Intervention for Cryptocurrency Emissions: A China

Case Study
Scott Fan, Elliot Gyllensvärd, Erich Farkas, Julian Schutzner

To cite this version:

Scott Fan, Elliot Gyllensvärd, Erich Farkas, Julian Schutzner. Intervention for Cryptocurrency Emis-
sions: A China Case Study. 2022. �hal-03737234�

HAL Id: hal-03737234

https://hal.science/hal-03737234
Preprint submitted on 23 Jul 2022

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 Intervention for Cryptocurrency Emissions: A China Case
Study
Scott Fan Elliot Gyllensvärd Erich Farkas
Julian Schutzner

Department of Economics
University College London, UK
July 23, 2022

Abstract
This paper seeks to analyse the environmental effects of China’s recent regulatory steps taken
against cryptocurrencies and compare them with two policy alternatives. To evaluate the environ-
mental impact, two baseline scenarios are used - a ban scenario and a no-ban scenario - which are
then employed to compare China’s intervention with two alternative policies; an emissions trading
system or carbon taxes. It is estimated that China’s decision to ban cryptocurrencies will result
in a 25.7% reduction in CO2 emissions from the global Bitcoin network between mid-2021 and
2030. In comparison, under a hypothetical emission trading system (ETS) and carbon tax inter-
vention the most stringent scenario is estimated to reduce global Bitcoin CO2 emissions by 2.9%
and 8.5% between mid-2021 and 2030, respectively. Furthermore, this paper shows that the ETS
and carbon tax are both restricted in their absolute ability to reduce CO2 emissions due to the
difficulties associated with the practical implementation of such policies. This provides evidence
for why, in terms of an environmental perspective, a cryptocurrency ban is the most effective policy
in reducing the Bitcoin blockchain network’s CO2 emissions. However, the paper also shows that
the transition to Proof-of-Stake (PoS) blockchains may create an environment in which there is
less of an argument for active government intervention in the cryptocurrency markets due to the
protocol’s high energy efficiency.

1 Introduction
Cryptocurrencies have seen rising levels of popularity in the global financial system over the recent
years, but their regulation has remained largely heterogeneous across countries. Regulation has become
a pressing issue among governments due to various issues, including the environmental impact of
cryptocurrencies. The underlying technology can use significant amounts of energy, depending on the
consensus protocol used. As a result, Bitcoin’s Proof of Work (PoW) protocol has become a focal
point of discussion among regulators, given its substantial power consumption.
This paper aims to analyse the environmental effects of China’s recent regulatory steps against
cryptocurrencies, and compare them with two alternative intervention policies; an emissions trading
system and carbon taxes. Consequently, the environmental impact of Proof of Work (PoW) and Proof
of Stake (PoS) chains will be analysed through the examples of Bitcoin and Ethereum1 respectively.

1.1 Blockchain Primer

Blockchain technology builds the foundation for cryptocurrencies. A blockchain is a distributed ledger
that keeps continuously updating records of digital transactions. There are both permissionless and
permissioned blockchains, but the focus of this paper will be on permissionless blockchains where
anyone can participate in the network and participants validate new transactions.
To validate transactions on the blockchain, cryptocurrencies use different consensus algorithms. The
two primary consensus algorithms are PoW and PoS. PoW is utilised by the Bitcoin blockchain while
1 Ethereum is upgrading to PoS from PoW in Q3 2022.

1
Ethereum is currently transitioning from PoW to PoS. Amongst other advantages, the predicted decline
in energy usage of more than 99.95% under PoS has often been cited as one of the main considerations
for Ethereum’s transition (Beekhuizen, 2021). PoS will be further explained in Section 3.4.
Under PoW, anyone with enough computing power can participate in validating transactions on the
network. However, to successfully validate a transaction, one needs to compete against other miners
in terms of computing power to solve cryptographic puzzles to add new blocks and receive a reward.
The computing power is often referred to as hash rate which is the amount of computing power being
contributed to the network through mining to process transactions.
On the Bitcoin blockchain (and on most other blockchains), every transaction is made publicly
available for any participant to view, each participant then processes recent transactions in order of its
timestamped hash to keep a time-ordered ledger. A hash is a mathematical function that takes a file
and produces a relatively short code that can be used to identify that file. Bitcoin miners verify the
transaction and add it to the next block using the cryptographic hash value of the previous block. The
previous block’s hash value is hidden from the miner who has to solve it by brute force. Miners keep
trying over and over to generate new hashes, until they find one that fulfils the requirement. Once
the hash of the previous block is computed, it is declared to the network which verifies it and a new
block is added to the blockchain. The miner who solved the PoW puzzle then updates the blockchain
and gets a reward in Bitcoin which is created in accordance with the protocol whereby every 210,000
blocks (around every four years) the block reward is halved.
For the puzzle to be solved and rewarded, significant computing power is required to create a
hash that follows the blockchain’s format - a process which demands significant amounts of energy.
Furthermore, miners who were not successful do not receive any compensation and the energy used to
compete is essentially wasted.
Given the high energy consumption and ever-improving mining technology which has to be updated
by miners, there is a significant amount of resources required which has a corresponding impact on the
environment. According to the Cambridge Bitcoin Energy Consumption Index, the Bitcoin network
has an annualised energy consumption of 128.87 TWh as of 27 February 20222 (Cambridge, 2022), more
than the entire annual energy consumption of the Netherlands. It is not only the energy consumption
that concerns policymakers given recent turbulence in the technology supply chain. To compete in
the mining process, specialised hardware referred to as ASIC (Application Specific Integrated Circuit)
Miner is commonly used to perform the computations required for Bitcoin’s PoW consensus algorithm.
In the light of the recent “chip shortage” across the world, the demand for ASICs disrupts supply chains
while technological advancement of mining devices requires miners to update their devices (average
lifespan 1.29 years) to compete, creating 30.7 metric kilotons of e-waste per year (de Vries & Stoll,
2021). This, however, will not be tackled in this report and instead the focus will be on quantifying
the extent of electricity consumption and carbon emissions.
Whilst PoW requires a significant amount of computing resources, PoS offers a more power efficient
and scalable alternative. The consensus algorithm in PoS relies on the stakeholders within the system
to validate transactions rather than competing miners. Validators are algorithmically chosen and
don’t need to use large amounts of computing power. In the case of Ethereum, full validators have
to deposit a stake of 32 ETH (equivalent currency unit) and can then create blocks when they are
randomly chosen or validate proposed blocks when they’re not (Ethereum, 2022). If a validator attests
to a malicious block, his stake is lost.
Platt et al.(2021) find that the Bitcoin PoW algorithm’s energy consumption is “at least three orders
of magnitude higher than the highest consuming PoS system”, which will have to be considered by
policymakers going forward. To evaluate China’s policy decisions, this paper begins by understanding
China’s relevance in the world of cryptocurrency.

2 There is substantial estimation uncertainty in such estimates, given that the energy consumption information from

all nodes of the Bitcoin network, equipment and datacentres are not widely available.

2
1.2 China’s Place In The World of Cryptocurrency
In terms of cryptocurrency use, Chinese people used to lead the adoption. In 2013 China had already
surpassed the rest of the world with 84,000 Bitcoin wallets being created in a single month, about 30
percent of the global total for the month (Romann, 2013). Bitcoin exchanges such as BTC China were
also the world’s highest-volume exchanges in 2013 (Southurst, 2013).
The local popularity has largely been attributed to strict capital controls on the yuan and BTC
China’s no-fee trading. Capital controls make it difficult to invest or spend yuan abroad, while Bitcoin
can easily be transferred to a wallet in a foreign country and exchanged for local currency.
The success of Chinese cryptocurrency miners was largely driven by the abundance of cheap elec-
tricity and cheap land in north and western provinces (von Carnap, 2021) and the availability of
market-leading ASIC hardware. It has been estimated that the Chinese ASIC provider Bitmain had
a market share of 70-80 percent before the regulatory crackdown (Cheng, 2021). This combination
of cheap resources and direct access to leading technology put mining companies at a competitive
advantage. In 2018 Chinese mining pools alone accounted for more than 60% of the Bitcoin network’s
total hash power (Kaiser, Jurado, & Ledger, 2018). Table 9 in Section 2.3.8.1 demonstrates the cost
advantage Chinese miners had over the last years, with almost 40% lower energy prices in certain peri-
ods. Consequently, mining operations in China dominated the rest of the world until recent regulations
banned operators from mining in June 2021 (Shen & Galbraith, 2021b).

1.3 China’s Regulatory Steps

Whilst Bitcoin enjoyed widespread popularity among the Chinese population, the government became
increasingly concerned with its growing influence over time. Cryptocurrencies allow citizens to cir-
cumvent rules imposed by the government such as capital controls and threaten the central role of the
Chinese Yuan (CNY) whilst the growing energy consumption by miners constrained local governments
during China’s 2020 power crunch.
The first regulatory step that was taken by the People’s Bank of China (PBOC) occurred in
2013, which banned banks from holding Bitcoin to protect the status of the CNY. However, the most
significant restrictions were imposed from 2017 onwards. In 2017 both initial coin offerings (ICOs)
and cryptocurrency exchanges were banned from China. In April 2019, a notice issued by the Chinese
economic planning agency first discussed a possible ban on Bitcoin mining in the future (BBC, 2019).
The notice eventually turned into reality in June 2021 when Chinese authorities banned crypto mining
and started a countrywide crackdown on mining operations (Shen & Galbraith, 2021b) to eventually
ban all cryptocurrency transactions the following September. The ban on all transactions was justified
by the following statement:
Virtual currency-related business activities are illegal financial activities,” the People’s
Bank of China said, warning it ”seriously endangers the safety of people’s assets” (BBC,
2021a).
However, the concerns are not only centred around fraud and money laundering but also the
government’s push for its e-CNY, and environmental concerns which intensified during China’s power
shortage in 2021 (BBC, 2021b). This paper does not explore China’s regulatory policy and agenda in-
depth, and instead focuses on the comparison of China’s current approach with alternative approaches
while assuming an underlying environmental motive behind the policy.

1.4 Research Aims

This paper focuses on addressing the key points listed below:

1. Understanding the impact of Chinese intervention against a hypothetical case of no intervention.

2. Investigating the dynamics and impact of a hypothetical Emissions Trading System (ETS) di-
rected toward the Bitcoin industry.
3. Understanding the impact of a hypothetical carbon emission tax implemented in the Bitcoin
industry.
4. Understanding what the transition to PoS will mean for blockchain mining energy consumption.

3
 5. Comparing these approaches, and understanding if there were other alternatives for Chinese
intervention.

2 Methodology
2.1 Literature Review of Interventions
In recent years, the most popular policy interventions to successfully limit emissions and incentivise
industries and consumers to consider their consumption of carbon, have come through carbon pricing.
Before that, it is important to understand the historical Chinese energy mix and subsequently its
energy goals.

2.1.1 Motivation for China’s Intervention

Historically, China’s energy mix has been heavily dominated by coal and it still is today. As seen from
Figure 1, most of China’s historical growth in energy has been fuelled by its relative abundance of coal
resources. This will contribute to a more polluting mix (see Section 2.3.4.1) but relative cheapness.
This relative cheapness contributes to the dominance of Chinese miners in the Bitcoin mining scene,
where mining profitability is a function of the cost of electricity.
By 2020, coal still represents 56.8% of energy consumption by source (ChinaPower, 2022). However,
China has become the top global investor in energy transition (ChinaPower, 2022) and renewable energy
resources are becoming increasingly important to meet its energy goals.

Figure 1: China and US Energy Mix From Our World in Data.

2.1.2 Energy Goals

During COP26, China and the US agreed on a joint declaration on enhancing climate action in the
2020s. As the largest emitter of CO2 , this was an important step in the right direction. Although there
are still concerns such as China’s reluctance to tackle its use of coal, Chinese authorities announced
on October 24:
”By 2030, China’s carbon dioxide emissions will peak, stabilize and then decline, and by
2060, China will be carbon neutral and have fully established a green, low-carbon and
circular economy, it said, reiterating the country’s previous pledge” (The State Council,
2021).
Table 1 summarises important Chinese energy goals going forward, suggesting a conflict between the
substantial use of energy in cryptocurrency mining and its energy and carbon emission intensity targets.

4
 Key Performance indicator Targets Target Year
Energy Intensity (Energy consumed per 13.5% lower than 2020 level 2025
Yuan of GDP)
Carbon Emission Intensity (CO2 emit- 65% lower than 2005 levels 2030
ted per Yuan of GDP)
Non-fossil fuels’ share in energy con- Around 25% 2030
sumption mix
Non-fossil fuels’ share in energy con- Above 80% 2060
sumption mix

Table 1: Key Chinese Energy Targets.

Source: The State Council (2021)

2.1.3 Emissions Trading System

The first alternative policy implementation will be an emissions trading system. As of April 2020,
there are 23 emissions trading systems implemented around the world. These range from markets
implemented at the city level (Beijing, Chongqing, Saitama, Shanghai, Shenzhen, Tianjin and Tokyo)
to the supranational European Union Emissions Trading System (IEA, 2020).

2.1.3.1 The Chinese Emissions Trading System and its Design

The Chinese ETS is designed to put a price on carbon to reward companies that succeed in cutting their
emissions, while punishing lagging firms and industries. CO2 -emitting firms are allocated tradable
permits that grant them permission to a certain amount of CO2 emitted over a specified period.
Importantly, the market design allows for mutual gain as firms with less emission-efficient processes
can buy permits from more emission-efficient firms. Further, an ETS is theoretically believed to
produce similar outcomes to carbon taxes, however, with potentially less associated political backslash
(Hanley, Shogren, & White, 2019). The theoretical idea is attractive, and as a result, these markets
are becoming increasingly common for regulating carbon emissions and helping markets facilitate a
shift towards alternative energy sources through technological innovation (IEA, 2020).
The permits are currently allocated for free based on the CO2 emission of each site. However, while
the Chinese ETS is in many regards similar to the European cap-and-trade system it differs in one
important regard. Namely, while the EU ETS operates with a declining annual cap on CO2 emissions
(European Commission, 2021), the current Chinese iteration does not have a cap on the total permits
allowed in the economy. China has chosen to rely on intensity-based climate goals instead of absolute
value caps of CO2 emissions.
The flexible amount of permits in the Chinese ETS poses two issues. Firstly, it is currently unclear
whether the Chinese ETS will lead to a reduction in total CO2 emissions. In a report published by the
IEA they conclude that the ETS could help China decrease CO2 emissions and achieve its ambitious
climate targets given that the permits allocated are based on emission intensities decreasing over time
(IEA, 2021). However, this has not been confirmed to be implemented in the Chinese ETS.
Furthermore, a report released by the think-tank TransitionZero argues that the current version
of the Chinese ETS will not have any significant impact on emissions. They argue that the lack of a
declining absolute cap and reliance on intensity-based permits are expected to create an oversupply of
permits which leads to inefficiently low prices (Gray, 2021). Too low prices remove the incentive for
firms to reduce their CO2 emissions.
The second issue with the flexible amount of permits is instability in the carbon permit price. As
supply constantly varies under the current iteration of the Chinese ETS, it is more likely that the
price will fluctuate more heavily compared to an absolute cap that ensures a synthetic scarcity of the
permits and thus a more stable price. In the case where permits are too volatile, firms will greatly
suffer as future investment decisions will become drastically more difficult to make. The theoretical
benefit of such a system over carbon taxes is uncertain (Hanley et al., 2019).

5
2.1.3.2 Has ETS Contributed Towards a Reduction In Emissions?

The practical success of ETS has been widely debated. Specifically, commentators have associated the
effectiveness of an ETS with current market prices as a common belief is that high prices of carbon
permits are crucial for reducing emissions. Therefore, the market price of established ETSs is often
thought to be too low as it differs markedly from most estimations of the Social Cost of Carbon (further
discussed in Section 2.1.5).
Perhaps the most widely analysed ETS is the EU ETS, where notable contributions include Laing,
Sato, Grubb, and Comberti (2013), Bayer and Aklin (2020), and recent reports published by the
European Commission (2021). The former provides a comprehensive framework for evaluating the
success of an ETS based on three aspects:
1. Has the implementation of ETS contributed towards a reduction in emissions?
2. Have the ETS induced incentives for investment in low-carbon technology?
3. What has been the impact of the ETS on profits and product prices?
For the purpose of this paper, the analysis proceeds with the literature review by only investigating
point 1 and refer readers interested in the other two to the original paper (Laing et al., 2013). Key
results are presented below in Table 2,

Study Methodology Key Results

GHG emissions from electricity production
and industrial installations covered by
(European Commission, 2021) Bottom-up the ETS fell by almost 29%, contributing
to an overall decrease of around 43% since
the system was set up in 2005.
EU carbon markets saved cumulative
emissions of about 1.2 billion
(Bayer & Aklin, 2020) Top-down tons CO2 from 2008 to 2016, or roughly
3.8% relative to total emissions
over these years.
China’s regional ETS pilots were effective
in reducing firm emissions in the early
trading phase (2013 to 2015) despite low
carbon prices and allowance liquidity.
(Cui, Wang, Zhang, & Zheng, 2021) Top-down
The magnitude of the effect, a 16.7%
reduction in carbon emissions, is on par
with that of the EU ETS (8 to 12%)
in the second trading phase (2008 to 2012).

Table 2: Summary of Studies.

The studies find strong evidence that an implementation of an ETS can successfully reduce emis-
sions by putting an effective price on carbon emissions and thereby make polluting activities more
costly for regulated firms. This is withstanding the low prices commonly observed in both the EU
and Chinese ETS. Specifically for the EU ETS, Bayer and Aklin (2020) argue that low carbon per-
mit prices are compatible with a reduction in carbon emissions as firms interpret the regulation as
a strong and credible signal for future environmental intervention. The signalling has led regulated
firms to invest in green technologies and reduce their carbon intensity above what would be expected
from the current market prices of carbon. While no specific literature for China currently exists, it is
hypothesised that the reduction in emissions observed by Cui et al. (2021) can be partly driven by the
same signalling mechanism illustrated by Bayer and Aklin (2020). However, while both schemes have
seemingly achieved their primary objective of reducing emissions, it is difficult to directly attribute a
reduction to the ETS due to the many sector-specific environmental policies in place in both China
and the EU (IEA, 2020).
As previously mentioned, both the EU and Chinese ETS have received criticism for their low
prices. In the EU ETS case, the underlying cause of the low prices was a result from one or several

6
of the following reasons: 1) low demand for permits due to the economic environment, 2) competing
environmental policies deflating demand for permits, and 3) oversupply of permits (Bayer & Aklin,
2020). In response to this issue, the architects behind the interventions are continuously implementing
measures to increase the flexibility of the systems, such as the Market Stability Reserve implemented
in 2021 (Phase IV) with the purpose of reducing surplus allowances (European Commission, 2021).
Similar improvements are expected to be implemented in the Chinese ETS in the future, however, it
remains uncertain when and how.
Lastly, while the absolute reduction in carbon emissions for the Chinese and the EU ETS are shown
to be on par by Cui et al. (2021), the response has differed markedly in terms of the specific channels
firms have adopted to reduce emissions. Namely, due to China’s heavy reliance on coal for power
generation, regulated firms can switch to natural gas to reduce carbon emissions while firms covered
by the EU ETS have been shown to reduce emissions by curbing consumption of natural gas and
petroleum products altogether (Petrick & Wagner, 2014). This is achieved by low-carbon innovation
which reduces the cost of firms to comply with environmental policies, and studies have shown that
ETS intervention has led to increased production efficiency in regulated European - as well as Chinese
- firms (Calel & Dechezleprêtre, 2016; Cui et al., 2021).

2.1.4 Carbon Tax

As reported by the World Bank (2021a), of the 64 globally active carbon pricing instruments, 36 were
carbon taxes. Similar to the ETS, a carbon tax is designed to put a price on the amount of CO2
emitted, usually measured in metric tonnes. This is regardless of whether the tax is an upstream or
downstream tax as either way the embedded greenhouse gas emissions in goods are captured. Due to
the fact that carbon taxes reduce negative externalities of climate change caused by greenhouse gases,
they can be considered as a type of Pigovian tax.
In principle, a carbon tax can achieve the same level of reduction in emissions as an ETS with an
absolute cap, but this would require a continuous review and adaptation of the tax to deal with the
consumer’s response. Additionally, a carbon tax does not provide the same level of emission certainty
as an ETS with an absolute cap. With a carbon tax, for example, firms will seek ways to reduce
their emissions, as long as the used solutions are relatively cheaper than paying the carbon tax. This
provides some flexibility as firms can pay the carbon tax if untaxed alternatives become more expensive
due to external factors (development costs, weather, prices), and vice-versa if the other way around.
The required level of emission certainty could be achieved by setting tax rates in accordance to meet
long-term emission targets or by providing price boundaries, which are more commonly found in cap-
and-trade systems such as the ETS. Yet, carbon taxes are in principle quicker to implement and easier
adaptable if the tax rate needs adjustment, therefore one notes the trade-off between certainty and
simplicity when considering carbon taxes against ETS.
Similarly, as with an ETS, a carbon tax can be subject to political pressure, thus again providing a
source of emission uncertainty. The effectiveness of the policy could also be reduced by offering certain
powerful industrial stakeholders exemptions or reductions to the tax. As such policymakers may also
be pressured to set extremely low carbon tax prices. A prominent example of such a situation occurred
recently and will be considered in the next section.
As carbon taxes also generate significant revenue for governments, the carbon tax price is often
directly related to the use of said revenue. Thus the revenue usage is also heavily politicized. Whilst
ideally carbon tax revenue is used for further climate mitigation investments, some governments choose
to employ the revenue for their general budget as shown by Ye (2020) in Table 3.

7
 Country Start Tax Rate Annual Revenue Distribution
Date Revenue
Government budget; accompanied
Finland 1990 US$30 per US$750 mil- by independent cuts in income
metric tonne lion taxes
of CO2
Reductions in other taxes;
Netherlands 1990 approx. US$4.819
Climate mitigation programs
US$20 per billion
metric ton of
CO2 in 1996
Norway 1991 US$15.93 to $900 million Government budget
US$61.76 per
ton CO2
Initially government
budget; Starting in 2000,
Sweden 1991 Standard US$3.665
revenue used to offset
rate: billion
labor taxes
US$104.83
per metric
ton of C02
Industry rate:
US$23.04 per
metric ton of
CO2

Table 3: Select carbon taxes by country and their revenue use.

2.1.5 Setting a Carbon Tax

Because carbon taxes aim at limiting and to some extent remedying the negative effects of greenhouse
gas emissions, carbon taxes and thus the price of carbon are often set equal to the Social Cost of
Carbon (SCC). The SCC can be defined as follows (Ye, 2020):

“The social cost of carbon is the present value of estimated environmental damages over
time caused by an additional ton of CO2 emitted today.”

The SCC is therefore a marginal cost of the damages caused by emissions.

Because Integrated Assessment Models (IAM) allow for the forecasting of climate change effects
and the quantification of the monetary value of environmental damages caused by climate change, they
are used to compute estimates of the SCC. Institutions such as the Environmental Protection Agency
(EPA) in the US have used three models to determine the value of the SCC: DICE (Dynamic Integrated
Model of Climate and the Economy) (Nordhaus & Sztorc, 2013), FUND (Climate Framework for
Uncertainty, Negotiation and Distribution) (Waldhoff, Anthoff, Rose, & Tol, 2014) and PAGE (Policy
Analysis of the Greenhouse Effect) (Hope, 2011). A general dynamic of all three models is that SCC
gradually increases over time. This originates since as time progresses the damages of climate change
accumulate, becoming more numerous and stronger and thus justifying an increase. Generally, the
SCC of countries with high emissions such as the US, China or India is estimated to be higher. Yet,
different IAMs provide different estimates of the SCC, due to their varying underlying assumptions and
model structures. As shown in Nordhaus (2016), even for a single model different SCC estimates are
considered under multiple discount rates for the future damages of climate change. Table 4 illustrates
the different DICE 2013R SCC estimates Nordhaus illustrated in his journal article:
Studies have shown that excessively high discount rates between 3% and 7% are among the most
common reasons for SCC values that are not representative and consistent with global emission reduc-
tion targets such as the Paris Agreement (Stern, Taylor, Karlsson, & Stiglitz, 2022).
As all three models (DICE, FUND, PAGE) yield different estimates and discount rate choice is
influential, the ”Interagency Working Group on the Social Cost of Greenhouse Gases” (IWG) of the

8
 Scenario Assumption 2015 2020 2025 2030 2050
Baseline 31.2 37.3 44.0 51.6 102.5
Base parameters
Optimal controls 30.7 36.7 43.5 51.2 103.6
Maximum 184.4 229.1 284.1 351.0 1006.2
2.5 degree maximum
Max for 100 y 106.7 133.1 165.1 203.7 543.3
The stern review discounting Uncalibrated 197.4 266.5 324.6 376.2 629.2
2.5% 128.5 140.0 152.0 164.5 235.7
3% 79.1 87.3 95.9 104.9 156.6
Alternative discount rates
4% 36.3 40.9 45.8 51.1 81.7
5% 19.7 22.6 25.7 29.1 49.2

Table 4: DICE 2013R estimates (Nordhaus, 2016).

Discount Rate and Statistic

Emissions Year 5% Average 3% Average 2.5% Average 3% 95th Percentile
2020 14 51 76 152
2025 17 56 83 169
2030 19 62 89 187
2035 22 67 96 206
2040 25 73 103 225
2045 28 79 110 242
2050 32 85 116 260

Table 5: SCC estimates from IWG of EPA, measured in $ t/CO2 (Whitehouse, 2021).

EPA has employed a very sensible approach of combining estimates obtained from all three models,
under different discount rates (Whitehouse, 2021). As such they consider the three models under
five different scenarios, yielding 15 frequency distributions of SCC estimates for each discount rate
considered. Further, they combined the distributions under emission scenarios and selected a set of
values for cost-benefit analysis: an average value resulting from the three models under 2.5%, 3% and
5% discount rates as well as the 95th percentile of the 3% discount rate estimate. The latter value
was included to capture more severe than expected economic damages from climate change. Their
resulting values are shown in Figure 5.
As with all results from IAMs, they serve as a guide in policy decisions, but ultimately the choice
of the actual SCC is political. The values represented in Figure 5 were used by the administration of
US President Joe Biden to set the US SCC value. Within his first weeks in office, the administration
set an interim value of US$51 t/CO2 (Cleetus, 2021). The order was accompanied by the promise
of a thorough review of the SCC through an inter-agency working group that would provide updated
SCC estimates. In the previous administration of President Donald Trump, several actions caused
the dismantlement of the working group tasked with calculating the SCC and the cost was reduced to
US$1-7 t/CO2 (Krane & Finley, 2022). This was a fraction of the value set before. Ultimately, this
move was attributed to the aim of the administration to reduce environmental policies in the energy
and automobile sector.
Concluding, there exists a wide variety of SCC estimates based on theoretical aspects such as
different models and discount rates but also due to political influences. Therefore carbon taxes tend
to vary in their aggressiveness.

9
2.2 Review of Current Models
Given its relevance in academia and popular media, various estimates for calculating the electricity con-
sumption and the carbon emissions of the Bitcoin network have been developed by various researchers.
Generally, they can be broken down into either a top-down or bottom-up approach.

2.2.1 Top-down Estimates

Top-down approaches are mathematically simple, and aim to extract trends from various network-
level statistics, such as ‘Total Bitcoin Transactions’ for a fixed period (Mora et al., 2018). This is then
multiplied by a time-dependent coefficient that considers the per-transaction energy requirement and
thus derives the energy transaction for the overall network. Given its simplicity, such models have not
been as popular as they do not consider other factors that might influence how much energy is used.
These factors include (but are not limited to):
• Hashrate
• Number of blocks produced daily
• Difficulty of cryptographic puzzles
• Mining reward
All of these factors will be explained more thoroughly in Section 2.3 where the use of a bottom-up
model in this analysis will be explained. While the main analysis will be conducted with a bottom-
up model, this report will also employ a top-down approach in Section 3.4 for the analysis of PoS
blockchains, given the lack of alternative methods of calculation.

2.2.2 Bottom-up Estimates

As a means of accounting for the influences of key factors that have a direct impact on the amount
of energy the Bitcoin network consumes more precisely, bottom-up models have become the most
popular approach to perform such an estimation. At its core, bottom-up models include factors that
can directly determine the energy consumption on a per-miner level, and work upwards to finally arrive
at an estimate for the electricity consumption or carbon emissions of the Bitcoin network. While most
bottom-up models focus on producing a point-in-time estimate of emissions, there have been two
notable models that can allow live tracking of Bitcoin’s electricity load and consumption on a daily
frequency. These approaches are:
1. ‘Bitcoin Energy Consumption Index’ (BECI) from de Vries.
2. ‘Cambridge Bitcoin Electricity Consumption Index’ (CBECI) from the University of Cambridge
Judge Businesses School’s Centre for Alternative Finance (Cambridge, 2022).
Table 6 reproduces a summary of approaches from previous studies of Bitcoin electricity consump-
tion, found on the CBECI website.

2.2.3 Caveats
It is important to note that most of the studies tend to produce diverging findings and provide a spec-
trum of possible outcomes, reflecting the use of different estimation techniques. Given that there is no
’ground truth’ or record of electricity consumption stored within the Bitcoin blockchain, each approach
will make different assumptions that amount to non-negligible estimation error. It is paramount that
readers are aware of these sources of differences introduced into the estimation, which in most cases
come from:
• Linear interpolation
• Non-linear interpolations
• Simple averaging
• Growth rate assumptions and projections for ‘miner-level’ factors

10
 Authors Date of Title Approach
Publication
Stoll, Lena Klaaßen, and June 2019 The Carbon Footprint of Bitcoin Bottom-up
Gallersdörfer
Zade, Myklebost, March 2019 Is Bitcoin the Only Problem? A Bottom-up
Tzscheutschler, and Scenario Model for the Power
Wagner Demand of Blockchains
Krause and Tolaymat November Quantification of energy and car- Bottom-up
2018 bon costs for mining cryptocur-
rencies
Mora et al. October 2018 Bitcoin emissions alone could Top-down
push global warming above 2°C
McCook August 2018 The cost sustainability of Bitcoin Bottom-up
de Vries May 2018 Bitcoin’s Growing Energy Prob- Top-down
lem
Vranken October 2017 Sustainability of Bitcoin and Bottom-up
blockchains
Bevand February 2018 Electricity consumption of Bit- Bottom-up
coin: a market-based and tech-
nical analysis
Hayes March 2015 A Cost Production Model for Bottom-up
Bitcoin
O’Dwyer and Malone September Bitcoin Mining and its Energy Bottom-up
2014 Footprint

Table 6: Summary of Previous Studies.

These estimations will also be presented in this analysis where they will be stated explicitly at each
step. Regrettably, this will also introduce a non-negligible amount of estimation error.
However, the estimation error does not render such an analysis pointless. While factors influencing
the unobserved true electricity consumption, such as the energy usage and efficiency of cooling solutions
in the mining data centres, which cannot be determined, different estimates are able to capture a similar
change in the magnitude and direction of either electricity or emissions in most circumstances. Readers
should therefore consider these estimates as a rough approximation that captures some of the effects
of electricity consumption and carbon emission of the various cryptocurrency networks. This analysis
will also be closer to a lower-bound estimate on the true emissions from the Bitcoin network given some
of the relatively conservative assumptions, which are explained fully in their respective subsections.
Next, this paper explains the Krause-Tolaymat (2018) model, and extends it to provide us with
forward projections for the network-level energy consumption of Bitcoin, and later Ethereum in Sec-
tion 3.4.

2.3 Multi-period Extension of Krause-Tolaymat Model for Bitcoin

The Krause-Tolaymat (2018) model is a bottom-up approach that has seen a considerable amount
of research attention. It was the first paper to provide a general framework that allowed for the
‘quantification of energy and carbon costs’ for PoW blockchains (Bitcoin, Ethereum, Litecoin and
Monero). Further research has also been built on top of the original analysis done by Krause and
Tolaymat (2018), such as ‘Cryptodamages: Monetary value estimates of the air pollution and human
health impacts of cryptocurrency mining’ from Goodkind, Jones, and Berrens (2020).
This report assumes that this model correctly captures the different dynamics in miners, and
thereby uses it as a base for analysis. For the sake of brevity, this section only provides the mathematical
formulation of the base model, and explains the intuition behind each factor briefly. For a more
comprehensive explanation of the motivation behind the selection of each variable, the original paper
from Krause and Tolaymat (2018) provides a detailed explanation.

11
2.3.1 Network Power Requirement (P)
Firstly, one needs to derive the power requirement of each (PoW) blockchain. This term will be referred
to as P(MW) (measures in Megawatts) and calculated in the following form:
       
hashes Joules −10 MWh seconds
P(MW) = HR × PE × 2.78 × 10 × 3, 600 (1)
second Hash Joule hours

where HR is the hashrate and PE is the power efficiency of mining rigs (the computers used for
writing blocks onto the blockchain). Both factors are determined by the hardware used to mine the
coins. Hashrate has a proportional relationship with time as semiconductors become more effective at
computations, while power efficiency has an inverse relationship with time, given that the advances in
semiconductor technology imply that mining rigs are more power efficient per calculation.

2.3.1.1 Hashrate (HR)

Historical observations for the hashrate are easily obtained online at https://www.blockchain.com/
charts/hash-rate. Given that this analysis focuses on forward forecasts of Bitcoin’s emissions and the
analysis of China’s intervention, the first step is to break down Bitcoin’s historical hashrate into contri-
butions from China and the rest-of-world (RoW). This is done by using estimates for the distribution of
Bitcoin mining hashrate by country from de Best (2022), separating out China individually. The RoW
contribution is attributed to the US and subsequently used as a proxy for subsequent calculations.
With the goal of coming up with forward projections for hash rate, a method inspired by Jiang et
al. (2021) was used and a simple function that reflected how investments in mining equipment (which
affects hashrate) was implemented, which is dependent on miner profits. This profit function is the
same for both Chinese miners and RoW miners, and is of the following form:
 Pt 
HRi={China,RoW },t+1 = HRi,t ×
1 + 2%, if j=t−6 miner profiti,j ≥ 0 (2)
1 − 3%, otherwise

which basically translates into the assumption that says: If a miner type i (either China or Row) has
made a profit over the past six months, they will be willing to upgrade or increase their mining rigs in
the hopes of making more profits further. Conversely, if a miner type is unprofitable over six months,
it is assumed that they will turn off a certain percentage of their equipment (most often the oldest
and least power-efficient) such that they can focus on reducing their excessive expenditures. This
assumption is likely realistic as miners ultimately still have to convert their cryptocurrency holdings
into local currency periodically to pay for utility bills. Miner’s profits will be subsequently explained
in Section 2.3.8.

2.3.1.2 Power Efficiency (PE)

Given that the Bitcoin network has been dominated by ASIC miners, and that estimates for the
upgrade cycle of mining rigs are 1.29 years (de Vries & Stoll, 2021), most researchers use ASIC
miner statistics for the estimation of power efficiency ratings. Historical data can be obtained from
https://www.asicminervalue.com/. In this analysis, the power efficiency of all mining equipment is
aggregated by year of release, and forecasted ahead by extrapolating with a polynomial best-fit line3 .
Since miners might not have the same upgrade cycles as the product update cycle of ASIC miners
released onto the market, these estimates would likely overestimate the power efficiency of mining
equipment, and likely be closer to the lower-bound of network power requirement estimates. These
estimates can be found in Table 7.

3 The polynomial best-fit line has the equation HR= 0.7967 × t−1.439 and R2 = 0.997.

12
 Year Average of Efficiency (j/Gh)
2014 0.7555
2015 0.2730
2016 0.1780
2017 0.1150
2018 0.0993
2019 0.0542
2020 0.0412
2021 0.0554
2022 0.0250
2023 0.0290
2024 0.0253
2025 0.0223
2026 0.0199
2027 0.0179
2028 0.0162
2029 0.0147
2030 0.0135

Table 7: Power Efficiency Estimates of ASIC Mining Equipment. Forward forecast numbers are
coloured in blue.

2.3.2 Network Velocity (v)

Network velocity, as defined by Krause and Tolaymat (2018) is the US$ generated per hour from the
multiplication of the number of blocks generated per hour, the coin reward for block completion and
the conversion from coin to US$. It takes the following form:

         
US$ coin rewarded US$ minutes minutes
v =R × EX ÷ tB × 60 (3)
h block coin block created hours

2.3.2.1 Reward Per Block (R)

The reward per block is clearly defined in the Bitcoin protocol, where coins are routinely issued
at a decreasing rate (on a per block basis) such that there can only ever be 21 million Bitcoins in
circulation. On average this happens every four years. Starting from 25 Bitcoins per block initially,
the first ‘halving’ happened in November 2012, the second happened in July 2016, and the most recent
in May 2020. The next two halving events are estimated to be in June 2024 and June 2028, where the
reward per block to decrease to 3.125 and 1.5625 respectively.
Later, it is shown that such halving events will lead to an adjustment period which will bring down
miner profitability, hashrate and electricity consumption in Section 3.

2.3.2.2 BTC/USD Exchange Rate (EX)

The BTC/USD spot exchange rate can be observed historically and will be used to inform miner profits
subsequently. It is impossible to come up with forward estimates of BTC/USD and this report does
not attempt to try and be a magical Bitcoin oracle. However, it has been noted in some papers that
the ‘fundamental price’ of Bitcoin (Jiang et al., 2021) tends to have an inverse relationship with the
halving events. That is, price of Bitcoin tends to double at every halving event (which occurs every
four years). This is just made as a working assumption for us to understand the dynamics of this
model, and very conservatively assume that Bitcoin price increases by 10% annually.

13
2.3.2.3 Block Time (tB )

Block time is the average amount of time (in minutes) it takes for a new block to be written on the
Bitcoin blockchain. This has historically hovered around the 10 minute mark and use this as a forward
estimate. Given that this is roughly the limit of the throughput for the Bitcoin blockchain, this is
likely an accurate estimate.

2.3.3 Energy Required Per Coin (NC )

As a precursory step towards determining the amount of carbon dioxide emitted by each country’s
miners, NC is the energy required for a coin to be generated on a given month.
     
kWh minutes kWh
NC = P × tB ÷ 60 × 1, 000 ÷R (4)
coin mined hours MWh

2.3.4 CO2 Emitted (kg) Per Coin (TC )

In this step, the effects of carbon emission are attributed to the countries in which the miners reside.
In this model, only China and RoW (with US as a proxy) are considered. The sum of both countries
will provide us with the network level carbon emission per coin.
X
TC = Ti,C (5)
i={China,RoW }

 
CO2 emitted (kg)
Ti,C = Yi × NC × % hashrate contributioni
kW h
where Yi reflects the CO2 emission from the country reflecting the usage of renewable and non-
renewable sources in each country and each miner type’s being considered through the % hashrate
contribution.

2.3.4.1 YChina and YRoW

Historical estimates for the carbon emitted per unit of energy were obtained from Our World in Data4 .
Forward projections were calculated very simply for china and the US through applying the long-run
linear time trend (going back to 1965) onto the last observation (2019). The long-run time trend
for the China is −0.000975 per year and RoW is −0.000418 per year. One has to note that with
China’s nuclear energy ambitions, the estimates for China could be much lower towards the end of the
estimation period.
The estimates can be found in Table 8.

4 The estimates can be found at https://ourworldindata.org/CO$ 2$/country/china#carbon-intensity-how-much

-carbon-does-it-emit-per-unit-of-energy

14
 Year China USA
1965 0.3265 0.2334
··· ··· ···
2019 0.2665 0.1999
2020 0.265525 0.199482
2021 0.26455 0.199064
2022 0.263575 0.198646
2023 0.2626 0.198228
2024 0.261625 0.19781
2025 0.26065 0.197392
2026 0.259675 0.196974
2027 0.2587 0.196556
2028 0.257725 0.196138
2029 0.25675 0.19572
2030 0.255775 0.195302

Table 8: Estimates for Yi . Forward forecast numbers are coloured in blue.

2.3.5 Monthly Rate of Coin Generated (rC )

The monthly rate of coins generated is a simple calculation that takes into account block time and
reward per block.
       
coins days hours minutes
rC = 30 × 24 × 60 ÷ tB × R (6)
day month day hours

2.3.6 Monthly Electricity Consumption of Bitcoin Network

The monthly electric consumption of the Bitcoin network is simply the sum of the parts for each miner
type. X
EC = Ei,C (7)
i={China,RoW }

Ei,C = NC × % hashrate contributioni × rC

2.3.7 Monthly CO2 Emission of Bitcoin Network (CC )

The CO2 Emission of the Bitcoin network is simply the sum of the parts for each miner type.
X
BT C BT C
CC = Ci,C (8)
i={China,RoW }

BT C
Ci,C = Ti,C × rC

2.3.8 Miner’s Monthly Profit

The miner’s profit contributes to the hashrate forward forecast. This is defined as the miner’s monthly
profit contribution as the difference between the number of coins attributed to them and the price cost
of electricity required.
   
US$ US$
miner profiti,t = EX × rC × % hashrate contributioni − pi,t × Ei,C (9)
coin kWh

2.3.8.1 Price of Electricity in Miner’s Country (pi,t )

The price of electricity in the RoW (proxied by the US) was found historically from the US Bureau of
Labour Statistics (U.S. Bureau of Labor Statistics, N.A.). For the forward forecasts (in blue), simple
averaging of electricity prices from 2012 to 2021 was performed, as the prices of electricity in the US
has been fairly consistent and the energy mix is unlikely to change.

15
 When it comes to Chinese electricity prices, some discretion was applied in finding a representative
price. Given that miners tend to migrate to where electricity is cheaper, one finds that Guangzhou
tends to have cheaper prices for industrial usage and thereby applied this number as an estimate.
When it comes to the forward projections, a linear interpolation was applied between the last observed
price and 0.075 US$/kWh, as this price reflects the long-term benefits China would benefit from when
it fully realises its nuclear energy goals. This is a discretionary choice given comparisons to other
countries which have a higher nuclear mix, and does not diverge too far from China’s long-term price
trend.

Year (CN) Price Exchange (CN) Price US$/kWh (US) Price US$/kWh
RMB/kWh Rate
2014 0.577 6.160 0.094 0.137
2015 0.870 6.280 0.139 0.138
2016 0.647 6.650 0.097 0.135
2017 0.640 6.760 0.095 0.138
2018 0.800 6.630 0.121 0.136
2019 0.588 6.910 0.085 0.136
2020 0.620 6.900 0.090 0.135
2021 0.620 6.450 0.096 0.141
2022 0.620 6.350 0.098 0.147
2023 0.095 0.137
2024 0.092 0.137
2025 0.089 0.137
2026 0.086 0.137
2027 0.083 0.137
2028 0.081 0.137
2029 0.078 0.137
2030 0.075 0.137

Table 9: China and US Electricity Prices.

3 Models
Through the equations seen earlier, one can model out various dynamics between the change in elec-
tricity consumption and carbon emission of the Bitcoin network. The next part looks at various cases
of policies and interventions to quantitatively evaluate the impacts of each.

3.1 What The Model Implies About Bitcoin Emissions

Before proceeding to look at cases for intervention, two baseline scenarios are shown, where one can
understand what the electricity consumption and carbon emission of the Bitcoin network would be
under:
1. The current case where China has banned cryptocurrency mining.
2. A hypothetical case where China allowed cryptocurrency to continue being mined.
Subsequently, every intervention policy should fall somewhere between these two cases. This is because
the intervention set will only be considered through the lens of Chinese regulators, so the banned case
being the least polluting, as it removes all Chinese miners who are more polluting than RoW miners.
By a similar logic, the allowance of Chinese miners will mean that a more polluting mix of energy is
used by Chinese miners, who dominate the market given their lower price of electricity. As such, the
no ban scenario will lead to the highest emissions. The cases are elaborated below.

3.1.1 Ban Case - China Bans Cryptocurrency Mining

As mentioned in Section 1.2, the current scenario sees crytocurrency mining being made illegal in
China as of June 2021 (Shen & Galbraith, 2021a). Given that around 70% of mining activity was

16
previously attributed to China (de Best, 2022), this means that a very immediate drop in both the
hashrate and electricity usage was observed in the later half of 2021. This also leads to the case with
the lowest projected annual and total emission from the Bitcoin network, as polluting Chinese miners
that have access to cheap electricity are removed from the network.
This leads to the following dynamic seen in Figure 2. In this case, one sees that the first peak
happens in 2022-2023. This occurs largely due to the slack provided by the departure of Chinese
miners, as RoW miners pick up mining and experience a period of high profitability. However, there
is also a sharp drop off in 2024, given the halving event in that year (explained in Section 2.3.2.1).
The model accounts for this in a very naive way, with actual transition likely being much smoother
than indicated by the model, given that such events are known. A similar dynamic is observed in 2027
and 2028, where electricity consumption and CO2 usage peak again after the drop off in hashrate from
the 2024 halving.

Figure 2: Base Case - China Bans Cryptocurrency Mining.

The model also suggests that post 2028, both electricity usage and mining should begin falling off, in
no small part due to the decline in reward of the Bitcoin network and the general increase in hashrate.
Given that the network will decreasingly contribute to global electricity consumption and emissions,
this is likely a representative base case to compare against. The total electricity consumption and the
annual CO2 emissions for the period 2014-2030 can be found in Table 10, and the estimate is 540 TWh
and 108 Mt respectively. The model estimates tend to be smaller than estimates from other scholars,
in no small part due to the conservative PE assumptions (see Section 2.3.1.2).

17
 Year Global An- Global Annual Chinese Miner RoW Miner
nual Electric- CO2 Emission Profit (US$) Profit (US$)
ity Consump- (Mt)
tion (TWh)
2014 0.91 0.22 141841265 404919748
2015 0.96 0.24 130407693 101268722
2016 2.35 0.60 246052181 76491375
2017 6.23 1.58 1509792032 492267467
2018 31.42 7.99 805449265 172312191
2019 31.25 7.80 1631014292 134363295
2020 43.01 10.42 816701017 -169539279
2021 70.32 15.12 1619622014 4458026292
2022 47.57 9.45 0 3078807966
2023 70.23 13.92 0 1227907192
2024 71.05 14.05 0 -1774515150
2025 48.33 9.54 0 -56533711
2026 48.60 9.57 0 563396050
2027 55.40 10.89 0 353986311
2028 56.74 11.13 0 -1581833907
2029 39.35 7.70 0 -584700869
2030 32.89 6.42 0 780354757
Global 540.48 107.80 1619622014 6464894931
Total
(2021-
2030)

Table 10: Ban Case - China Bans Cryptocurrency Mining. Global annual estimates are for the whole
Bitcoin network.

3.1.2 No Ban Case - If China Allowed Cryptocurrency Mining

The case where China does not ban cryptocurrency is perhaps the worst case for the environment
currently, given its more polluting energy mix (see Figure 1, where China’s coal contribution to energy
was around 56% in 2020) and lower energy prices (see Table 9). As seen in Figure 3, the amount of
electricity consumed and CO2 emission is much above the ban case, as seen in the gray lines. While
there is some convergence after the 2024 halving, the tail end of the forecasts still show both variables
rising, unlike the ban case. This is an interesting dynamic that can be explained by the relative
cheapness of Chinese electricity in the model, which means that Chinese miners can still profitably
mine even as rewards are halved, driving out RoW Miners. This can also be observed in Table 11,
where the Chinese miners are profitable while RoW miners are unprofitable from 2028 onwards. Given
the above dynamics, the forecasted the electricity consumption of the Bitcoin network (2021-2030)
is 638 TWh. When it comes to emissions, the relatively more polluting mix of the Chinese miners
are likely to lead to the worst environmental outcome. This leads to the CO2 emission of the global
Bitcoin network to be 148 Mt.
An important thing to note is that going forward, one would expect that even with the Chinese ban,
and with the RoW miners picking up the slack from Bitcoin to the same hash rate, the environmental
outcome will likely be more desirable than weaker Chinese intervention cases. This again is a result of
the modelling assumptions made, where the RoW miners will always have a less polluting energy mix,
but face higher energy prices. This is a result that will be shown next.

18
 Figure 3: No Base Case - China Does Not Ban Cryptocurrency Mining.

Year Global An- Global Annual Chinese Miner RoW Miner

nual Electric- CO2 Emission Profit (US$) Profit (US$)
ity Consump- (Mt)
tion (TWh)
2014 0.91 0.22 141841265 404919748
2015 0.96 0.24 130407693 101268722
2016 2.35 0.60 246052181 76491375
2017 6.23 1.58 1509792032 492267467
2018 31.42 7.99 805449265 172312191
2019 31.25 7.80 1631014292 134363295
2020 43.01 10.42 816701017 -169539279
2021 90.43 20.44 2782632660 1361529874
2022 66.79 14.60 1087986235 175654256
2023 79.41 17.68 1252365884 -14171807
2024 73.29 16.57 504914733 -1130624842
2025 55.83 12.86 828209687 -523615952
2026 51.80 12.26 1730105219 48881178
2027 55.72 13.27 2198397761 102913854
2028 60.59 14.51 885095910 -596650249
2029 51.84 12.55 582613838 -547286961
2030 52.65 12.99 1128245515 -282972443
Global 638.35 147.73 12980567442 -1406343092
Total
(2021-
2030)

Table 11: No Ban Case - China Reverses Ban in June 2021. Global annual estimates are for the
whole Bitcoin network.

19
3.2 Implementation of Emissions Trading System
China formally introduced an emissions trading system (ETS) on the 16th of July 2021, which instantly
became the world’s biggest of its kind by effectively covering 12% of global CO2 emissions. While
currently covering 40% of China’s total CO2 emissions, President Xi has ambitiously vowed to rapidly
expand the market to 80% by 2025 in order to curtail polluting industries (Refinitiv, 2021).
The introduction of the ETS adds an additional tool in the wide array of policies that are being
implemented in China in order to achieve its dual carbon climate goal. As such, this section investigates
its potential in mitigating the environmental consequences caused by the Bitcoin mining industry.

3.2.1 Mathematical Formulation

To investigate the effect of a carbon trading system, a simplified agent-based model is implemented.
Three main agents are involved: the government, the Bitcoin industry, and the rest of economy (RoE).
The model can be depicted by the diagram in Figure 4. Note that given the discussion in Section 2.1.3.1,
this report will proceed by implementing a cap-and-trade ETS with a declining absolute cap instead
of an intensity-based specification similar to the one currently in place in China.

Allocation Rule Government

Rest
Bitcoin
of Permit Market
Industry
Economy

Figure 4: Schematic diagram of the simplified ETS model.

3.2.2 Agents Description

In the model there are three agents,
1. The government’s responsibility is to (1) determine the extent of the ETS as a total of Chinese
CO2 emissions, (2) calculate its desired absolute cap of CO2 emissions at the start of every
year, (3) allocate free tradeable carbon permits to the Bitcoin industry and RoE according to its
allocation rule.
2. The Bitcoin industry will interact with the model by purchasing or selling permits from RoE
needed for them to conduct their mining operations.

3. RoE is assumed to be in constant equilibrium and can conduct its operations at the current
market price of carbon permits. The RoE will interact with the Bitcoin industry by supplying
and buying carbon permits for any excess demand or supply at the current market price.

20
3.2.2.1 Government Agent
The government’s objective is to curtail the CO2 emissions of the economy by decreasing the total
allowable permits. The dynamic modelling assumes the permit market is effectively covering 40% of
the total CO2 emissions in China, where the Bitcoin industry is fully covered by the scheme. However,
as time passes it is assumed that the ETS is linearly expanding to ultimately cover 80% of total Chinese
CO2 emissions by 2025.
Furthermore, a grandfather rule is applied for the allocation of permits based on historical carbon
emissions of the Bitcoin industry (Knight, 2013). For the purpose of this model, it is assumed that the
government has perfect insight into the exact CO2 emissions attributed to the Bitcoin industry. The
initial amount of permits allocated will be:
t−1
1 X BT C
CP ermitsBT C,t=0 = C (10)
6 i=t−6 China,i

After the initial allocation, the government decides on an emissions reduction factor σT 5 which
corresponds to the decrease in permits allocated to the Bitcoin industry at the start of every year. As
such, at month t the Bitcoin industry is endowed with the following amount of permits:

(1 − σT )CP ermitsBT C,T −1

CP ermitsBT C,t = (11)
12

3.2.2.2 Bitcoin Industry

The general agent will behave similarly to the benchmark case. However, at the end of the period
the Bitcoin industry has to submit an equal amount of permits as the number of tonnes of CO2 it has
emitted. This means that any CO2 emission exceeding the CP ermitst needs to be purchased from
the market. In such a case, an additional emissions cost will be incurred:
BT C
ECosti,t = µt × (CChina,t − CP ermitsBT C,t ) (12)
where µt is the current market price of one unit CP ermit. The miner’s new profit function becomes:
   
US$ US$
miner profiti,t = EX × rC × % hashrate contributioni − pi,t × Ei,C − ECosti,t (13)
coin kWh

Furthermore, in the case of over-allocation or effective reduction in emissions, a situation could

arise where the Bitcoin industry will be left with more permits than required to cover its emissions.
In such cases the Bitcoin miners will choose to sell any excess permits in the market. ECosti,t will
then act as a windfall to the Bitcoin industry.
Note that this model does not allow for second-level considerations such as switching to low-carbon
fuels or investment in emission abating technologies.6

3.2.2.3 Rest of Economy

The Rest of Economy is assumed to remain in equilibrium at any prevailing market price. Put differ-
ently, the industries covered by the ETS are able to effectively conduct their operations and optimise
their profits given the current market price, leaving them unwilling to trade any permits. This will be
implemented in the model by setting:
RoE
CP ermitsRoE,t = CChina,t (14)
As mentioned, President Xi vowed to expand the ETS market to 80% of total CO2 emissions by
2025. This is incorporated into the model by incrementally increasing the relative size of the RoE to
the Bitcoin industry and, as a result, any supply/demand imbalance arising from the Bitcoin industry
will result in smaller price movements over time. Specifically, the Bitcoin industry’s market share of
the total permit market will decrease to 0.05% by the end of the year 2030 from 0.20% in 2021. This
5T refers to the year corresponding to month t.
6 Refer to min Yu, Fan, Zhu, and Eichhammer (2020) for an Agent Based Model incorporating similar considerations.

21
is a result of the continued expansion of the ETS market, as well as the continued growth of China’s
total CO2 emissions.7

3.2.3 Market Dynamics

The model includes a simple trading mechanism which elicits a price increase to any excess demand
and a price decrease to excess supply (Lebaron, 2001):
RoE BT C
µt+1 − µt = α(CChina,t + CChina,t − CP ermitsRoE,t − CP ermitsBT C,t ) (15)
which given equation (14) becomes:
BT C
µt+1 − µt = α(CChina,t − CP ermitsBT C,t ) (16)
where α is a constant reflecting the sensitivity in price to supply and demand imbalances.
Notably, the model is highly sensitive to the α parameter. A too large parameter leads to excessive
and unrealistic price movements, while a too small parameter would lead to a sluggish market unable
to react to excess demand. To mitigate this issue, the permits demanded from the Bitcoin industry
relative to the total market are considered, and the parameter is tuned to result in price movements
associated with such an agent disturbing the market equilibrium. For example, assuming a 2% yearly
additional increase of CO2 emissions from RoE leads to a quadrupling in carbon permit price from
2021 to 2035. Existing ETS implementations have seen comparable price movements in even shorter
time frames (Trading Economics, 2022). Lastly, the model is sensitive to the initial specified carbon
permit price, µ0 , which here is assumed to be US$50. The reason for the deviation from China’s
launching price of 48.01 CNY (US$7.53) (Refinitiv, 2021) is due to this paper’s ETS using an absolute
cap on carbon emissions similar to the EU ETS. It is therefore expected, given similar abatement
policies, that the launching ETS price would be in the range of the EU ETS trading range of €0-€90
(US$0-US$98.15).8 Furthermore, in its Sustainable Development Scenario9 the International Energy
Association assumes a carbon price (SCC) of US$43 per tonne of CO2 emitted in 2025 for developing
countries and US$63 per tonne for advanced countries. This paper will proceed by assuming µ0 = 50
by weighting the above considerations.

3.2.4 Scenarios of ETS Intervention

Three scenarios are introduced, with a corresponding linear yearly reduction in permits allocated to
the Bitcoin industry as shown in Table 12. The intervention is implemented in June 2021 with an
initial endowment of permits decided according to Equation 10, and by the beginning of each year the
number of permits will be reduced by σT . The total reduction in permits is then calculated by the
endowed amount of permits in year 2030 relative to the initial amount endowed in 2021.

Case Reduction Factor (σT ) Total Reduction in Permits (2021-2030)

A 5% per year 34%
B 15% per year 73%
C 30% per year 94%

Table 12: The three ETS cases and the associated yearly reduction in allocated carbon permits.

7 The forward projection for China’s total CO emissions is calculated simply by a linear regression on historical
2
log-values.
8 See Trading Economics for EU ETS Carbon trading price.
9 An integrated scenario specifying a pathway aiming at: ensuring universal access to affordable, reliable, sustainable

and modern energy services by 2030; substantially reducing air pollution; and taking effective action to combat climate
change.

22
3.2.5 Results of ETS
The result of the implementation of the ETS and subsequent effect on the global Bitcoin network’s elec-
tricity consumption and CO2 emissions can be shown for the three cases A-C in Figure 5. Furthermore,
the carbon permit price development is shown in Figure 6.

Figure 5: Electricity consumption and CO2 emissions forecasts for the global Bitcoin network for
cases A-C.

Figure 6: Permit price forecast (µt ) for cases A-C.

3.2.6 Analysis
Figures 5 and 6 illustrate the result of the implementation of the ETS on the global Bitcoin network’s
electricity consumption, annual CO2 emission, and market price for carbon permits. It is observed that
given a 5% yearly reduction (Case A) in endowed carbon permits to the Bitcoin industry, the global
Bitcoin CO2 emission is expected to decrease relative to the China No Ban Case with a forecasted
CO2 emission of 175 Mt between 2014-2030 vs. 177 Mt. This, rather small effect, can be ascribed to
mainly two calibrations of the model, namely 1) the initial starting price of a carbon permit being set
at US$50 and 2) the allocation rule’s initial allocation and decrease in permits year over year. While
an initial carbon permit price of US$50 is well within the observed range of actual trading schemes
it markedly differs from a firm’s average cost of carbon as the permits are allocated for free. This,

23
together with the allocation rule, will therefore ultimately determine the average cost of carbon for
the Bitcoin industry and any subsequent investment decisions.
Thus, it is not surprising to observe from Figure 5 that as the government decreases the yearly
allocated permits, so do the total CO2 emissions. However, a maximum reduction of 93% carbon
permits endowed to the Bitcoin industry (Case C) only results in a 2.89% reduction in global CO2
emissions from the Bitcoin Network compared to the No Ban Case between mid-2021 to December 2030.
This result is mainly due to the ability of Chinese Bitcoin miners to extract significant and sustainable
economic rents in current market conditions. The miners can therefore continue their business while
purchasing the required amount of carbon permits without suffering from any significant losses that
would impact their future investment decisions. Secondly, the increased demand arising from the
Bitcoin industry is relatively small compared to the entire ETS market, and as such, an increase in
demand from Bitcoin miners will only affect the market price in relation to its size. This leads to a
sluggish price that is not as responsive to changes in demand from the Bitcoin industry as perhaps
would be socially optimal, and the dynamic will aid the industry to maintain its operations even
while incurring additional costs and the endowed permits are significantly reduced year over year. For
example, a 30% yearly reduction (Case C) in allocated carbon permits results in a modest increase
in carbon permit price, reaching US$58.73 (+17.46%) in 2030. Meanwhile, the amount of permits
purchased per month by the Bitcoin industry rise from 279,255 in 2023 to 749,230 (+168.30%) in 2030
in Case C.10 .
Further, Case A results in a prolonged period where the carbon permit price µt decreases below
the initial launching price of US$50 as seen from Figure 6. This can be explained by the specified
allocation rule endowing the Bitcoin industry with an excess amount of permits which can later be
sold for additional profit to the RoE. The additional number of carbon permits being supplied to
the market decreases the market price as determined by equation (16), aptly illustrating the issue of
over-allocation mentioned in 2.1.3.2 and confirming the importance of efficient allocation and artificial
scarcity of the number of allocated permits, as highlighted by Gray (2021). Lastly, in all three scenarios
the intervention result in noticeable effects on electricity consumption and emissions starting in 2024.
This result can be ascribed to the Halving Mechanism, which will be further discussed in Section 4.1.

3.3 Implementation of Carbon Tax

Before implementing the emission trading system, in 2010 it was reported that the Chinese Ministry of
Finance had proposed a carbon tax which would come into effect in 2012 (Zhang, 2010). The proposal
suggested introducing a carbon tax of 10 CNY ($1.44) t/CO2 which would increase to 50 CNY ($ 7.25)
t/CO2 by 2020 (Ros, 2013). The tax proposal never passed however.
Given the differences in SCC prices discussed in 2.1.5, in the subsequent sections the paper inves-
tigates how different levels of Chinese carbon taxes would mitigate the environmental consequences
caused by the global Bitcoin mining network, if they had been passed instead of the ETS.

3.3.1 Mathematical Formulation

To investigate the effect of a potential carbon tax, the Chinese miner’s profit function is adjusted by
including an additional cost element originating from the carbon tax.
The cost of the tax τ , in month t, can be expressed by multiplying the monthly CO2 emission of
the Chinese Bitcoin network CChina by the price per metric tonne of CO2 at time t, denoted Pt , as
follows in equation (17).
τChina,t = CChina Pt (17)

This additional cost is then included in the miner’s profit function as follows:
   
US$ US$
miner profitChina,t = EX × rC × % hashrate contributioni − pChina,t × Ei,C − τChina,t
coin kWh
(18)
10 The time period is adjusted in order to avoid including 2021-2022 which includes significant selling of carbon

permits by the Bitcoin industry. The comparison is therefore not directly applicable but should provide the reader with
an indication of the discrepancy between the increase in demand and increase in price.

24
 Given this adjusted miner profit equation, one can analyze the effect of carbon taxes as an emission
reduction policy intervention. The differences in profit of Chinese Miners will have an effect on the
electricity consumption and emissions of the global Bitcoin network.

3.3.2 Carbon Tax Scenarios for Analysis

For the carbon tax analysis, three price scenarios were chosen. In each scenario, the prices increase
by varying degrees between June 2021 to December 2030 are shown in Figure 7. Linear interpolation
between the initial and terminal prices was applied to obtain the monthly prices required. Thus the
three chosen scenarios are as follows:

• Scenario 1 (50%): The first scenario will be referred to as the 50% case. In this case, China will
set it’s carbon prices at precisely half the price of the upper price targets required to achieve
the temperature pathway set by the Paris Agreement, thus lying at the lower bound. As one
will see in scenario 2, this price also corresponds to half the price set by the US. Given the low
price setting tendency of the 2010 carbon tax proposal described in Section 3.3, this tendency
is extended to the 50% scenario. Additionally, China will aim to set as low as possible carbon
tax prices in order not to set a precedent of high carbon taxes that could further be applicable
to other industries of higher importance to their domestic economy. Thus in this scenario, in
June 2021 the initial price will be US$40, increasing to US$50 by 2030.
• Scenario 2 (Paris Agreement): The second scenario will be referred to as the ”Paris Agreement”
case. The report of the World’s Bank High Level Commission in 2017 on carbon prices establishes
that in order to achieve the 2 degrees Celsius temperature goal of the Paris Agreement, global
carbon prices should reach between US$40-80 t/CO2 by 2020, and then increase to between
US$50-100 t/CO2 by 2030 (World Bank, 2021b). This price evolution is similar to the prices set
in the US by the Biden administration. Recall from Section 2.1.5, that studies considered a 3%
discount rate to high for the SCC. Looking at the US technical guidance document (Whitehouse,
2021) for the calculation of the SCC, at a 2.5% average discount rate, the price of the SCC
increases from US$76 t/CO2 in 2020 to US$89 t/CO2 in 2030. Thus the SCC estimated by the
Biden administration seems in line with the Paris Agreement. Hence, this scenario considers the
hypothetical Chinese implementation of a carbon tax at the upper price recommendations of the
World Bank and thus in line with the Paris Agreement and the US. In June 2021, the initial
price is set at US$80, increasing to US$100 by 2030.
• Scenario 3 (High): The last scenario will be referred to as the ”High” case. Multiple studies
suggest that the carbon prices should be set at well above US$100 t/CO2 (Carleton & Greenstone,
2021) as soon as possible, in order to truly achieve the emission reduction and temperature targets
set by the Paris Agreement. The World Bank itself mentions this fact in their 2021 report:
”State and trends in carbon pricing”, by referencing a study by Woodmac, an energy research
and consultancy founded in the UK. This study suggests that global carbon prices should reach
US$160 t/CO2 to successfully achieve the Paris Agreement pathways (Mackenzie, 2021). Since
China is one of the largest emitters of CO2 , who’s SCC should naturally be higher, this scenario
considers a price increase all the way up to US$160 t/CO2 . Thus, the initial June 2021 price is
set at US$100, increasing to US$160 by 2030.

25
 Figure 7: 3 scenario carbon price evolution.

3.3.3 Analysis of Carbon Tax Intervention

Figure 8: Carbon Tax 50 %.

Figure 9: Carbon Tax Paris.

26
 Figure 10: Carbon Tax High.

Figures 8 to 10 illustrate the development of annual electricity consumption and CO2 emissions of
the global Bitcoin mining network under each carbon tax scenario. The figures contrast the results
of the different scenarios to the ”No Ban” and ”Ban” cases in China previously discussed. As the
carbon prices increase towards US$160 t/CO2 , the line’s colours become progressively more red, given
starting points of US$0 t/CO2 coloured in green.
As indicated by the figures, the higher the carbon tax, the lower the CO2 emissions become. As such,
the “High” scenario carbon tax provides the greatest reduction in emissions followed by the “Paris” and
“50 %” case. As the price evolution is also greatest in the “High” scenario, more significant emission
reductions become apparent at 2025 and 2026 compared to the other two scenarios. The different
price evolutions also become apparent in the tail end of the prediction interval. In the years 2029 and
2030, for the “50%” case the carbon emission line flattens out, whilst in the other two cases the line is
downward sloping. This is because in the “50%” case, over the prediction interval price increases by
US$10 t/CO2 , compared to the US$20 t/CO2 and US$60 t/CO2 increases in the other two scenarios.
Thus the step-wise monthly increases become greater in the higher-priced carbon tax scenarios, which
leads to the aforementioned higher emission reductions at the tail end of the prediction interval.
A further notable finding presents itself in the annual electricity consumption in Figure 10. In 2026,
under the ”High” scenario, electricity momentarily falls beneath the ”China Ban” case, indicating high
effectiveness of aggressive carbon taxes in reducing electricity consumption. Also, under the other
scenarios electricity consumption is effectively reduced, especially in the described year range, further
highlighting the efficacy of carbon taxes in electricity consumption reduction.
Nevertheless, in all three scenarios, the carbon taxes only become truly effective after 2024. Up
to that year, the carbon emissions produced by Chinese miners are little to no difference between the
scenarios and the “No ban” case in which China takes no action against Bitcoin mining. After 2024,
the taxes become effective and reduces annual CO2 emissions. The reason for the occurrence of this
phenomenon is the planned Bitcoin “halving” in March in 2024, where the reward is reduced to 3.125
of new Bitcoins per block. Halvings are significant, because they decrease the supply of Bitcoins. For
miners this means that less reward is available for the same amount of costs. Therefore, the carbon
tax by further increasing costs, decreases the miner’s profitability even more which in turn leads to
less mining and therefore less CO2 emissions.

3.3.4 Carbon Tax Revenue

Another interesting consideration, is the revenue carbon taxes generate. Often carbon tax revenue
comes under political pressure. Depending on the price, carbon taxes can generate a significant amount
of revenue which can be invested in climate change mitigation measures and transitional support for
industries and communities dependent on high carbon energy sources.
Each analysis illustrates how much revenue was generated under each analyzed scenario. As such,
Table 13 illustrates the annualized revenue generated by each carbon tax scenario whilst also putting
the amount of emitted CO2 in context:

27
 Scenario Chinese Annualized Annualized Annualized Annualized Annualized
Carbon Global Global Chinese Tax Rev- Tax Rev-
Price Electrcity Carbon Tax Rev- enue/Total enue/Total
Evolu- (TWh) Emission enue (M Chinese Chinese
tion (Mt/CO2 ) US$) Miner Miner
Revenue Earnings
before
Tax
50 Per- 40 to 50 61.52 14.12 371.17 9.4% 30.1%
cent
Paris 80 to 100 60.38 13.74 678.64 18.2% 55.6%
High 100 to 59.22 13.35 875.21 25.1% 74.1%
160

Table 13: Revenue Table by Carbon Tax Scenarios Period 2021.06.01-2030.12.01

As apparent from the table, the larger the carbon tax, the greater the revenue generated by the
tax. As such, the revenue generated in by the “High” carbon tax is $200 trillion more than the revenue
generated by the “Paris” case for China.
Another interesting point comes when looking at the percentage that the carbon tax revenue makes
up of the Chinese miner profit. By considering the earnings before tax of the Chinese mining network
compared to the carbon tax revenue generated, it can be illustrated how penalizing different carbon
taxes are. The increase in percentages from 26.9% in the 50 % case all the way up to 65.1% nicely
illustrates how carbon taxes also manage to reduce the Chinese miner’s overall revenue. As such in
the “High” case, the high carbon price and the reduction of overall profit for miners work in tandem.
Therefore, this analysis concludes that in all three cases the government manages to recoup a significant
amount of Chinese miner’s profits.
Nevertheless, under the assumptions made and even with aggressive carbon pricing as shown by the
analysis and the revenue consideration, Chinese miners are still likely to dominate Bitcoin mining. In
all three cases, significant profits will still be available by 2030. With further price increases specifically
for coal-based energy, which is very popular in China as well as stricter international policies, in the long
run the pollution of the Chinese Bitcoin mining network could further be reduced. Simultaneously, it
can be expected the Chinese government will use the revenue generated to transition industries reliant
on high carbon fuels such as coal and oil, to cleaner energy sources such as liquefied natural gas first
and then eventually to renewable energy. In the future, it can also be foreseen that carbon-capture
will become of heightened importance. Revenue from carbon taxes could also be used to develop this
technology for application in areas with high emissions. As most policies will reduce profits for miners,
they are likely to figure out how to mine in an environmentally favourable fashion, or mine in times
where electricity costs for example are lower.
As such, there already exist select approaches to cleaner mining (Singh, 2021). Mining operations
that align with sustainability goals currently provide a great opportunity for miners globally. As such,
miners may also explore partnering with local energy companies and start-ups that provide affordable
access to renewable energy. This will allow them to avoid carbon taxed energy sources.

28
3.4 Inevitability of Proof-of-Stake (PoS)
The energy consumption of PoS blockchains should also be considered by regulators. This could be
seen in the example of Ethereum, arguably the second most notable blockchain, which is making a
transition to PoS approximately in Q3 2022. Given the difference in its security protocol, the protocol
might affect electricity consumption over a forecasted period and thus require a different modelling
approach.

3.4.1 Changes from Ethereum 1.0 to Ethereum 2.0

The main change involves the actual specification of who can become a validator. As explained in the
introduction, the Ethereum 2.0 protocol will rely on Proof-of-Stake rather than Proof-of-Work. Under
PoS validators will be chosen by the network rather than having to compete against another.

3.4.2 What This Means For Emissions

Given that there is no longer a need to calculate cryptographic puzzles, the PoS algorithm similarly
does not require the constant computation by all nodes in the network to determine who gets to write a
block. Therefore, equipment of lower computational power can be used and still be profitable. Rough
estimates for the energy usage of Ethereum 2.0 (PoS based) were calculated and many believe it will
lead to around 99.95% reduction in energy usage (Ethereum Foundation, 2022).
Given this consideration, the calculation for the total electricity consumption of the network will
take a different form as elaborated below.

3.4.3 Modified Powell et al. Model

The Powell, Hendon, Mangle, and Wimmer (2021) model povides a very simple mathematical approx-
imation for the total network electricity consumption of any PoS blockchain. Fundamentally, it only
depends on two input factors:

Energy Per Server × Number of Servers × 24(hours per day) × 365 days (Annualization) (19)

The energy per server is simply estimated from the Thermal Design Power (TDP) of designated servers
which run as a validator for the network. TDP stands for Thermal Design Power, in watts, and refers
to the power consumption under the maximum theoretical load. This will provide one with the upper
bound for the energy use per validator. A high energy usage per server approximation is used, following
the convention from Platt et al. (2021) and Powell et al. (2021), by utilizing the TDP of representative
enterprise rack servers (see Table 14).

Year Model Full Load (Using Wh Conversion

TDP of Proces- (Annualizing)
sor)
2015 R730 - 2014 344 3013440
2016 R730 - 2015 344 3013440
2017 R740 - 2017 205 1795800
2018 R740 - 2017 205 1795800
2019 R740 - 2018 205 1795800
2020 R7515 - 2019 280 2452800
2021 R7515 - 2019 280 2452800
2022 ··· R7515 - 2019 280 2452800
2030

Table 14: Dell Server Power Rating (Proxy For All Server Used For Mining).

For the forward forecasts, the model assumes that the equipment will not be upgraded past 2022.
This is because there is no more requirement for these equipment to be on the cutting edge of com-
putational ability after the Ethereum 2.0 merge. Practically, there is no need to make an assumptions
around what the TDP of future equipment will be.

29
 The next input is the number of servers. The model diverges from the Powell et al. model, where
they use actual network statistics. To make this calculation an actual upper bound, one observe the
historical number of Ethereum addresses with ≥ 32 ETH11 , which is the stake required to become a
validator on the Ethereum network.
With this, the estimates from other researchers are plotted and PoS electricity consumption statis-
tics are applied retrospectively to understand what the electricity consumption of Ethereum 2.0 would
have looked like if it were implemented at its genesis. In the plots, one see that on average, Ethereum
2.0 would have used approximate 97% less electricity compared to Ethereum 1.0 (From January 2020
to March 2022), on average, since inception. The most important thing to note here is that unlike the
previous Krause-Tolaymat model, the model calculates electricity consumption on a daily annualized
measure to make it easily comparable against all the other estimates, as done in McDonald, (2018).

Figure 11: Various Estimates From Researchers Vs. PoS.

11 This information is easily found on https://studio.glassnode.com/metrics?a=ETH&m=addresses.Min32Count

30
3.4.4 Potential On-Chain Intervention
One intervention method that was not considered before was the ability to perform on-chain inter-
ventions. This report defines on-chain intervention as the application of policies by the network to
influence the electricity consumption and carbon output of the network.
This is likely the most direct and impactful method that could be implemented would be a hardware
power efficiency requirement. Through equation (19) of the Powell et al. model, it will directly reduce
the electricity consumption of the network while being fundamentally unchanged as a network. This
does not just have to target enterprise-level servers, but could also be required of consumer-grade
hardware, which is consistent with what the Ethereum 2.0 merge will allow for. This model omits
the calculation of any period, as this will simply be a linear scaling of the results seen in Figure 11.
However, this should not undermine the effectiveness of such an approach, and could potentially be
considered as one of the many methods that could bring emissions further in line with other green
agendas.

3.4.5 Unnecessary Regulatory Interventions

In the case of Ethereum and other PoS blockchains, the electricity consumptions savings are significant,
and would likely mean that blockchains like Ethereum will not fall under the radar of regulators. As
mentioned in Section 3.4, the potential for an on-chain intervention would likely be much more im-
pactful and meaningful rather than any policy from government intervention. That said, interventions
will likely still be relevant for the class of PoW blockchains.

4 Discussion and Conclusion

4.1 Discussion of Various Interventions
While the ban on Bitcoin mining in China may be driven by a wide range of different motives, for the
purpose of this paper it will be assumed (perhaps naively) that an ambition to support the country’s
environmental transition underlies the policy’s primary motive. As such, on the aggregate level it can
be observed that the direct impact of various interventions on the Bitcoin network is likely to bring
down electricity consumption to some extent, however, not nearly as extensive as the ban that China
has opted for. Not only does the ban bring down average electricity consumption, but it also appears
to have avoided a peak consumption year in 2021 (see Figure 12 black bar) which likely resulted from
the positive traction around Bitcoin price in 2021.
The implementation of the Chinese ETS successfully led to a reduction in carbon emissions and
electricity consumption by 2030, however, even a highly aggressive reduction in allocated permits
results in a relatively limited reduction of 8.49% of Chinese CO2 emissions arising from the Bitcoin
network. This is due to the Chinese Bitcoin miners’ ability to counteract the policy intervention by
simply purchasing additional permits from a highly liquid market. While the exact magnitude of
the profits from Bitcoin mining is hard to quantify, it has been reported that it has been ”one of
the most profitable businesses on the planet” (Fortune, 2021). This supports the model’s findings as
any additional costs associated with carbon permits do not significantly reduce the Chinese Bitcoin
industry’s profits. Furthermore, the Bitcoin industry’s limited share of the total Chinese ETS market
implies that additional demand from the Bitcoin industry is translating into far from a proportional
increase in the market price of carbon permits. Instead, the governing dynamics of the ETS market
will be determined by considerably larger industries covered by the ETS and the individual decisions
of the firms constituting those industries. For example, the power sector in China is reported to be
responsible for 38% of the country’s total CO2 emissions (Liu, Zheng, Zhao, Chen, & Lugovoy, 2018)
while this paper estimates that the Chinese Bitcoin miners contribute only 0.09% of China’s total
CO2 emissions in 2021. However, it is to be noted that the decrease in carbon emissions from the
Chinese Bitcoin industry under a more stringent case is in close proximity to the estimated decrease in
carbon emissions from the Chinese power sector under a Chinese ETS. Specifically, the International
Energy Association estimates that under a firmer ETS specification the CO2 emissions from electricity
generation would decrease by 12% between 2020-2030 (IEA, 2021).
The more aggressively priced carbon taxes generally achieved a greater amount of emission reduc-
tion. Under the carbon tax ”high” scenario, Chinese Bitcoin emissions are reduced by up to 27.7%

31
 Figure 12: Bitcoin and World (2019) Electricity Consumption.

compared to the no ban case. But the implementation of carbon prices at the height of the carbon
tax ”high” scenario in China is ambitious, as only Lichtenstein, Switzerland, and Sweden had carbon
taxes above $100 t/CO2 in 2021 (World Bank, 2021b). For both the ETS and carbon taxes, however,
emission and electricity reductions become more significant post the 2024 Bitcoin halving, momentarily
even providing an even lower electricity consumption reduction than the ”Ban” case. The reason, as
explained in the end of Section 3.3.3, is due to halving reducing the supply of Bitcoins, thereby decreas-
ing the reward available for miners given the same amount of costs. Therefore the efficiency of both
interventions shows dependency on the state of the global Bitcoin network. Yet, it is important that
emission reductions occur sooner than later (Luz, 2022), in order to avoid compounding environmental
damages over time and the surpassing of irreversible global climate states.

Figure 13: Electricity consumption and CO2 emission forecasts for the global Bitcoin network.

The effect of the Carbon Tax intervention and the ETS on Chinese CO2 emissions from Bitcoin
mining can be seen in Table 15. Evidently, the two schemes differ markedly in the actual price that
Bitcoin mining firms are required to pay for every tonne of CO2 emitted. In the most aggressive
case, the Bitcoin mining industry is being incurred a cost of US$130.0 and US$34.2 per tonne carbon

32
dioxide emitted for the Carbon Tax and ETS respectively. This illustrates a potential benefit of
the Carbon Tax intervention as policymakers can decide and implement the exact average price of
emissions. However, this also presumes that policymakers have an adequate ability of determining the
optimal carbon price (SCC) which is a notoriously difficult task due to both theoretical and political
circumstances as discussed in Section 2.1.5.
It is worth noting the Absolute Reduction in Chinese BTC Emissions vs. No Ban for scenario CT
50 Percent and ETS Case A. While the reduction is essentially equal (-8.4% vs. -8.5% respectively), the
two cases differ markedly in the magnitude of the reduction in profit (-27.6% vs. -21.4% respectively).
This is the result of a combination of factors, such as the balancing act of production and an increasing
price of carbon emissions. For example, the exact timing of when the carbon tax increases can have
a varying effect on Bitcoin mining operations depending on the current market price of Bitcoin, hash
rate contributions, and previous investment decisions by firms. While this result is not conclusive on
which intervention can weigh those considerations most effectively, it illustrates that under current
assumptions an ETS would be able to reduce emissions with less associated economic costs.
In summary, the analysis of the three policy interventions has illustrated an important point.
While the theoretical implementation of an ETS or a carbon tax can provide significant benefits
over an outright ban, the result of a practical implementation can differ markedly due to a number
of factors. Specifically, for the ETS the relatively small share of the Bitcoin industry of the ETS
and the industry’s ability to generate large profits limit the ETS’s effectiveness in delivering a targeted
intervention and subsequent reduction in CO2 emissions. Similarly, aggressive carbon prices are needed
to counteract this profit-generating ability. As carbon taxes are based on the SCC which is calculated
using Integrated Assessment Models that capture global climate change effects, some might argue that
modelling adjustments are needed to provide specific SCC estimates in relation to Bitcoin mining.
Otherwise, carbon taxes will continue to be set according to political agendas. As agendas that
actively promote emission reductions are arguably rarer, this will likely result in low carbon taxes
that keep industrial stakeholders happy (Dyarto & Setyawan, 2020). The significant revenue carbon
taxes generate, which mustn’t be necessarily used for further climate change mitigation efforts, may
nevertheless incentive governments to set higher carbon taxes. Given the ban, China is willing to turn
its back on the Chinese Bitcoin Mining industry but will avoid setting a high carbon tax specific to
the industry in order not to set precedent for other industries in the domestic economy.
These considerations are crucial for policymakers when deciding on an adequate policy. If, as
assumed, the main objective was to reduce the negative environmental impact associated with Bitcoin
mining, then an outright ban is shown to provide the most benefit. Both the ETS and Carbon Tax
intervention under their current specifications suffers from their own shortcomings in their absolute
ability to reduce CO2 emissions, which would warrant the need for a more drastic policy solution.

Average Absolute Absolute Average Absolute Emissions

Price for Reduction Reduction Profit Per Reduc- Reduced
Tonne in Chinese in Global Tonne tion in / Profit
Carbon BTC Emis- BTC Emis- CO2 Profit Reduced
Emitted sions vs. sions vs. Emitted
(US$) No Ban No Ban (US$)
CT 50 % 45.0 -8.4% -3.5% 105.0 -27.6% 0.31
CT Paris 90.0 -17.4% -5.9% 71.9 -54.5% 0.32
CT High 130.0 -27.8% -8.5% 43.3 -75.0% 0.37
ETS A 5.6 -1.3% -0.7% 137.8 -4.0% 0.33
ETS B 20.3 -5.5% -2.0% 130.3 -12.7% 0.43
ETS C 34.2 -8.5% -2.9% 120.9 -21.4% 0.39
China 0.0 -100.0% -25.7% 0.0 -100.0% 1.00
Ban

Table 15: Comparative statistics for carbon tax, ETS, and ban cases for the period June 2021 to
December 2030

33
4.2 Conclusion
4.2.1 Summary
This paper investigated the environmental effects of China’s recent regulatory steps against cryptocur-
rencies and attempts to bridge a gap in today’s academic literature by offering an insight into the
varying environmental consequences of a set of hypothetical country specific regulations on the global
cryptocurrency ecosystem. The paper uses two baseline scenarios; one considered the ban scenario and
one a world without the ban taking place. They were then evaluated against two alternative policies,
an emission trading system and a carbon tax.
It is found that under China’s policy of banning cryptocurrencies, the total emissions from the
global Bitcoin network between June 2021 to December 2030 will be 25.7% lower than under the
no-ban scenario. In comparison, if the most stringent ETS case (Case C) is considered, an ETS
implementation would achieve a total reduction of 2.9% in global CO2 emissions over the same period.
The relatively low reduction in emissions stems from the inherent problem that cryptocurrency mining
will remain a small player in the permit market and given its profits will be able to keep buying permits.
Meanwhile, under the Paris Agreement carbon tax case the global Bitcoin network’s CO2 emissions
would decline by 5.9% relative to the no-ban scenario. Nevertheless, for more significant emission
reductions even higher and thus more aggressive carbon prices are needed to limit the excessive profit
generating ability of the Chinese Bitcoin network. Further if the Chinese Bitcoin network was subjected
to some of the highest carbon taxes observed globally, as seen in the carbon tax ”High” scenario, this
would only lead to a 8.5% reduction in the global Bitcoin network’s CO2 emissions.
Overall, this paper suggests that when the ban scenario is evaluated against alternative policies,
and the primary focus lies on environmental objectives, a cryptocurrency ban is the most efficient and
fastest policy option. Between June 2021 and December 2030 the total reduction in CO2 emissions is
forecasted to be 22.8% higher than under the most stringent ETS case and 19.8% higher than under
a carbon tax consistent with the Paris Agreement.
As a result, China’s cryptocurrency ban appears sensible from a solely environmental perspective
and the current state of the cryptocurrency environment. However, a ban comes at a significant
economic cost that may no longer be necessary under energy efficient protocols. Going forward the
transition towards less energy intensive protocols such as PoS will play a vital role in the decision
making process of governments. Although the proposal was rejected, the EU already signalled a
more positive stance towards more environmentally friendly protocols with its proposal to limit the
use of PoW protocols (CoinDesk, 2022). These findings could subsequently be used to investigate
the economic opportunity cost of a cryptocurrency ban to present a policy case that extends beyond
factors other than the environment.

4.2.2 Difficulty of Implementation

There is a variety of top-down and bottom-up models that have been used in the past to estimate the
environmental impact of blockchain protocols. As a result, findings can diverge significantly reflecting
the use of different estimation techniques and assumptions. Given the uncertainty around climate
change and the evolving politics surrounding climate policy, assumptions have to be made that can
be subject to drastic change at any point in time. For instance, China’s energy mix, a change in the
CCP’s stance towards its energy goals would have a substantial impact on the CO2 emissions of miners
under the no-ban scenario.
Furthermore, financial markets and specifically cryptocurrency markets have been subject to boom
and bust cycles. The possibility of a significant downturn or shift in consumer behaviour away from
cryptocurrencies should not be neglected but cannot be accounted for by the model. Therefore, the
model has to make an assumption about consumer and miner behavior which is inherently difficult.
Additionally, when it comes to the implementation of a carbon tax or a ETS, policymakers will be
guided by the estimated social cost of carbon (SCC). This paper acknowledges that determining the
most adequate SCC is complicated, and has therefore rested most of its analysis on existing research
on the matter.
Lastly, modelling market dynamics is a notoriously difficult task and is an area under active re-
search. For the sake of simplicity, the ETS model assumes an overly simplified trading mechanism
that elicit price responses as a result of supply and demand imbalances. The ETS and Carbon Tax
model and its underlying Agent Based Model also disregards the intertemporal considerations by the

34
economic agent commonly observed in established IAMs. Specifically, Chinese Bitcoin miners only
consider their investment decision in future production of Bitcoin and disregards considerations such
as switching to low-carbon fuels or investment in emission abating technologies. The model ultimately
tries to understand the aggregate result of these dynamics through the sensitisation of hyperparame-
ters, but acknowledges there still exist difficulty in not only fine tuning the model, but representing a
complicated process with a simple and limited set of mathematical equations.

4.2.3 What This Means

Given the limitations mentioned, this model relies on a set of relatively conservative assumptions that
position the estimates towards the lower end of the spectrum compared to other papers. However, the
model still provides valuable insights into the analysis that policymakers have to make. By comparing
the Chinese ban scenario with two policy alternatives, the findings on the effectiveness of alternative
interventions can be used for other countries in their decision making process.

4.2.4 Further Research

Above this analysis, there is value in using the various intervention frameworks to understand the
economic value of the Chinese cryptocurrency ban further. For example, using the carbon tax frame-
work, one can understand the opportunity cost of their actions against a no-ban scenario. This could
have been a revenue stream that the Chinese government could have used for investments in green
technology and might have further positive externalities for the wider economy.
As with the many iterations of IAMs, finding calibrated parameters for models that reflect empirical
findings is also an area that would warrant further research. It would also be worthwhile to further
refine profit functions for miners, and help determine their optimal reinvestment decisions, a question
first brought up by researchers like Jiang et al. (2021).

35
References
Bayer, P., & Aklin, M. (2020). The european union emissions trading system reduced co¡sub¿2¡/sub¿
emissions despite low prices. Proceedings of the National Academy of Sciences, 117 (16), 8804-
8812. Retrieved from https://www.pnas.org/doi/abs/10.1073/pnas.1918128117 doi: 10
.1073/pnas.1918128117
BBC. (2019). Bitcoin mining ban considered by china’s economic planner. Retrieved from https://
www.bbc.co.uk/news/technology-47867031
BBC. (2021a). China declares all crypto-currency transactions illegal. Retrieved from https://
www.bbc.co.uk/news/technology-58678907
BBC. (2021b). China power cuts: Coal miners ordered to boost output, say reports. Retrieved from
https://www.bbc.co.uk/news/business-58839894
Beekhuizen, C. (2021). Ethereum’s energy usage will soon decrease by 99.95/. Retrieved from
https://blog.ethereum.org/2021/05/18/country-power-no-more/
Bevand, M. (2017). Electricity consumption of bitcoin: A market-based and technical analysis. Re-
trieved from http://blog.zorinaq.com/bitcoin-electricity-consumption/#fnref:refD:
1
Calel, R., & Dechezleprêtre, A. (2016). Environmental policy and directed technological change:
evidence from the european carbon market. Review of economics and statistics, 98 (1), 173–191.
Cambridge, U. (2022). Cambridge bitcoin electricity consumption index. Retrieved from https://
ccaf.io/cbeci/index/ (Accessed: 2022-02-27)
Carleton, T., & Greenstone, M. (2021, Nov). Working paper updating the united states government’s
social cost of carbon. Retrieved from https://bfi.uchicago.edu/wp-content/uploads/2021/
01/BFI WP 202104.pdf
Cheng, E. (2021). Secretive chinese bitcoin mining company may have made as much money as nvidia
last year. Retrieved from https://www.cnbc.com/2018/02/23/secretive-chinese-bitcoin
-mining-company-may-have-made-as-much-money-as-nvidia-last-year.html
ChinaPower. (2022). How is china’s energy footprint changing? Retrieved from https://chinapower
.csis.org/energy-footprint/
Cleetus, R. (2021, Jun). The social cost of carbon gets an interim update from the biden admin-
istration. Retrieved from https://blog.ucsusa.org/rachel-cleetus/the-social-cost-of
-carbon-gets-an-interim-update-from-the-biden-administration/
CoinDesk. (2022). Proposal limiting proof-of-work is rejected in eu parliament committee vote. Re-
trieved from https://www.coindesk.com/policy/2022/03/14/proposal-limiting-proof-of
-work-is-rejected-in-eu-parliament-committee-vote-sources/
Cui, J., Wang, C., Zhang, J., & Zheng, Y. (2021). The effectiveness of china’s regional
carbon market pilots in reducing firm emissions. Proceedings of the National Academy of
Sciences, 118 (52), e2109912118. Retrieved from https://www.pnas.org/doi/abs/10.1073/
pnas.2109912118 doi: 10.1073/pnas.2109912118
de Vries, A. (2020). Bitcoin’s energy consumption is underestimated: A market dynamics approach.
Energy Research Social Science, 70 , 101721. Retrieved from https://www.sciencedirect.com/
science/article/pii/S2214629620302966 doi: https://doi.org/10.1016/j.erss.2020.101721
de Vries, A., & Stoll, C. (2021). Bitcoin’s growing e-waste problem. Resources, Conservation and
Recycling, 175 , 105901. Retrieved from https://www.sciencedirect.com/science/article/
pii/S0921344921005103 doi: https://doi.org/10.1016/j.resconrec.2021.105901
de Best, R. (2022). Bitcoin mining by country 2021.
Dyarto, R., & Setyawan, D. (2020, Sep). Understanding the political challenges of introducing
a carbon tax in indonesia - international journal of environmental science and technology.
Springer Berlin Heidelberg. Retrieved from https://link.springer.com/article/10.1007/
s13762-020-02925-4
Ethereum. (2022). Proof-of-stake (pos). Retrieved from https://ethereum.org/en/developers/
docs/consensus-mechanisms/pos/
Ethereum Foundation. (2022, Mar). Ethereum energy consumption. Author. Retrieved from https://
ethereum.org/en/energy-consumption/
European Commission. (2021). Emissions cap and allowances. https://ec.europa.eu/clima/eu
-action/eu-emissions-trading-system-eu-ets/emissions-cap-and-allowances en. (Ac-
cessed: 2022-03-10)

36
European Commission. (2021). Report from the commission to the european parliament and the
council. (962 final). Retrieved from https://ec.europa.eu/clima/system/files/2021-10/
com 2021 962 en.pdf
Fortune. (2021). Bitcoin mining is suddenly one of the most profitable businesses on the planet.
Retrieved 2022-04-14, from https://fortune.com/2021/08/05/bitcoin-mining-is-suddenly
-one-of-the-most-profitable-businesses-on-the-planet/
Goodkind, A. L., Jones, B. A., & Berrens, R. P. (2020). Cryptodamages: Monetary value estimates
of the air pollution and human health impacts of cryptocurrency mining. Energy Research So-
cial Science, 59 , 101281. Retrieved from https://www.sciencedirect.com/science/article/
pii/S2214629619302701 doi: https://doi.org/10.1016/j.erss.2019.101281
Gray, M. (2021). Turning the supertanker. ThinkZero. https://www.transitionzero.org/reports/
turning-the-supertanker.
Hanley, N., Shogren, F., & White, B. (2019). Introduction to environmental economics (3rd ed.). OUP
Oxford.
Hayes, A. (2015). A cost of production model for bitcoin. Retrieved from http://www
.economicpolicyresearch.org/econ/2015/NSSR WP 052015.pdf
Hope, C. (2011, Apr). Working paper series - cambridge judge business school. Retrieved from
https://www.jbs.cam.ac.uk/wp-content/uploads/2020/08/wp1104.pdf
IEA. (2020). Implementing effective emissions trading systems. https://www.iea.org/reports/
implementing-effective-emissions-trading-systems. (Accessed: 2022-03-10)
IEA. (2021). The role of china’s ets in power sector decarbonisation. IEA Publications. https://
www.iea.org/reports/the-role-of-chinas-ets-in-power-sector-decarbonisation.
Jiang, S., Li, Y., Lu, Q., Hong, Y., Guan, D., Xiong, Y., & Wang, S. (2021). Policy assessments for
the carbon emission flows and sustainability of bitcoin blockchain operation in china. Nature
Communications, 12 , 1938. doi: https://doi.org/10.1038/s41467-021-22256-3
Kaiser, B., Jurado, M., & Ledger, A. (2018). The looming threat of china: An analysis of chinese
influence on bitcoin. arXiv preprint arXiv:1810.02466 .
Knight, C. (2013). What is grandfathering? Environmental Politics, 22 (3), 410-427. Retrieved from
https://doi.org/10.1080/09644016.2012.740937 doi: 10.1080/09644016.2012.740937
Krane, J., & Finley, M. (2022, Mar). What is the ’social cost of carbon’ ? 2 energy experts explain
after court ruling. Retrieved from https://theconversation.com/what-is-the-social-cost
-of-carbon-2-energy-experts-explain-after-court-ruling-176255
Krause, M., & Tolaymat, T. (2018). Quantification of energy and carbon costs for mining cryptocur-
rencies. Nature Sustainability, 1 , 711-718. doi: https://doi.org/10.1016/j.erss.2019.101281
Laing, T., Sato, M., Grubb, M., & Comberti, C. (2013, January). Assessing the effectiveness of the
EU Emissions Trading System (GRI Working Papers No. 106). Grantham Research Institute
on Climate Change and the Environment. Retrieved from https://ideas.repec.org/p/lsg/
lsgwps/wp106.html
Lebaron, B. (2001, 02). A builder’s guide to agent-based financial markets. Quantitative Finance, 1 ,
254-261. doi: 10.1088/1469-7688/1/2/307
Liu, Q., Zheng, X.-Q., Zhao, X.-C., Chen, Y., & Lugovoy, O. (2018). Carbon emission scenarios of
china’s power sector: Impact of controlling measures and carbon pricing mechanism. Advances
in Climate Change Research, 9 (1), 27-33. Retrieved from https://www.sciencedirect.com/
science/article/pii/S1674927817300916 (Including special topic on China Energy Modeling
Forum) doi: https://doi.org/10.1016/j.accre.2018.01.002
Luz, S. (2022, Apr). The evidence is clear: The time for action is now. we can halve emissions by
2030. Retrieved from https://www.ipcc.ch/2022/04/04/ipcc-ar6-wgiii-pressrelease/
Mackenzie, W. (2021, Mar). Significant increase in carbon pricing is key in 1.5-degree
world. WoodMac.Site.Features.Shared.ViewModels.Metadata.Publisher. Retrieved from
https://www.woodmac.com/press-releases/significant-increase-in-carbon-pricing
-is-key-in-1.5-degree-world/
McCook, H. (2018). The cosst amp; sustainability of bitcoin.
McDonald, K. (2021, Dec). Ethereum emissions: A bottom-up estimate.
doi: https://doi.org/10.48550/arXiv.2112.01238
min Yu, S., Fan, Y., Zhu, L., & Eichhammer, W. (2020). Modeling the emission trading scheme from an
agent-based perspective: System dynamics emerging from firms’ coordination among abatement

37
 options. European Journal of Operational Research, 286 (3), 1113-1128. Retrieved from https://
www.sciencedirect.com/science/article/pii/S0377221720303131 doi: https://doi.org/10
.1016/j.ejor.2020.03.080
Mora, C., R.L., R., Taladay, K., Kantar, M., Chock, M., Shimada, M., & Franklin, E. (2018). Bitcoin
emissions alone could push global warming above 2°c. https://www.nature.com/articles/
s41558-018-0321-8.pdf. (Accessed: 2022-03-10)
Nordhaus, W. (2016, Nov). Revisiting the social cost of carbon — pnas. Retrieved from https://
www.pnas.org/doi/10.1073/pnas.1609244114
Nordhaus, W., & Sztorc, P. (2013, Oct). Dice 2013r - yale university. Retrieved from http://www
.econ.yale.edu/~nordhaus/homepage/homepage/documents/DICE Manual 100413r1.pdf
O’Dwyer, K. J., & Malone, D. (2014, Jun). Bitcoin mining and its energy footprint. ISSC .
Petrick, S., & Wagner, U. (2014). The impact of carbon trading on industry: Evidence from german
manufacturing firms (Kiel Working Papers No. 1912). Kiel Institute for the World Economy
(IfW Kiel). Retrieved from https://EconPapers.repec.org/RePEc:zbw:ifwkwp:1912
Platt, M., Sedlmeir, J., Platt, D., Xu, J., Tasca, P., Vadgama, N., & Ibanez, J. (2021, Sep). Energy
footprint of blockchain consensus mechanisms beyond proof-of-work. arXiv . Retrieved from
https://arxiv.org/pdf/2109.03667v5.pdf
Powell, L. M., Hendon, M., Mangle, A., & Wimmer, H. (2021). Awareness of blockchain usage,
structure, generation of platform’s energy consumption: Working towards a greener blockchain.
Issues in Information Systems, 22 (1), 114–123. doi: https://doi.org/10.48009/1 iis 2021 114
-123
Refinitiv. (2021). China’s national carbon market exceeds expectations. https://
www.refinitiv.com/perspectives/future-of-investing-trading/chinas-national
-carbon-market-exceeds-expectations/. (Accessed: 2022-03-10)
Romann, A. (2013). China leading the bitcoin charge. Retrieved from https://www.chinadailyasia
.com/business/2013-11/22/content 15100696.html
Ros, D. (2013, Feb). Why we risk overinterpreting china’s carbon tax statements. Retrieved
from https://www.carbonbrief.org/why-we-risk-overinterpreting-chinas-carbon-tax
-statements#:~:text=In%202010%2C%20the%20ministry%20suggested,5)%20per%20tonne%
20by%202020.
Shen, S., & Galbraith, A. (2021a, Jun). China’s ban forces some bitcoin miners to flee overseas,
others sell out. Thomson Reuters. Retrieved from https://www.reuters.com/technology/
chinas-ban-forces-some-bitcoin-miners-flee-overseas-others-sell-out-2021-06-25
Shen, S., & Galbraith, A. (2021b). Why china is leading the global rise of bitcoin.
Retrieved from https://www.reuters.com/technology/chinas-ban-forces-some-bitcoin
-miners-flee-overseas-others-sell-out-2021-06-25/
Singh, R. (2021, Nov). Cop26: How to make cryptocurrency sustainable. Retrieved from
https://www.weforum.org/agenda/2021/11/how-to-make-bitcoin-and-cryptocurrency
-sustainable-renewable-energy-blockchain-crypto-mining/
Southurst, J. (2013). Why china is leading the global rise of bitcoin. Retrieved
from https://www.coindesk.com/markets/2013/11/18/why-china-is-leading-the-global
-rise-of-bitcoin/
Stern, N., Taylor, C., Karlsson, K., & Stiglitz, J. (2022, Jan). A social cost of carbon consistent with a
net-zero climate ... Retrieved from https://rooseveltinstitute.org/wp-content/uploads/
2022/01/RI Social-Cost-of-Carbon 202201-1.pdf
Stoll, C., Lena Klaaßen, L., & Gallersdörfer, U. (2019, July). The carbon footprint of bitcoin. Joule(3).
doi: https://doi.org/10.1016/j.joule.2019.05.012
The State Council. (2021). China maps path to carbon peak, neutrality under new development philos-
ophy. Retrieved from http://english.www.gov.cn/policies/latestreleases/202110/24/
content WS61755fe9c6d0df57f98e3bed.html
Trading Economics. (2022). Eu carbon permits. Retrieved 2022-04-14, from https://
tradingeconomics.com/commodity/carbon
U.S. Bureau of Labor Statistics. (N.A.). Bureau of labor statistics data on price of electricity. U.S.
Bureau of Labor Statistics. Retrieved from https://data.bls.gov/timeseries/APU000072610
?amp%5C%253bdata tool=XGtable&output view=data&include graphs=true
von Carnap, K. (2021). Unchaining blockchain in china. Retrieved from https://merics.org/sites/

38
 default/files/2021-03/MERICS%20Primer%20Unchaining%20Blockchain%20in%20China%
202021.pdf
Vranken, H. (2017). Sustainability of bitcoin and blockchains. Current Opinion in Environmental
Sustainability, 28 , 1-9. Retrieved from https://www.sciencedirect.com/science/article/
pii/S1877343517300015 (Sustainability governance) doi: https://doi.org/10.1016/j.cosust
.2017.04.011
Waldhoff, S., Anthoff, D., Rose, S., & Tol, R. S. J. (2014, Dec). The marginal damage costs of different
greenhouse gases: An application of fund. De Gruyter Open Access. Retrieved from https://
www.degruyter.com/document/doi/10.5018/economics-ejournal.ja.2014-31/html
Whitehouse, U. S. (2021, Feb). Technical support document: Social cost of car-
bon, methane, and nitrous oxide interim estimates under executive order 13990. In-
teragency Working Group on Social Cost of Greenhouse Gases, United States Gov-
ernment. Retrieved from https://www.whitehouse.gov/wp-content/uploads/2021/02/
TechnicalSupportDocument SocialCostofCarbonMethaneNitrousOxide.pdf
World Bank. (2021a, May). State and trends of carbon pricing 2021. Washington, DC: World Bank.
Retrieved from http://hdl.handle.net/10986/35620
World Bank. (2021b, May). State and trends of carbon pricing 2021. Washington, DC: World Bank.
Retrieved from http://hdl.handle.net/10986/35620
Ye, J. (2020, Feb). Options and considerations for a federal carbon tax. Retrieved from https://
www.c2es.org/document/options-and-considerations-for-a-federal-carbon-tax/
Zade, M., Myklebost, J., Tzscheutschler, P., & Wagner, U. (2019). Is bitcoin the only prob-
lem? a scenario model for the power demand of blockchains. Frontiers in Energy Research,
7 . Retrieved from https://www.frontiersin.org/article/10.3389/fenrg.2019.00021 doi:
10.3389/fenrg.2019.00021
Zhang, J. (2010, May). Retrieved from http://www.chinadaily.com.cn/china/2010-05/11/
content 9834525.htm

39