61 10.08 5.43 11.42
Note: The findings presented in this study are derived from a 200-day rolling-window Quantile Vector Autoregression (QVAR) model, with a lag length determined by
the Bayesian Information Criterion (BIC). To assess the forecasting performance, a 10-step-ahead generalized forecast error variance decomposition is conducted. Each
cell in the results represents the outcome based on the 25th, 50th, and 75th quantiles, providing insights across different levels of uncertainty. In the results, the
diagonal values represent the shares of the own-variance, indicating the proportion of shocks that can be attributed to the variable itself. On the other hand, the off-
diagonal values represent the cross-variance shares, reflecting the proportion of shocks transmitted from other variables to the specific variable under consideration.
When examining the “FROM” values, it highlights the influence that all other variables (denoted as j) have on the variable of interest (denoted as i). Conversely, the
“TO” measures indicate the influence that the variable i has on all other variables (denoted as j). By comparing the TO and FROM measures, the NET spillovers can be
calculated. A positive (negative) NET spillover implies that the variable i acts as a net transmitter (receiver) of shocks. Moreover, the Total Connectedness Index (TCI)
provides a measure of network interconnectedness. It is derived by averaging the TO or FROM values, indicating the degree of influence between variables. The TCI
ranges between zero and unity, where higher values signify a greater level of interconnectedness within the network. All connectedness indices have been scaled by a
factor of 100 to enhance interpretability and facilitate a clearer comprehension of the connections.
To begin our study, we considered the start dates of the respective

Table 3 cryptocurrencies. DASHG, the latest understudied cryptocurrency, was
MAE for the prediction part of the network (trainer: trainrp (Resilient Back
introduced on January 14, 2019. Henceforth, our research period was
propagation) approach).
slated to commence on January 14, 2019, extending until May 12, 2023,
Quantile 25th 50th 75th utilizing a daily frequency and a natural logarithmic format. For BITG
Bitcoin Green 0.2222 0.0530 0.1176 and DASHG, data were extracted from Yahoo! Finance (https://finance.
Bitcoin 2.9111 2.0261 3.2362 yahoo.com/). BITG is a cryptocurrency that users can generate through
Dash Green 0.1059 0.0363 0.2728
mining. BITG, an open and permissionless blockchain, is designed to
Dash 2.8885 1.5059 2.9185
EverGreenCoin 2.8827 1.9952 3.2119 meet the needs of non-governmental organizations; environmental, so
Greencoin 1.7679 1.2993 2.2580 cial, and corporate governance groups; and purpose-driven innovation
Stellar 2.1831 1.1225 2.3848 on Web3. By combining blockchain technology with green innovation,
X8X 1.2032 0.6319 1.4392 BITG aims to drive capital toward critically important sustainability
Note: The Mean Absolute Error (MAE) is a performance metric used to assess the initiatives.
accuracy of predictions made by an Elman Neural Network (ENN) in the context DASHG represents an eco-friendly digital money system where coins
of regression tasks for dynamic total connectedness of RECT and each are created through the “minting” process. This approach has negligible
cryptocurrency-based digital innovation. It quantifies the average magnitude of environmental effects and has zero cost. All DASHG sites are hosted in
the errors between the predicted outputs and the actual outputs. renewable-energy data centers, thus reflecting a commitment to a safer
and greener environment. Transactions are facilitated through proof-of-
8
Fig. 3. Dynamic total connectedness of RECT and each cryptocurrency-based digital innovation.
Fig. 4. Average total connectedness of RECT and each cryptocurrency-based digital innovation.
stake and master nodes, enabling instant and private transactions both producers, thus capturing the environmental and public policy values of
online and in physical stores. Each transaction incurs a near-zero fee, carbon-reduced human productivity. EverGreenCoin, however, goes
thereby ensuring affordability for users. Moreover, the Zerocoin Proto beyond being a mere currency and emphasizes raising funds for green
col enhances privacy and anonymity. environmental projects. It inherits traits from Bitcoin, such as low-cost
For Greencoin and EverGreenCoin, data are extracted from Inve international transfers and zero risk of loss of personal information or
sting.com (https://www.investing.com). Greencoin is a peer-to-peer manipulation by governments and banks.
cryptographically secure digital asset generated by a distributed The X8X Token and Stellar data were obtained from Yahoo! Finance
network of miners. It features a compulsory sharing mechanism with (https://finance.yahoo.com/). X8 AG, the Switzerland-based blockchain
certified renewable energy providers. The base value of Greencoin assets technology company, has obtained approval from Islamic scholars for
is tied to the aggregate carbon sequestered by renewable energy marketing X8X Token as a Sharia-compliant digital currency. This
9
certification allows for X8X Token’s expansion in the Middle East, where capture the dynamic and time-varying nature of the data, which has
regulators and financial exchanges are increasingly encouraging inno been supported by previous studies (Abakah et al., 2022; Abakah et al.,
vation in finance. Stellar Coin leverages the XRP Ledger, which is built 2023a; 2023b; Abdullah et al., 2023; Madaleno et al., 2023; Mensi et al.,
with sustainability in mind. It was one of the first major carbon-neutral 2022). The lag length for the model was determined using the Bayesian
blockchains to provide energy-efficient and low-cost transactions. The information criterion to ensure an appropriate representation of the
GreenFuel ecosystem focuses on climate change, climate security, the relationships. To evaluate the forecasting performance, a 10-step-ahead
prevention of illicit financial transactions, and carbon emission initia generalized forecast error variance decomposition was conducted. This
tives using blockchain technology for transparent carbon accounting. To decomposition allowed us to understand the contribution of each vari
capture the broader landscape of green investments, we employ the able to the overall forecast error variance across the different quantiles
S&P/TSX renewable energy and clean technology index (abbreviated as (25th, 50th, and 75th), providing a comprehensive view of the un
RECT) as a proxy. This index serves as a reliable measure of performance certainties involved.
and trends in the green investment sector. Further details regarding this In Table 2, the diagonal values represent the share of variance,
index can be obtained from S&P Global (https://www.spglobal. indicating the proportion of shocks that can be attributed to each vari
com/en/). able. By contrast, the off-diagonal values represent cross-variance
Table 1 presents an in-depth analysis of the variables, using shares, revealing the extent to which shocks are transmitted from
descriptive statistics to shed light on their central tendency, variability, other variables to the specific variable of interest. The “FROM” values
skewness, kurtosis, normality, and stationarity. The mean values pro highlight the influence of all other variables (denoted as j) on the vari
vide valuable insights into the average performance of each variable. able under consideration (denoted as i). Conversely, the “TO” measures
Bitcoin has a relatively positive average performance, as indicated by its indicate the influence of the variable i on all other variables (denoted as
highest mean value (0.183). Conversely, BITG exhibits the lowest mean j). By comparing the TO and FROM measures, we can calculate the NET
value (− 0.304), suggesting a relatively negative average performance. spillovers, which indicate whether the variable i acts as a net transmitter
The remaining variables, including Dash, DASHG, EverGreenCoin, or receiver of shocks. Additionally, the TCI serves as a measure of
Greencoin, Stellar, X8X, and RECT, demonstrate mean values that hover network interconnectedness. It is derived by averaging the TO or FROM
around zero, suggesting a relatively neutral average performance. values, providing an overall assessment of the degree of influence be
In terms of variability, the standard deviation provides a measure of tween variables. The TCI ranges between zero and unity, where higher
dispersion and volatility among variable values. Greencoin demon values indicate a higher level of interconnectedness within the network.
strates the highest standard deviation (10.302), indicating a higher level Fig. 3 shows the dynamic TCI of RECT alongside each
of volatility. DASHG exhibits a standard deviation of 36.623, indicating cryptocurrency-based digital innovation, which served as the central
a relatively significant variability. By contrast, the other variables focus of this study. The interrelationships among the variables sub
display lower standard deviations, indicating lower volatility. stantially intensified during periods characterized by heightened market
Skewness, which assesses the asymmetry of variable distribution, volatility. This phenomenon is depicted in Fig. 3, encompassing both the
reveals interesting patterns among the variables. BITG demonstrates a duration of the COVID-19 pandemic and the period of heightened global
positive skewness of 1.294, indicating a longer right tail and positive inflation coinciding with the Russo-Ukrainian war. The graphical rep
deviation from a symmetric distribution. Stellar and X8X also exhibit resentation in Fig. 3 serves as a tangible testament to the heightened
positive skewness, albeit to a lesser extent. Conversely, Bitcoin and intricacy and interplay among diverse factors amid these turbulent
Evergreen show negative skewness, signifying a longer left tail and a events. Fig. 4 shows the average TCI for these innovations. By examining
negative deviation from a symmetric distribution. The remaining vari cryptocurrency-based digital innovations, it is evident that the nexus
ables demonstrate skewness values close to zero, indicating a relatively between digital assets and green investment intensified during extreme
symmetric distribution. market conditions. Conversely, the middle quartile exhibited the
Kurtosis, which measures the peakedness or flatness of a variable weakest association between crypto-driven technological advancements
distribution, provides further insight. BITG had the highest kurtosis and environmentally conscious investments. Notably, the interconnec
value of 21.646, suggesting a relatively peak distribution with heavy tedness between cryptocurrency innovations and sustainable in
tails. Similarly, Stellar, DASHG, and EverGreenCoin display relatively vestments reached its pinnacle within bull markets characterized by
high kurtosis values, indicating a peak distribution. Conversely, the upper tails and remarkably high returns, with the exception of DASHG.
remaining variables exhibited kurtosis values of approximately 15, Our findings partially converge with those of Kamal and Hassan (2022),
indicating less peaked distributions compared to the variables. who observed a high (low) level of connectedness between the crypto
The Jarque-Bera (JB) test statistic enables evaluation of the depar currency environment attention index and green assets at higher (lower)
ture of variable distributions from a normal distribution. Greencoin quantiles based on their quantile connectedness analysis. While there is
exhibits the highest JB test value (24,862.8), signifying a significant some inconsistency between our study and theirs, both underscore the
departure from normality. Similarly, BITG, DASHG, and Stellar have complexity and variability of the relationship between assets, particu
relatively high JB test statistics, suggesting a departure from normality. larly across different quantile levels. These differences highlight the
Overall, the variables showed a general departure from normality. need for further exploration and analysis to fully understand the dy
In terms of stationarity, the Augmented Dickey-Fuller, Phillips- namics of asset interconnectedness under different market conditions. A
Perron, and Kwiatkowski-Phillips-Schmidt-Shin tests offer valuable in meticulous comparative analysis of cryptocurrency-based digital tech
sights. None of the variables provide evidence for the rejection of the nologies revealed that Bitcoin and Stellar exhibited the most robust
stationarity hypothesis, as all values are negative, suggesting the pres interplay with eco-friendly financial ventures across diverse market
ence of stationarity. Fig. 2 shows a graphical representation of the var states. Considering that the average connectedness between
iables, reinforcing the insights presented in Table 1. cryptocurrency-based technologies and green investments fell below the
25 % threshold, we posit that cryptocurrency, as a nascent asset class,
4. Results has significant potential as a valuable diversification tool for sustainable
portfolios. This supposition aligns harmoniously with the findings of
Table 2 presents the average dynamic connectedness of each bivar Irfan et al. (2023), Omane-Adjepong et al. (2019), Chuen et al. (2017),
iate model, providing important insights into the relationships and in and Corbet et al. (2018), who elucidate the demonstrably low correla
terdependencies among the studied variables. The 200-day rolling- tion and relative isolation between cryptocurrencies and other financial
window QVAR model served as the analytical framework for this assets.
study. This window size was selected based on its ability to effectively Furthermore, we found that the level of interconnectedness between
10
Bitcoin Green (Q=25th)
Fig. 5. The performance of the network in forecasting and validating the RECT-Bitcoin Green case (Note: There is a blue line in the first column which indicates
prediction of dynamic total connectedness for the latest 100 days, and a red line indicating the target (i.e., real data). The second and third column, revealed the
validation performance results). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
cryptocurrency innovations and sustainable investments was most pro devoid of crisis or turmoil. The latter’s research provides additional
nounced in bull markets characterized by upper tails and exceptionally support for the understanding that the association between Bitcoin and
high returns. However, it is worth noting that DASHG deviated from this the clean energy sector remains constrained and fragile under normal
pattern. A thorough comparative analysis of cryptocurrency-based circumstances. Furthermore, our research reveals significant associa
technologies shows that Bitcoin and Stellar exhibited the strongest tions between environmentally conscious investment practices and
nexus with ecofriendly financial ventures across various market states. other cryptocurrencies, such as Dash, EverGreenCoin, Greencoin, and
Our findings demonstrate that the type of technology employed by X8X. These findings suggest that environmental awareness and the
cryptocurrencies does not significantly alter the strength of the rela Shariah-compliant nature of cryptocurrency-driven innovations do not
tionship between these technologies and sustainable investment. Bit substantially alter the nature and structure of the relationship between
coin, commonly regarded as a “dirty” digital currency due to its use of a advancements in crypto-based technologies and green investment ini
proof-of-work mechanism during its production process, and Stellar, tiatives. Our findings highlight the complexity of the relationship be
which employs proof-of-stake technology and is recognized as a “clean” tween cryptocurrency innovation and sustainable investments. Factors
form of digital currency (Patel et al., 2023), both exhibited substantial such as market dynamics, market capitalization, and the distinct char
associations with environmentally sustainable stocks. This finding sug acteristics of cryptocurrencies play influential roles in establishing
gests that factors such as market capitalization and the distinctive connections with climate-oriented investment strategies. Bitcoin and
characteristics of Bitcoin and Stellar, particularly their prominent roles Stellar are strongly associated with environmentally sustainable stocks,
in the FinTech sector, exert a more significant influence on establishing thereby surpassing their intrinsic climate-conscious qualities. However,
connections with climate-oriented investment strategies in the financial the correlation between Bitcoin and the clean energy market remains
domain, surpassing their intrinsic climate-conscious qualities. tenuous, and the association between crypto-based technologies and
Moreover, the outcomes of our study are consistent with those of sustainable investment is not significantly altered by environmental
Syed et al. (2022), who examine the relationship between Bitcoin and awareness or Shariah compliance.
green bonds, an environmentally friendly financial instrument. Simi Additionally, we found that during periods of extreme market con
larly, our findings align with those of Attarzadeh and Balcilar (2022), ditions, a notable relationship emerged between digital currencies and
further reinforcing the notion that the correlation between Bitcoin and environmentally sustainable investments. Several factors could explain
the clean energy market remains tenuous, especially during periods this correlation. First, forward-thinking environmentally conscious
11
Bitcoin (Q=25th)
Bitcoin (Q=50th)
Bitcoin (Q=75th)
Fig. 6. The performance of the network in forecasting and validating the RECT-Bitcoin case (Note: see note of Fig. 5).
enterprises, particularly those operating in the clean energy and eco- through which green enterprises can access additional financial re
friendly technology sectors, demonstrate a proclivity for portfolio sources and navigate challenging monetary environments (Pilloni et al.,
diversification as a risk-management strategy. Meanwhile, digital cur 2022). In summary, the association between digital currencies and
rencies offer a compelling avenue for achieving diversification objec environmentally sustainable investments during market volatility arises
tives. When conventional financial markets exhibit heightened volatility from their capacity to mitigate risk, facilitate capital acquisition, align
or uncertainty, they may allocate a portion of their capital to digital with societal concerns, and provide resilience under challenging finan
currencies to reduce risk (Arfaoui et al., 2023; Le et al., 2021). Given the cial circumstances. This connection underscores the synergistic inter
differential behavior of digital currencies compared to traditional fiat action between digital currencies and ecofriendly investments,
currencies, their inclusion in portfolios can effectively safeguard assets particularly when conventional markets are uncertain and unstable.
during periods of market turbulence. BITG and DASHG, green-synthesized cryptocurrency-driven in
Second, green companies that engage in ecological projects often novations, exhibit the lowest degree of association with eco-friendly
require substantial capital for research, development, and expansion. investment approaches. The perceptions and reputations of BITG and
Digital currencies present alternative channels for fundraising, such as DASHG within the investment community may contribute to their
initial coin offerings and security token offerings (Chou et al., 2023). limited connections with eco-friendly investment approaches. Despite
During challenging market conditions When conventional fundraising exhibiting distinct herding behavior and differing responses during
avenues are constrained, green enterprises may turn to digital currency market downturns (Ren and Lucey, 2022b), the escalating pressure of
to acquire the financial resources necessary to sustain their eco-projects. climate concerns has led to a clear dichotomy in the cryptocurrency
Third, the nexus between digital currencies and eco-friendly in market based on the energy consumption involved in mining or pro
vestments can be attributed to broader global trends and societal shifts. duction (Ren and Lucey, 2022a). Notably, our study’s novel findings
As society increasingly emphasizes environmental sustainability and regarding the TCI results contradict those of recent studies. For instance,
responsible practices, investments in green initiatives have become Ren and Lucey (2022a) propose that clean energy can serve as a hedge
increasingly significant. The convergence of these evolving preferences and weak safe haven against the downside risk associated with “dirty”
and the growing acceptance of digital currencies engender a mutually cryptocurrencies. Similarly, Naeem and Karim (2021) suggest that clean
reinforcing relationship, particularly during periods of market volatility. energy can act as a valuable hedge against downside risks in Bitcoin.
Finally, innovative green companies may leverage digital currencies to Furthermore, our findings offer new insights that supplement the work
capitalize on financial opportunities or enhance their liquidity positions of Patel et al. (2023), who emphasize that portfolio managers should not
under turbulent market conditions. Digital currencies serve as conduits treat green and dirty cryptos as interchangeable assets, particularly
12
Dash Green (Q=25th)
Dash Green (Q=50th)
Dash Green (Q=75th)
Fig. 7. The performance of the network in forecasting and validating the RECT-Dash Green case (Note: see note of Fig. 5). (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)
during crises, when they exhibit divergent behavior. Many recent the impact of the educational function on the outcomes and determine
studies investigate the connectedness of green investments and other the susceptibility of the model to this determinant. The training pro
assets (Annamalaisamy and Vepur Jayaraman, 2023; Arfaoui et al., tocols provided insightful gradient feedback that facilitated model
2023; Ndubuisi and Urom, 2023; Omane-Adjepong and Alagidede, enhancement through strategic weight and bias alterations. This judi
2020; Sharif et al., 2023), but none evaluate the non-linear information cious selection and comparison of training algorithms allowed for a
contained in their connectivity. Therefore, we proceed to the second rigorous evaluation of the modeling sensitivity and offered vital intent to
stage of our analysis by evaluating the non-linear dynamic information propel further refinement of the network architecture and
contained in the TCI of each pair using an Elman neural network. hyperparameters.
Figs. 5–12 each contain three panels denoting different quantiles To assess the ability of the Elman network to generalize and mitigate
(25th, 50th, and 75th). In the leftmost column, the blue line represents the risk of overfitting, a separate validation dataset containing instances
the dynamic total connectedness prediction for the previous 100 days, that were not observed during training was used. The best validation
whereas the red line represents the actual target data. Columns 2 and 3 performance was determined by identifying the epoch at which the
present validation results. Within the context of the Elman network, the Elman network achieved the lowest or highest accuracy in the validation
validation section evaluates network performance on a dedicated vali dataset. Training typically comprises multiple epochs representing the
dation dataset during the training process. As the network underwent complete iterations of the training dataset. The epoch associated with
training, it iteratively adjusted its weights and biases to minimize the optimal performance can be identified by continuously monitoring the
disparity between its predictions and the actual values in the training validation performance at each epoch. This epoch indicated when to
dataset. In our preliminary investigations, we leveraged the training complete the training process or select the optimal set of weights and
protocol as a neural-network educational operation that capitalized on biases for the Elman network. This represents the point at which the
the resilient back-propagation technique to amend the system’s weights network achieves optimal performance on the unseen data, suggesting
and biases. This algorithm modifies the load and bias values in line with that further training may result in diminished returns or potential
the gradient information to optimize network execution. To assess the overfitting. The value of the gradient at a specific epoch reflects both its
sensitivity of the consequences of the educational function selection, we magnitude and direction, which are computed using a back-propagation
conducted a robust examination using a training protocol. The training algorithm during training. The gradient captures the slope of the loss
algorithm employs a Levenberg-Marquardt optimization strategy to function with respect to the network parameters, including the weights
rework the load and bias valuations. By comparatively analyzing the and biases. As training progressed, the Elman network adjusted these
results obtained from different educational functions, we aimed to gauge parameters by evaluating the magnitude of the gradient and updating
13
Dash (Q=25th)
Dash (Q=50th)
Dash (Q=75th)
Fig. 8. The performance of the network in forecasting and validating the RECT-Dash case (Note: see note of Fig. 5).
the parameters in opposite directions to navigate the loss landscape accuracy. In addition to the MAE, we implemented the root mean square
effectively. The gradient value at a particular epoch indicates the error (RMSE) as an additional metric for cross validation. The RMSE was
steepness of the landscape loss during the training phase. A larger used to assess the precision of the network forecasts and serve as a
magnitude implies a steeper slope, indicating a rapid parameter complementary measure for evaluating predictive accuracy during
adjustment within the network. Conversely, a smaller magnitude sug robustness testing. By incorporating the RMSE, we aimed to achieve a
gests a flatter slope, indicating that the network is closer to a minimum more comprehensive evaluation of the sensitivity of the model to
or a plateau in the loss landscape. different training algorithms. This allowed us to assess the repeatability
During the training process, data are typically divided into three sets: of the predictions across various training routines and obtain a well-
training, validation, and test sets. The training set was used to update the rounded understanding of the model’s performance.
network parameters, whereas the validation set was crucial for assessing When analyzing the MAE of the predictions extracted from the first
the performance and making decisions during training. Validation columns of Figs. 5–12 (summarized in Fig. 13), compelling observations
checks were conducted periodically on the validation set to monitor the emerged. Notably, the non-linear dynamic information was associated
ability of the network to generalize and perform on unseen data. A with green-synthesized cryptocurrency-driven innovations, while their
validation failure occurs when the network performance on the valida connection to green investments exhibited remarkably higher values
tion set falls below a predefined threshold, as measured by metrics such than other innovations. Furthermore, other digital innovations within
as the validation error or accuracy. Identifying validation failures helps the cryptocurrency domain demonstrated considerably higher MAE
guide decisions such as terminating training or employing early- values, indicating larger disparities between their predicted and actual
stopping techniques to prevent overfitting. In the prediction phase of values, while exhibiting comparable non-linear dynamic information
the network, the mean absolute error (MAE) is computed by comparing during the prediction phase of the Elman network. However, an
the predicted values with the actual values. The MAE quantifies the exhaustive exploration of the literature revealed a paucity of studies
average absolute difference between the predicted and actual values and directly addressing the focal points of the present study. Nonetheless,
is commonly used to measure prediction accuracy. In the Elman our empirical findings exhibit remarkable congruence with the asser
network, the prediction phase corresponds to the output layer, where tions of Ye et al. (2023a, 2023b) regarding the intricate and non-linear
the network generates predictions based on the input data, learned ramifications arising from the intersection of blockchain technology and
weights, and biases. The MAE was calculated by summing the absolute green investment. Their scholarly discourse underscores the pressing
differences between the predicted and actual values and dividing by the need for judicious regulatory measures, be it through stringent control
total number of predictions. A lower MAE indicates a higher prediction mechanisms governing the volume of cryptocurrencies or through the
accuracy, reflecting closer agreement between the predicted and actual adoption of environmentally conscious practices such as the leveraging
values, whereas a higher MAE signifies larger discrepancies and reduced of renewable energy sources in the mining process. Similarly, our
14
EverGreenCoin (Q=25th)
Fig. 9. The performance of the network in forecasting and validating the RECT-EverGreenCoin case (Note: see note of Fig. 5).
findings align with those of Migliorelli (2021), providing comprehensive Furthermore, financial markets are subject to significant effects from
support for the emergence of green cryptocurrencies as pivotal assets in the ongoing COVID-19 pandemic and the Russo-Ukrainian conflict,
the domain of sustainable finance, wherein holistic considerations resulting in potential oscillations between bear and bull market phases.
encompass environmental, social, and governance effects that inform The Russo-Ukrainian conflict, a geopolitical confrontation, introduced a
investment decisions. heightened level of uncertainty in the market landscape. This uncer
In response to the exigencies of the current environmental landscape, tainty increased the perceived risk associated with investments, partic
the blockchain industry has championed the development of BITG as an ularly those tied to affected regions or industries. Therefore, investors
avant-garde solution aimed at mitigating environmental pollution and may exhibit caution and adopt defensive strategies that contribute to
safeguarding our planet (Jacquet and Mans, 2019). Notably, antecedent increased market volatility, and in certain instances, bearish sentiment
research indicates that prior to the onset of the COVID-19 pandemic, (Bossman and Gubareva, 2023; Bossman et al., 2023; Khalfaoui et al.,
Bitcoin demonstrated comparatively diminished efficiency levels (Mnif 2023). Simultaneously, significant events such as the COVID-19
et al., 2023). pandemic have had a profound impact on various aspects of the econ
In contrast to the relative prominence of BITG, DASHG has not omy, including economic activities, corporate earnings, and investor
received significant attention in the literature. The comparatively sentiment. The implementation of lockdown measures, disruptions in
modest valuation of BITG provides practical advantages for prospective global supply chains, and a decline in consumer spending collectively
traders. By committing to relatively modest financial resources, traders contributed to market contractions, potentially triggering a shift toward
amass substantial quantities of coins to fortify their holdings. Conse a bear market regime. Meanwhile, the adverse effects on economic
quently, in the event of an appreciable increase in the value of crypto fundamentals and investor confidence amplify market downturns (Ali
currencies, the resultant profits are magnified because of extensive et al., 2021; Chiappini et al., 2021; Conlon and McGee, 2020; Maheu
cryptocurrency holdings. The heightened visibility of energy-conserving et al., 2021). Given these circumstances, our results—particularly those
cryptocurrencies has the potential to catalyze an upward trajectory in pertaining to bear market conditions relative to normal and bull mar
the valuation of BITG. Moreover, traders are afforded a singular op kets—offer valuable insights for investors and policymakers facing
portunity to own and operate the BITG master nodes, which actively similar challenges. Understanding the dynamics of bear markets, their
contribute to the maintenance and fortification of the BITG network triggers, and their implications can assist investors in making informed
through mining. These master nodes epitomize virtual private servers decisions, effectively managing risks, and developing appropriate stra
that meticulously store an accurate replica of the BITG blockchain and tegies to navigate uncertain market conditions. Policymakers can also
confer rewards to their owners and managers, akin to those reaped by benefit from these insights when formulating measures to mitigate the
traditional Bitcoin miners (Mnif et al., 2023). adverse effects of geopolitical conflicts and pandemics on financial
15
Greencoin (Q=25th)
Greencoin (Q=50th)
Greencoin (Q=75th)
Fig. 10. The performance of the network in forecasting and validating the RECT-Greencoin case (Note: see note of Fig. 5).
markets and broader economies. observations, providing insights into the model’s overall performance.
However, it assigns a greater weight to larger errors because of its
5. Robustness tests squared-error calculation, making it sensitive to outliers. By employing
both the MAE and RMSE, researchers have gained a more nuanced un
Robustness checks that consider multiple aspects of errors and the derstanding of the model’s strengths and weaknesses. This approach is
exploration of different optimization algorithms in the training process advantageous because it permits model evaluation from diverse per
offer significant benefits in assessing the performance of the model. By spectives, allowing stakeholders to make informed decisions based on
examining various error dimensions, including the magnitude and di the specific objectives and requirements of the predictive task. Table 4
rection, these tests provide a comprehensive and nuanced evaluation of and Fig. 14 provide supporting evidence for the RMSE, confirming the
a model’s ability to capture different types of errors and make accurate findings derived from the MAE. A consistent ranking of predictive errors
predictions. Furthermore, by employing diverse optimization algo exists when assessing the dynamic connectedness of various
rithms during training, these tests enable the investigation of algo cryptocurrency-based digital innovations using a resilient back
rithmic sensitivity and identification of the most suitable optimization propagation method. This consistency highlights the stability and
approach for achieving optimal model performance. These approaches durability of the observed error patterns across the different evaluation
enhance the rigor and reliability of the robustness tests, enabling a more metrics.
robust and comprehensive assessment of the model performance and its
potential for generalization across diverse scenarios.
5.2. Optimization algorithms in the training process
5.1. Predictive model: errors Another crucial facet of assessing model robustness is the exploration
of different optimization algorithms during the model-training phase.
In predictive modeling, evaluating model performance is a critical Optimization algorithms play a pivotal role in fine-tuning the model
step in ensuring the reliability and utility of a model’s predictions. A parameters to minimize prediction errors. Researchers have adopted this
commonly employed approach to assess model robustness involves approach to scrutinize the sensitivity of the model’s performance to the
considering different aspects of errors. This approach involves the choice of optimization method. By training the model with multiple
calculation of multiple error metrics to gauge the accuracy of the model algorithms, such as gradient descent or resilient backpropagation,
from various perspectives. Two prominent metrics often utilized for this Levenberg-Marquard gained insights into the behavior of the model
purpose are the MAE and the RMSE. The MAE quantifies the average under varying training conditions. This approach is beneficial because it
absolute difference between the predicted values and actual helps identify the consistency of the model’s performance across
16
Stellar (Q=25th)
Stellar (Q=50th)
Stellar (Q=75th)
Fig. 11. The performance of the network in forecasting and validating the RECT-Stellar case (Note: see note of Fig. 5).
different optimization techniques. This method may require additional variable’s incremental impact provides a nuanced understanding of the
computational resources and time; however, the insights gained from it financial repercussions of their interactions, thereby elucidating the
are invaluable for making informed decisions regarding training pro subtleties of their symbiotic relationships.
cedures. As evident in Tables 5 and 6 and Figs. 14–16, the resilient
backpropagation and Levenberg-Marquardt methods produced similar 6. Conclusion and policy recommendations
results. This highlights the strength and reliability of our findings for the
different training algorithms. Importantly, our analysis demonstrated a The escalating global imperative to address environmental chal
lack of significant variation caused by trainers in the dataset and lenges and mitigate the effects of climate change underscores the sig
resulting outcomes, thus confirming the stability and consistency of our nificance of environmentally conscious investment. These investments
experimental results. encompass diverse initiatives aimed at fostering sustainability,
This study contributes substantially to our understanding of the including renewable-energy undertakings, clean technologies, and eco-
complex and fluid relationship between green investment and digital friendly infrastructure projects. To expedite the global transition to
crypto-based innovation. It examines several types of cryptocurrencies, ward a low-carbon economy, it is important to acquire a comprehensive
including green, green-synthesized, Islamic, and conventional ones. By understanding of the effect of different cryptocurrency categories on
empirically, using advanced quantitative approaches, such as the QVAR environmentally conscious investments. These categories include green,
extended joint connectedness approach and the Elman neural network, green-synthesized, Islamic, and conventional cryptocurrencies. By
the present study offers valuable knowledge regarding the changing meticulously examining their impact, effective pathways can be iden
dynamics of these associations. These sophisticated methodologies have tified to expedite the adoption of sustainable practices.
enabled a comprehensive analysis of the interconnectedness and varia Further, the ethical dimension of investment decisions has gained
tions between environmentally friendly investments and the previously prominence among both individuals and institutions. Islamic finance
described forms of digital currency. From a marginal economic effect operates in accordance with Shariah principles, which emphasize ethical
perspective, our study elucidates the complex financial ramifications and socially responsible investments. Analyzing the alignment between
stemming from the intricate interconnectedness between digital cur Islamic cryptocurrencies and environmentally conscious investments
rencies and environmentally sustainable investments, viewed through offers a distinctive opportunity to pursue sustainable financing that
the lens of marginal economic effects. A meticulous examination of each adheres to Islamic principles while simultaneously addressing
17
X8X (Q=25th)
X8X (Q=50th)
X8X (Q=75th)
Fig. 12. The performance of the network in forecasting and validating the RECT-X8X case (Note: see note of Fig. 5).
Fig. 13. MAE for the prediction of dynamic total connectedness of RECT and each cryptocurrency-based digital innovation. Note: trainer is trainrp (Resilient
Backpropagation) approach.
18
Table 4 Table 5
RMSE for the prediction part of the network (trainer: trainrp (Resilient Back MAE for the prediction part of the network (trainer: trainlm (Levenberg-Mar
propagation) approach). quardt) approach).
Quantile 25th 50th 75th Quantile 25th 50th 75th
Bitcoin Green 0.40881 0.092771 0.22764 Bitcoin Green 0.22462 0.070617 0.11393
Bitcoin 4.0843 2.0261 4.6124 Bitcoin 2.9109 2.0261 3.2362
Dash Green 0.21062 0.074845 0.34576 Dash Green 0.10674 0.036544 0.25142
Dash 3.801 1.9149 3.8435 Dash 2.8929 1.5004 2.9015
EverGreenCoin 4.0416 2.8256 4.5879 EverGreenCoin 2.878 2.0031 3.2084
Greencoin 2.3436 1.7698 3.1519 Greencoin 1.7679 1.2953 2.255
Stellar 2.9436 1.585 3.2506 Stellar 2.3848 1.1254 2.3848
X8X 1.839 1.0513 2.2066 X8X 1.1806 0.63143 1.4391
Note: The Root Mean Squared Error (RMSE) is a holistic measure of predictive Note: see note of Table 3.
error that takes into account both the magnitude and direction of errors. It is
considered a robust metric for evaluating the accuracy and performance of a innovations by focusing on green, green-synthesized, Islamic, and con
model. The RMSE calculates the square root of the average of the squared dif
ventional cryptocurrencies. By applying sophisticated quantitative
ferences between the predicted values and the actual values. By incorporating
methodologies, such as the QVAR extended joint connectedness
squared differences, the RMSE places more emphasis on larger errors, providing
a more sensitive measure of model performance. approach and the Elman neural network, we gained invaluable insights
into the evolving dynamics of these relationships. These advanced
techniques enabled a thorough examination of the interdependencies
and fluctuations between green investments and the aforementioned
environmental concerns. types of cryptocurrencies. This analytical framework provided a
Additionally, the advent of cryptocurrencies, synthesized with a comprehensive understanding of the nuanced and evolving nature of
focus on environmental sustainability, has unveiled possibilities for these relationships. These findings offer profound insight into the impact
sustainable investment. These cryptocurrencies are expressly engi of different digital cryptocurrency-based innovation categories on green
neered to support initiatives that are environmentally friendly and investment. This study elucidates the intricate synergistic and distinct
frequently incorporate mechanisms such as carbon credits, tokenized dynamics within this complex nexus. These findings serve as a founda
renewable-energy assets, or the traceability of sustainable supply chains. tion for informed decision- and policymaking, thus facilitating the
An assessment of the impact of cryptocurrencies, synthesized with a
focus on environmental sustainability on environmentally conscious
investments, can enable the identification of innovative approaches for Table 6
financing and supporting projects that prioritize environmental RMSE for the prediction part of the network (trainer: trainlm (Levenberg-Mar
quardt) approach).
awareness.
In summary, comprehending the implications of various categories Quantile 25th 50th 75th
of cryptocurrencies on environmentally conscious investments is vital Bitcoin Green 0.40928 0.11949 0.21959
for facilitating the global transition toward a low-carbon economy. By Bitcoin 4.0841 2.8632 4.6124
embracing strategies for sustainable financing and examining the po Dash Green 0.20865 0.072594 0.3177
Dash 3.8068 1.9114 3.8213
tential synergies between cryptocurrencies and environmentally
EverGreenCoin 4.0416 2.838 4.5835
conscious initiatives, we can reveal new avenues for promoting positive Greencoin 2.3286 1.7628 3.1504
environmental change. Stellar 3.2506 1.586 3.2506
This study advances our understanding of the intricate and dynamic X8X 1.8102 1.0499 2.2064
interplay between green investments and digital crypto-based Note: see note of Table 4.
4.5
3.5
2.5
1.5
0.5
0
Bitcoin Green Bitcoin Dash Green Dash EverGreenCoin Greencoin Stellar X8X
25th 0.40881 4.0843 0.21062 3.801 4.0416 2.3436 2.9436 1.839
50th 0.092771 2.0261 0.074845 1.9149 2.8256 1.7698 1.585 1.0513
75th 0.22764 4.6124 0.34576 3.8435 4.5879 3.1519 3.2506 2.2066
Fig. 14. RMSE for the prediction of dynamic total connectedness of RECT and each cryptocurrency-based digital innovation. Note: see note of Fig. 13.
19
Fig. 15. MAE for the prediction of dynamic total connectedness of RECT and each cryptocurrency-based digital innovation. Note: trainer is trainlm (Levenberg-
Marquardt) approach.
Fig. 16. RMSE for the prediction of dynamic total connectedness of RECT and each cryptocurrency-based digital innovation. Note: see note of Fig. 15.
development of strategies to expedite the transition toward a sustainable and algorithmic proof-of-work) exhibited the highest levels of inter
and environmentally conscious economy. connectivity. This can be attributed to their robust presence in the fin
Our findings highlight Bitcoin and Stellar’s significant interconnec tech industry and their substantial market capitalization. These results
tedness with green investments. Bitcoin (a conventional cryptocurrency indicate that cryptocurrencies have emerged as influential facilitators of
characterized by mining activities and a proof-of-work consensus algo green investment by effectively aligning financial activities with sus
rithm) and Stellar (an Islamic cryptocurrency devoid of mining features tainability objectives. Furthermore, our analysis highlights the distinct
20
characteristics of DASHG and BITG. These cryptocurrencies provide the Moreover, our analysis reveals that adopting sustainable technolo
most accurate non-linear information on green energy and technology, gies contributes to environmental preservation and offers substantial
particularly in bear-and-bull markets. This suggests that digital crypto- economic returns to green investors, thus highlighting the dual benefits
based innovations effectively capture and reflect the evolving dy of eco-friendly investments. Our results suggest that allocating resources
namics between green investments and broader market conditions. to green technologies is linked to reduced risk profiles and improved
Their capacity to capture these dynamics implies their potential to guide long-term financial performance, thereby underscoring the value prop
investors and market participants to make informed decisions and sus ositions of such investments for environmentally aware investors. These
tainably navigate the digital asset landscape. green investors seek companies that can dramatically reduce operating
The fusion of green synthesis and digital assets, with a specific focus expenses and increase their profits.
on cryptocurrency-based innovations, has emerged as a compelling Our findings also have valuable implications for financial market
strategy for bolstering the adoption of clean energy. These innovations developers, who can leverage these insights to shape the design and
exhibit the inherent capacity to establish accurate and intricate non- functionality of digital asset platforms, thus creating ecosystems that
linear linkages between renewable energy and clean technology. This actively facilitate green investments. By integrating the knowledge ob
symbiosis is facilitated by the seamless integration of environmentally tained from this study into the development process, these platforms can
conscious blockchain and sustainable production processes. By har be customized to provide investors with opportunities to actively
nessing the power of these interconnected systems, the potential to drive contribute to renewable energy and clean technology projects. This, in
the transition toward a greener future is significantly amplified. This turn, can drive sustainable advancement in the digital asset industry.
paradigm shift not only reinforces the viability of clean energy initia This study provides valuable insight into the dynamic and non-linear
tives but also highlights the transformative impact of deploying digital nature of the relationship between digital cryptocurrency-based inno
assets as dynamic agents for advancing environmental sustainability. vation and green investments. These findings underscore the potential
The findings have significant implications for sustainability policy for leading cryptocurrency platforms to gain environmental awareness,
makers and financial market developers, particularly regulators, who thereby contributing to a more sustainable digital-asset environment.
play a pivotal role in shaping the sustainable landscape of digital assets. Future studies should extend the analysis to cover a longer time
Specifically, the insights gained from this study are highly relevant for horizon, spanning multiple market cycles, which would enable an ex
policymakers because they devise strategies and regulations to facilitate amination of the stability and robustness of the interconnections be
the seamless integration of green investments and environmentally tween digital crypto-based innovations and green investments under
friendly digital currencies. By gaining a comprehensive understanding diverse economic conditions. Moreover, future studies could broaden
of the interconnections and non-linear dynamics within the market, the scope of digital cryptocurrency-based innovations by exploring the
policymakers can formulate measures to foster the growth of ecofriendly inclusion of other innovative digital assets such as decentralized finance
digital assets and drive sustainable development. These policies tokens or emerging stablecoins. This expansion could enable the
encompass a range of incentives, guidelines, or frameworks that incen assessment of their impact on green investments and interconnections
tivize the use of green cryptocurrencies and facilitate their seamless with conventional digital cryptocurrency-based innovations.
integration into sustainable investment practices. Policymakers can also Furthermore, the incorporation of additional variables in the anal
explore regulatory approaches to ensure transparency, accountability, ysis may be beneficial. Factors such as regulatory development, investor
and environmental responsibility in the digital assets sphere. This in sentiment, and technological advancements in renewable energy and
cludes establishing frameworks that promote responsible innovation, clean technologies should be considered when examining the market-
foster collaboration among stakeholders, and align financial market driven nexus between digital cryptocurrency-based innovation and
practices with overarching sustainability objectives. Ultimately, these green investment. This broader perspective could offer a more
efforts would contribute to the broader objective of transitioning toward comprehensive understanding of the multifaceted dynamics at play.
a more sustainable and environmentally conscious economy. Our find Incorporating these suggestions into future research could contribute to
ings provide a valuable foundation for sustainability policymakers to a more nuanced and comprehensive understanding of the relationship
build and refine their strategies, leading to a future in which green in between digital cryptocurrency-based innovation and green in
vestments and digital crypto-based innovations work harmoniously to vestments. Continually expanding knowledge is expected to better
drive positive environmental changes. Further, regulators play a crucial inform decision-making and foster the development of sustainable and
role in establishing a sustainable sphere of influence over digital assets. resilient digital asset ecosystems.
Our findings underscore that it is crucial for regulators to consider the
interplay between different classes of digital cryptocurrency-based CRediT authorship contribution statement
innovation and green investment when formulating policies and regu
lations. We recommend that regulators establish reporting requirements Mahdi Ghaemi Asl: Writing – original draft, Software, Methodol
for green investments, fostering the development of eco-friendly digital ogy, Formal analysis, Conceptualization. Sami Ben Jabeur: Writing –
currencies and implementing mechanisms to track the environmental review & editing, Writing – original draft, Supervision, Formal analysis,
impact of digital-asset transactions. Thus, regulators can ensure that the Data curation, Conceptualization. Younes Ben Zaied: Writing – review
digital-asset industry operates in a manner that supports sustainable & editing, Writing – original draft, Software, Funding acquisition,
development and addresses environmental concerns. This study pro Formal analysis, Conceptualization.
vides valuable guidance for financial market developers and regulators
in shaping the future of digital assets. By incorporating sustainability Data availability
principles and practices into the design of digital asset platforms and by
establishing robust regulatory frameworks, we can create an ecosystem Data will be made available on request.
that fosters responsible and impactful green investments.
21
Appendix A
Estimation of Quantile Vector Autoregression

approach (QVAR)
Extraction of Extended Joint Connectedness

Measure
Applying the Elman Neural Network

(ENN)
Calculation of RMSE and MSE for

prediction of TCIs
Detection of non-linear information in each

interconnectedness
Fig. A1. Study design for clearing the process steps.
References Balcilar, M., Gabauer, D., Umar, Z., 2021. Crude oil futures contracts and commodity
markets: new evidence from a TVP-VAR extended joint connectedness approach.
Resour. Policy 73, 102219. https://doi.org/10.1016/j.resourpol.2021.102219.
Abakah, E.J.A., Tiwari, A.K., Sharma, A., Mwamtambulo, D.J., 2022. Extreme
Bossman, A., Gubareva, M., 2023. Asymmetric impacts of geopolitical risk on stock
connectedness between green bonds, government bonds, corporate bonds and other
markets: a comparative analysis of the E7 and G7 equities during the Russian-
asset classes: insights for portfolio investors. J. Risk Financ. Manag. 15 (10), 477.
Ukrainian conflict. Heliyon 9 (2), e13626.
Abakah, E.J.A., Ullah, G.W., Adekoya, O.B., Bonsu, C.O., Abdullah, M., 2023a.
Bossman, A., Gubareva, M., Teplova, T., 2023. Asymmetric effects of geopolitical risk on
Blockchain market and eco-friendly financial assets: dynamic price correlation,
major currencies: Russia-Ukraine tensions. Financ. Res. Lett. 51, 103440.
connectedness and spillovers with portfolio implications. Int. Rev. Econ. Finance 87,
Bouchmel, I., Ftiti, Z., Louhich, W., Omri, A., 2024. Financing sources, green investment,
218–243.
and environmental performance: cross-country evidence. J. Environ. Manag. 353,
Abakah, E.J.A., Tiwari, A.K., Ghosh, S., Doğan, B., 2023b. Dynamic effect of bitcoin,
120230 https://doi.org/10.1016/j.jenvman.2024.120230.
fintech and artificial intelligence stocks on eco-friendly assets, Islamic stocks and
Bouteska, A., Sharif, T., Abedin, M.Z., 2023. Volatility spillovers and other dynamics
conventional financial markets: another look using quantile-based approaches.
between cryptocurrencies and the energy and bond markets. Q. Rev. Econ. Financ.
Technol. Forecast. Soc. Change 192, 122566. https://doi.org/10.1016/j.
92, 1–13.
techfore.2023.122566.
Cainelli, G., D’Amato, A., Mazzanti, M., 2020. Resource efficient eco-innovations for a
Abdullah, M., Adeabah, D., Abakah, E.J.A., Lee, C.C., 2023. Extreme return and volatility
circular economy: Evidence from EU firms. Res. Policy 49, 103827. https://doi.org/
connectedness among real estate tokens, REITs, and other assets: the role of global
10.1016/j.respol.2019.103827.
factors and portfolio implications. Financ. Res. Lett. 56, 104062 https://doi.org/
Cao, S., Nie, L., Sun, H., Sun, W., Taghizadeh-Hesary, F., 2021. Digital finance, green
10.1016/j.frl.2023.104062.
technological innovation and energy-environmental performance: evidence from
Albitar, K., Al-Shaer, H., Liu, Y.S., 2023. Corporate commitment to climate change: the
China’s regional economies. J. Clean. Prod. 327, 129458 https://doi.org/10.1016/j.
effect of eco-innovation and climate governance. Res. Policy 52, 104697. https://
jclepro.2021.129458.
doi.org/10.1016/j.respol.2022.104697.
Chai, S., Zhang, K., Wei, W., Ma, W., Abedin, M.Z., 2022. The impact of green credit
Ali, F., Jiang, Y., Sensoy, A., 2021. Downside risk in Dow Jones Islamic equity indices:
policy on enterprises’ financing behavior: evidence from Chinese heavily-polluting
precious metals and portfolio diversification before and after the COVID-19 bear
listed companies. J. Clean. Prod. 363, 132458 https://doi.org/10.1016/j.
market. Res. Int. Bus. Financ. 58, 101502.
jclepro.2022.132458.
Ali, F., Bouri, E., Naifar, N., Shahzad, S.J.H., AlAhmad, M., 2022. An examination of
Chatziantoniou, I., Abakah, E.J.A., Gabauer, D., Tiwari, A.K., 2022. Quantile
whether gold-backed Islamic cryptocurrencies are safe havens for international
time–frequency price connectedness between green bond, green equity, sustainable
Islamic equity markets. Res. Int. Bus. Financ. 63, 101768.
investments and clean energy markets. J. Clean. Prod. 361, 132088.
Ali, S., Yousaf, I., Vo, X.V., 2023. Comovements and hedging effectiveness between
Chatziantoniou, I., Gabauer, D., Stenfors, A., 2021. Interest rate swaps and the
conventional and Islamic cryptocurrencies: evidence from the COVID-19 pandemic.
transmission mechanism of monetary policy: a quantile connectedness approach.
Int. J. Emerg. Mark. https://doi.org/10.1108/IJOEM-10-2021-1571.
Econ. Lett. 204, 109891 https://doi.org/10.1016/j.econlet.2021.109891.
Ando, T., Greenwood-Nimmo, M., Shin, Y., 2022. Quantile connectedness: modeling tail
Chiappini, H., Vento, G., De Palma, L., 2021. The impact of COVID-19 lockdowns on
behavior in the topology of financial networks. Manag. Sci. 68, 2401–2431. https://
sustainable indexes. Sustainability 13 (4), 1846.
doi.org/10.1287/mnsc.2021.3984.
Chou, S.C., Li, Z.A., Wang, T., Yen, J.C., 2023. How the quality of initial coin offering
Annamalaisamy, B., Vepur Jayaraman, S., 2023. Do cryptocurrencies integrate with the
white papers influences fundraising: using security token offerings white papers as a
indices of equity, sustainability, clean energy, and crude oil? A wavelet coherency
benchmark. Intell. Syst. Account. Finance Manag. 30 (1), 3–18.
approach. Int. J. Finance Econ. 29 (3), 3372–3392. https://doi.org/10.1002/
Chuen, D.L.K., Guo, L., Wang, Y., 2017. Cryptocurrency: a new investment opportunity?
ijfe.2843.
J. Altern. Invest. 20 (3), 16–40.
Anwer, Z., Farid, S., Khan, A., Benlagha, N., 2023. Cryptocurrencies versus
Conlon, T., McGee, R., 2020. Safe haven or risky hazard? Bitcoin during the COVID-19
environmentally sustainable assets: does a perfect hedge exist? Int. Rev. Econ.
bear market. Financ. Res. Lett. 35, 101607.
Finance 85, 418–431. https://doi.org/10.1016/j.iref.2023.02.005.
Corbet, S., Meegan, A., Larkin, C., Lucey, B., Yarovaya, L., 2018. Exploring the dynamic
Arfaoui, N., Naeem, M.A., Boubaker, S., Mirza, N., Karim, S., 2023. Interdependence of
relationships between cryptocurrencies and other financial assets. Econ. Lett. 165,
clean energy and green markets with cryptocurrencies. Energy Econ. 120, 106584.
28–34.
Asl, M.G., Jabeur, S.B., 2024. Could the Russia-Ukraine war stir up the persistent
Corbet, S., Lucey, B., Yarovaya, L., 2021. Bitcoin-energy markets interrelationships - new
memory of interconnectivity among Islamic equity markets, energy commodities,
evidence. Resour. Policy 70, 101916. https://doi.org/10.1016/j.
and environmental factors? Res. Int. Bus. Finance 69, 102260. https://doi.org/
resourpol.2020.101916.
10.1016/j.ribaf.2024.102260.
Cunado, J., Chatziantoniou, I., Gabauer, D., de Gracia, F.P., Hardik, M., 2023. Dynamic
Attarzadeh, A., Balcilar, M., 2022. On the dynamic return and volatility connectedness of
spillovers across precious metals and oil realized volatilities: evidence from quantile
cryptocurrency, crude oil, clean energy, and stock markets: a time-varying analysis.
extended joint connectedness measures. J. Commod. Mark. 30, 100327.
Environ. Sci. Pollut. Res. 29 (43), 65185–65196.
Dabbous, A., Barakat, K.A., Tarhini, A., 2024. Digitalization, crowdfunding, eco-
innovation and financial development for sustainability transitions and sustainable
22
competitiveness: insights from complexity theory. J. Innov. Knowl. 9, 100460 Mnif, E., Mouakhar, K., Jarboui, A., 2023. Energy-conserving cryptocurrency response
https://doi.org/10.1016/j.jik.2023.100460. during the COVID-19 pandemic and amid the Russia–Ukraine conflict. J. Risk
De Vries, A., 2023. Cryptocurrencies on the road to sustainability: Ethereum paving the Financ. 24 (2), 169–185.
way for bitcoin. Patterns 4. Moghul, U.F., Safar-Aly, S.H., 2014. Green sukuk: the introduction of Islam’s
Diebold, F.X., Yilmaz, K., 2012. Better to give than to receive: predictive directional environmental ethics to contemporary Islamic finance. Georget. Int. Environ. Law
measurement of volatility spillovers. Int. J. Forecast. 28, 57–66. https://doi.org/ Rev. 27, 1.
10.1016/j.ijforecast.2011.02.006. Moldan, B., Janoušková, S., Hák, T., 2012. How to understand and measure
Dogan, E., Majeed, M.T., Luni, T., 2022. Are clean energy and carbon emission environmental sustainability: indicators and targets. Ecol. Indic. 17, 4–13.
allowances caused by bitcoin? A novel time-varying method. J. Clean. Prod. 347, Naeem, M.A., Karim, S., 2021. Tail dependence between bitcoin and green financial
131089 https://doi.org/10.1016/j.jclepro.2022.131089. assets. Econ. Lett. 208, 110068.
Drucker, H., Wu, D., Vapnik, V.N., 1999. Support vector machines for spam Ndubuisi, G., Urom, C., 2023. Dependence and risk spillovers among clean
categorization. IEEE Trans. Neural Netw. 10 (5), 1048–1054. cryptocurrencies prices and media environmental attention. Res. Int. Bus. Financ.
Elman, J.L., 1990. Finding structure in time. Cognit. Sci. 14, 179–211. https://doi.org/ 65, 101953.
10.1016/0364-0213(90)90002-E. Omane-Adjepong, M., Alagidede, I.P., 2020. Dynamic linkages and economic role of
Ghisetti, C., Quatraro, F., 2017. Green technologies and environmental productivity: a leading cryptocurrencies in an emerging market. Asia-Pacific Finan. Markets. 27 (4),
cross-sectoral analysis of direct and indirect effects in Italian regions. Ecol. Econ. 537–585.
132, 1–13. Omane-Adjepong, M., Ababio, K.A., Alagidede, I.P., 2019. Time-frequency analysis of
Goodland, R., 1995. The concept of environmental sustainability. Annu. Rev. Ecol. Syst. behaviourally classified financial asset markets. Res. Int. Bus. Financ. 50, 54–69.
1–24. Patel, R., Kumar, S., Bouri, E., Iqbal, N., 2023. Spillovers between green and dirty
Hassanein, A., Mostafa, M.M., 2023. Bibliometric network analysis of thirty years of cryptocurrencies and socially responsible investments around the war in Ukraine.
islamic banking and finance scholarly research. Qual. Quant. 57 (3), 1961–1989. Int. Rev. Econ. Finance 87, 143–162.
Hastie, T., Tibshirani, R., Friedman, J.H., Friedman, J.H., 2009. The Elements of Pesaran, H.H., Shin, Y., 1998. Generalized impulse response analysis in linear
Statistical Learning: Data Mining, Inference, and Prediction, Vol. 2. Springer, New multivariate models. Econ. Lett. 58, 17–29. https://doi.org/10.1016/S0165-1765
York, pp. 1–758. (97)00214-0.
Haykin, S., 1998. Neural Networks: A Comprehensive Foundation. Prentice Hall PTR. Pham, L., 2019. Do all clean energy stocks respond homogeneously to oil price? Energy
He, G., Zhu, P., Wu, C., 2024. Financial innovation, green investment, and sustainable Econ. 81, 355–379. https://doi.org/10.1016/j.eneco.2019.04.010.
mineral extractions in China. Resour. Policy 90, 104696. https://doi.org/10.1016/j. Zhao, X., Zeng, B., Zhao, X., Zeng, S., Jiang, S., 2024. Impact of green finance on green
resourpol.2024.104696. energy efficiency: a pathway to sustainable development in China. J. Clean. Prod.
Hochreiter, S., Schmidhuber, J., 1997. Long short-term memory. Neural Comput. 9 (8), 450, 141943.
1735–1780. Pham, L., Huynh, T.L.D., Hanif, W., 2023. Time-varying asymmetric spillovers among
Huang, Y., Ahmad, M., Ali, S., Kirikkaleli, D., 2022. Does eco-innovation promote cleaner cryptocurrency, green and fossil-fuel investments. Glob. Finance J. 58, 100891
energy? Analyzing the role of energy price and human capital. Energy 239, 122268. https://doi.org/10.1016/j.gfj.2023.100891.
https://doi.org/10.1016/j.energy.2021.122268. Pilloni, M., Kádár, J., Abu Hamed, T., 2022. The impact of COVID-19 on energy start-up
Hung, N.T., 2023. Green investment, financial development, digitalization and economic companies: the use of global financial crisis (GFC) as a lesson for future recovery.
sustainability in Vietnam: evidence from a quantile-on-quantile regression and Energies 15 (10), 3530.
wavelet coherence. Technol. Forecast. Soc. Change 186, 122185. https://doi.org/ Rashid, A., Siddique, M.A., 2021. Circular economy, green economy, and sustainable
10.1016/j.techfore.2022.122185. development: Establishing the interconnections and discoursing the role of Islamic
Irfan, M., Rehman, M.A., Nawazish, S., Hao, Y., 2023. Performance analysis of gold-and finance. In: Islamic Finance and Circular Economy: Connecting Impact and Value
fiat-backed cryptocurrencies: risk-based choice for a portfolio. J. Risk Financ. Creation. Springer Singapore, Singapore, pp. 53–72.
Manag. 16 (2), 99. Raudys, S.J., Jain, A.K., 1991. Small sample size effects in statistical pattern recognition:
Jabeur, S.B., Serret, V., 2023. Bankruptcy prediction using fuzzy convolutional neural recommendations for practitioners. IEEE Trans. Pattern Anal. Mach. Intell. 13 (3),
networks. Res. Int. Bus. Finance 64, 101844. 252–264.
Jacquet, P., Mans, B., 2019, April. Green mining: Toward a less energetic impact of Ren, B., Lucey, B., 2022a. Do clean and dirty cryptocurrency markets herd differently?
cryptocurrencies. In: IEEE Infocom 2019-IEEE Conference on Computer Financ. Res. Lett. 47, 102795.
Communications Workshops (INFOCOM WKSHPS). IEEE, pp. 210–215. Ren, B., Lucey, B., 2022b. A clean, green haven?—examining the relationship between
Jia, H., Taheri, B., 2021. Model identification of solid oxide fuel cell using hybrid Elman clean energy, clean and dirty cryptocurrencies. Energy Econ. 109, 105951.
neural network/quantum pathfinder algorithm. Energy Rep. 7, 3328–3337. https:// Rizvi, S.K.A., Naqvi, B., Mirza, N., Umar, M., 2022. Safe haven properties of green,
doi.org/10.1016/j.egyr.2021.05.070. Islamic, and crypto assets and investor’s proclivity towards treasury and gold.
Jiang, F., Zhu, Q., Yang, J., Chen, G., Tian, T., 2022. Clustering-based interval prediction Energy Econ. 115, 106396 https://doi.org/10.1016/j.eneco.2022.106396.
of electric load using multi-objective pathfinder algorithm and Elman neural Rogers, E.M., Singhal, A., Quinlan, M.M., 2014. Diffusion of innovations. In: An
network. Appl. Soft Comput. 129, 109602 https://doi.org/10.1016/j. Integrated Approach to Communication Theory and Research. Routledge,
asoc.2022.109602. pp. 432–448.
Jordan, M., 1986. Attractor dynamics and parallelism in a connectionist sequential Ruiz, L.G.B., Rueda, R., Cuéllar, M.P., Pegalajar, M.C., 2018. Energy consumption
machine. In: Eighth Annual Conference of the Cognitive Science Society, 1986, forecasting based on Elman neural networks with evolutive optimization. Expert
pp. 513–546. Syst. Appl. 92, 380–389. https://doi.org/10.1016/j.eswa.2017.09.059.
Kamal, J.B., Hassan, M.K., 2022. Asymmetric connectedness between cryptocurrency Rumelhart, D.E., Hinton, G.E., Williams, R.J., 1986. Learning representations by back-
environment attention index and green assets. J. Econ. Asymmetries 25, e00240. propagating errors. Nature 323 (6088), 533–536.
Khalfaoui, R., Gozgor, G., Goodell, J.W., 2023. Impact of Russia-Ukraine war attention Schilling, L., Fernández-Villaverde, J., Uhlig, H., 2024. Central bank digital currency:
on cryptocurrency: evidence from quantile dependence analysis. Financ. Res. Lett. when price and bank stability collide. J. Monet. Econ. https://doi.org/10.1016/j.
52, 103365. jmoneco.2024.01.007.
Koop, G., Pesaran, M.H., Potter, S.M., 1996. Impulse response analysis in nonlinear Sharif, A., Brahim, M., Dogan, E., Tzeremes, P., 2023. Analysis of the spillover effects
multivariate models. J. Econ. 74, 119–147. https://doi.org/10.1016/0304-4076(95) between green economy, clean and dirty cryptocurrencies. Energy Econ. 120,
01753-4. 106594.
Le, T.L., Abakah, E.J.A., Tiwari, A.K., 2021. Time and frequency domain connectedness Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., Salakhutdinov, R., 2014.
and spill-over among fintech, green bonds and cryptocurrencies in the age of the Dropout: a simple way to prevent neural networks from overfitting. J. Mach. Learn.
fourth industrial revolution. Technol. Forecast. Soc. Change 162, 120382. Res. 15 (1), 1929–1958.
Li, G., Wu, H., Jiang, J., Zong, Q., 2023. Digital finance and the low-carbon energy Sun, B., Li, J., Zhong, S., Liang, T., 2023. Impact of digital finance on energy-based
transition (LCET) from the perspective of capital-biased technical progress. Energy carbon intensity: evidence from mediating effects perspective. J. Environ. Manage.
Econ. 120, 106623 https://doi.org/10.1016/j.eneco.2023.106623. 327, 116832 https://doi.org/10.1016/j.jenvman.2022.116832.
Li, J., Li, N., Peng, J., Cui, H., Wu, Z., 2019. Energy consumption of cryptocurrency Syed, A.A., Ahmed, F., Kamal, M.A., Ullah, A., Ramos-Requena, J.P., 2022. Is there an
mining: a study of electricity consumption in mining cryptocurrencies. Energy 168, asymmetric relationship between economic policy uncertainty, cryptocurrencies,
160–168. https://doi.org/10.1016/j.energy.2018.11.046. and global green bonds? Evidence from the United States of America. Mathematics
Madaleno, M., Taskin, D., Dogan, E., Tzeremes, P., 2023. A dynamic connectedness 10 (5), 720.
analysis between rare earth prices and renewable energy. Resour. Policy 85, 103888. Symitsi, E., Chalvatzis, K.J., 2018. Return, volatility and shock spillovers of bitcoin with
Maheu, J.M., McCurdy, T.H., Song, Y., 2021. Bull and bear markets during the COVID-19 energy and technology companies. Econ. Lett. 170, 127–130. https://doi.org/
pandemic. Financ. Res. Lett. 42, 102091. 10.1016/j.econlet.2018.06.012.
Medsker, L.R., 2012. Hybrid Intelligent Systems. Springer Science & Business Media. Tian, J., Sun, S., Cao, W., Bu, D., Xue, R., 2024. Make every dollar count: the impact of
Mensi, W., Shafiullah, M., Vo, X.V., Kang, S.H., 2022. Spillovers and connectedness green credit regulation on corporate green investment efficiency. Energy Econ. 130,
between green bond and stock markets in bearish and bullish market scenarios. 107307 https://doi.org/10.1016/j.eneco.2024.107307.
Financ. Res. Lett. 49, 103120. Tiwari, A.K., Abakah, E.J.A., Dwumfour, R.A., Mefteh-Wali, S., 2022. Connectedness and
Migliorelli, M., 2021. What do we mean by sustainable finance? Assessing existing directional spillovers in energy sectors: international evidence. Appl. Econ. 54,
frameworks and policy risks. Sustainability 13 (2), 975. 2554–2569. https://doi.org/10.1080/00036846.2021.1998326.
Mnif, E., Mouakhar, K., Jarboui, A., 2022. The co-movements of faith-based Tiwari, A.K., Abakah, E.J.A., Yaya, O.S., Appiah, K.O., 2023. Tail risk dependence, co-
cryptocurrencies in periods of pandemics. Rev. Financ. Econ. 40 (3), 300–311. movement and predictability between green bond and green stocks. Appl. Econ. 55,
201–222. https://doi.org/10.1080/00036846.2022.2085869.
23
Valero-Gil, J., Surroca, J.A., Tribo, J.A., Gutierrez, L., Montiel, I., 2023. Innovation vs. Ye, W., Wong, W.-K., Arnone, G., Nassani, A.A., Haffar, M., Faiz, M.F., 2023b. Crypto
standardization: the conjoint effects of eco-innovation and environmental currency and green investment impact on global environment: a time series analysis.
management systems on environmental performance. Res. Policy 52, 104737. Int. Rev. Econ. Finance 86, 155–169. https://doi.org/10.1016/j.iref.2023.01.030.
https://doi.org/10.1016/j.respol.2023.104737. Yongchun, L., 2010. Application of Elman neural network in short-term load forecasting.
Wang, X., Xu, X., 2024. Sustainable resource management and green economic growth: a In: 2010 International Conference on Artificial Intelligence and Computational
global prospective. Res. Policy 89, 104634. Intelligence, 2. IEEE, pp. 141–144. October.
Wang, Y., Zhao, Z., Zhang, S., Su, Y., 2023a. Research on the impact of digital inclusive Yousaf, I., Yarovaya, L., 2022. Spillovers between the Islamic gold-backed
finance on regional carbon emissions: based on the sustainable green innovation of cryptocurrencies and equity markets during the COVID-19: a sectorial analysis. Pac.
small and medium-sized enterprises. J. Clean. Prod. 428, 139513 https://doi.org/ Basin Financ. J. 71, 101705.
10.1016/j.jclepro.2023.139513. Yu, H., Wei, W., Li, J., Li, Y., 2022. The impact of green digital finance on energy
Wang, Z., Samuel, R.-N., Chen, X., Xu, B., Huang, W., 2023b. Central bank digital resources and climate change mitigation in carbon neutrality: case of 60 economies.
currencies: consumer data-driven sustainable operation management policy. Resour. Policy 79, 103116. https://doi.org/10.1016/j.resourpol.2022.103116.
Technol. Forecast. Soc. Change 196, 122867. https://doi.org/10.1016/j. Zhang, D., 2023. Can digital finance empowerment reduce extreme ESG hypocrisy
techfore.2023.122867. resistance to improve green innovation? Energy Econ. 125, 106756 https://doi.org/
Wani, T.A., Ali, S., 2015. Innovation diffusion theory. J. Gen. Manag. Res. 3, 101–118. 10.1016/j.eneco.2023.106756.
Wasiuzzaman, S., Azwan, A.N.M., Nordin, A.N.H., 2023. Analysis of the performance of
Islamic gold-backed cryptocurrencies during the bear market of 2020. Emerg. Mark.
Mahdi Ghaemi Asl received a Ph.D. degree in Economics from the Faculty of Economics
Rev. 54, 100920.
and Business Administration, the Ferdowsi University of Mashhad, Mashhad, Iran, in
Werbos, P.J., 1988. Generalization of backpropagation with application to a recurrent
2015. In 2016, he joined the Kharazmi University, Tehran, Iran as an official faculty
gas market model. Neural Netw. 1 (4), 339–356.
member at the Faculty of Economics. His research interests include modeling and simu
Witten, D., James, G., 2013. An Introduction to Statistical Learning with Applications in
lation (M&S), multi-market optimization, FinTech, and industrial organization.
R. Springer Publication.
Xie, K., Yi, H., Hu, G., Li, L., Fan, Z., 2020. Short-term power load forecasting based on
Elman neural network with particle swarm optimization. Neurocomputing 416, Sami Ben Jabeur is an associate professor at ESDES Business School (Lyon, France). He
136–142. https://doi.org/10.1016/j.neucom.2019.02.063. obtained his Ph.D. degree in finance in 2011 at the University of Toulon-Var. His main
Xu, D., Hu, X., Wang, W., Chau, K., Zang, H., Wang, J., 2024. A new hybrid model for research fields are bankruptcy prediction, machine learning models and environmental
monthly runoff prediction using ELMAN neural network based on decomposition- modeling. He has published in different academic journals, such as Journal of the Opera
integration structure with local error correction method. Expert Syst. Appl. 238, tional Research Society, Finance Research Letters, Journal of Retailing and Consumer
121719 https://doi.org/10.1016/j.eswa.2023.121719. Services, Technological Forecasting and Social Change, Environmental Modeling &
Xu, J., Yu, Y., Zhang, M., Zhang, J.Z., 2023. Impacts of digital transformation on eco- Assessment, Environmental and Resource Economics, Journal of Environmental Man
innovation and sustainable performance: evidence from Chinese manufacturing agement and other ranked journals.
companies. J. Clean. Prod. 393, 136278.
Yang, F., Luo, C., Pan, L., 2024. Do digitalization and intellectual capital drive
Younes Ben Zaied is a professor in Finance at the EDC Paris Business School in France. He
sustainable open innovation of natural resources sector? Evidence from China.
has published several papers in leading international journals. His-research interest fo
Resour. Policy 88, 104345. https://doi.org/10.1016/j.resourpol.2023.104345.
cuses especially on the sustainable development goals achievement, innovation, and green
Ye, F., Ouyang, Y., Li, Y., 2023a. Digital investment and environmental performance: the
investment. Research papers have been published in Technological forecasting and social
mediating roles of production efficiency and green innovation. Int. J. Prod. Econ.
change, annals of operation research, finance research letters, energy economics.
259, 108822 https://doi.org/10.1016/j.ijpe.2023.108822.
24

1 s2.0 S0040162524005134 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S0040162524005134 Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0040162524005134 Main

Uploaded by

Copyright:

Available Formats

Technological Forecasting & Social Change 208 (2024) 123715

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

Analyzing the interplay between eco-friendly and Islamic digital currencies

1. Introduction the carbon impact of blockchain technology. Many of these elements

1∑ k Elman networks are constructed using the backpropagation (BP)

Fig. 2. Plot of returns.

Bitcoin Green-RECT Bitcoin Green RECT FROM

To begin our study, we considered the start dates of the respective

Bitcoin Green (Q=25th)

Bitcoin Green (Q=50th)

Bitcoin Green (Q=75th)

Dash Green (Q=25th)

Dash Green (Q=50th)

Dash Green (Q=75th)

Estimation of Quantile Vector Autoregression

Extraction of Extended Joint Connectedness

Applying the Elman Neural Network

Calculation of RMSE and MSE for

Detection of non-linear information in each

Fig. A1. Study design for clearing the process steps.

You might also like