ArticlePDF Available

2020 CCAF 3rd Global Cryptoasset Benchmarking Study

Authors:

Abstract and Figures

This report reviews the impact of significant changes in the industry since the publication of the 2nd Global Cryptoasset Benchmarking Study in 2018. It provides novel insights into the state of the cryptoasset industry, having gathered data from 280 companies in 59 countries and across four main market segments – exchanges, payments, custody and mining.
Content may be subject to copyright.
Apolline Blandin, Dr. Gina Pieters, Yue Wu, Thomas Eisermann,
Anton Dek, Sean Taylor, Damaris Njoki
September 2020
3RD GLOBAL CRYPTOASSET
BENCHMARKING STUDY
supported by
Disclaimer: Data for this report has been gathered primarily from online surveys. While every reasonable effort has been made
to verify the accuracy of the data collected, the research team cannot exclude potential errors and omissions. This report should
not be considered to provide legal or investment advice. Opinions expressed in this report reect those of the authors and not
necessarily those of their respective institutions.
TABLE OF CONTENTS
FOREWORDS ..................................................................................................................................................4
RESEARCH TEAM ..........................................................................................................................................6
ACKNOWLEDGEMENTS ............................................................................................................................ 7
EXECUTIVE SUMMARY ........................................................................................................................... 11
METHODOLOGY ........................................................................................................................................14
SECTION 1: INDUSTRY GROWTH INDICATORS ......................................................................... 17
Employment gures ................................................................................................................................................ ..............................................................17
High-growth enterprises ................................................................................................................................................ ....................................................18
Financial performance of service providers .............................................................................................................................................................. 19
SECTION 2: MINING, THE BACKBONE OF THE INDUSTRY ................................................... 21
Hashing as a business ........................................................................................................................................................................................................... 21
Where are we on PoW’s energy consumption? ....................................................................................................................................................... 26
Mining pools .............................................................................................................................................................................................................................. 28
Mining hardware manufacturing ................................................................................................................................................. ...................................32
The nancialisation of mining............................................................................................................................................................................................34
SECTION 3: THE OFF-CHAIN STORY ................................................................................................ 36
On-chain and off-chain stories ................................................................................................................................................. ......................................36
Off-chain cryptoassets and at currency support ................................................................................................................................................ 37
Off-chain activity providers ................................................................................................................................................. ............................................. 39
SECTION 4: PROFILING CRYPTOASSET USERS ...........................................................................44
User number and activity .................................................................................................................................................................................................. 44
User geography ................................................................................................................................................. ......................................................................45
User types ................................................................................................................................................ ..................................................................................46
SECTION 5: REGULATORY AND COMPLIANCE ..........................................................................49
Compliance benchmarks.....................................................................................................................................................................................................49
Authorisation of service providers ................................................................................................................................................ ................................52
AML and KYC procedures ................................................................................................................................................................................................. 54
Regulation impact: redening geographies ................................................................................................................................................. ..............56
SECTION 6: IT SECURITY ........................................................................................................................ 59
The development of best market practices ............................................................................................................................................................... 59
Resources allocation to IT security ................................................................................................................................................ ................................60
Security audits ................................................................................................................................................ .........................................................................61
Insurance ................................................................................................................................................ ....................................................................................62
SECTION 7: BALANCING BETWEEN INTEGRATION AND INNOVATION ....................... 64
Enhanced transparency and compliance ....................................................................................................................................................................64
A decoupling of functions across the value chain ................................................................................................................................................. .65
The growth of ‘decentralised nance’ ................................................................................................................................................. .........................65
APPENDIX ...................................................................................................................................................... 67
Miners’ inuence ....................................................................................................................................................................................................................67
Operational risks for miners ............................................................................................................................................................................................. 67
Additional risks for miners ................................................................................................................................................................................................. 68
Operational risks for service providers ................................................................................................................................................ ....................... 69
Future developments ................................................................................................................................................ ...........................................................70
3rd Global Cryptoasset Benchmarking Study
4
FOREWORDS
The rapid pace of innovation and increased investment in the cryptoasset industry
is increasing the need for information analysing these developments. With the
publication of the rst edition of the Global Cryptoasset Benchmarking Study
three years ago, the CCAF set out to progressively track and take the pulse of
this nascent industry by transparently collecting, analysing and disseminating
knowledge about cryptoassets. Similarly, the 3rd Global Cryptoasset
Benchmarking Study seeks to shed light on the market dynamics of the
cryptoasset industry since late 2018.
The report collates data from entities operating in four main segments of the
industry: exchange, payments, custody, and mining. A total of 280 entities
from over 50 countries across various regions responded to the surveys. This
benchmarking report is compiled using data from one of the most comprehensive
and robust databases currently available in the cryptoasset industry.
The research ndings suggest that the industry has entered a growth stage
despite the notable headwinds the cryptoasset markets had encountered since
2018. Additionally, regulators’ collaborative dialogue and regulatory interventions
in the industry appear to be supporting its growth by providing regulatory clarity
and harmonisation on the treatment of cryptoassets and related activities. This
is an important development that has had immediate effects. For instance, the
publication of updated AML and CFT standards by the Financial Action Task Force
(FATF) in June 2019 encouraged compliance by industry participants, with an
increased share of the surveyed service providers performing KYC & AML checks
on their customers.
Nevertheless, our analysis has identied several hurdles – ranging from regulatory
compliance, IT security, and insurance – which need to be addressed for the
industry to grow to scale.
Our hope is that the ndings captured within this study will offer insight into the
evolution of the industry and inform the decisions that industry stakeholders
will face as the space matures. As with all of our research projects, we appreciate
that our ability to produce high quality research is highly dependent on the
cooperation of industry players and we extend our thanks to all the entities that
have contributed towards the publishing of this report. Finally, I want to gratefully
acknowledge the nancial support of Invesco as a long-standing supporter of
CCAF’s research and whose support made this study possible.
Dr. Robert Wardrop
Director
Cambridge Centre for Alternative Finance
3rd Global Cryptoasset Benchmarking Study
5
Despite the uncertainty and economic rollercoaster ride that 2020 has brought
us with the introduction of the global Covid-19 pandemic, we have learned that
even in trying times, businesses and markets have reached a critical point where
operations can sustain even a majority of their employees working remotely. Even
as the pandemic continues forward, nance still moves, and specically alternative
nance has its place in a post-pandemic world and the research and analysis of
trends in emerging still press forward.
2020 brings us the Cambridge Centre for Alternative Finance (CCAF) third
edition of its Global Cryptoasset Benchmarking Study. In this study the CCAF
gathered data points from approximately 280 entities including representation
from 59 countries across four main market segments: exchanges, payments,
custody, and mining. While most of the data was collected prior to the pandemic,
the aggregated learnings and insights from the report remain relevant in current
times.
This year’s cryptoasset benchmarking study comes at a particularly appropriate
time for Invesco as we completed a successful asset tokenization proof of concept
(POC) this year that explores various facets of the token lifecycle including the
creation and custody chain of real asset backed tokens and how they may be
distributed and exchanged in practice in the real world. The results of the POC
validate key ndings uncovered in the study in the areas of industry growth,
service providers, regulatory standards, and future outlook of cryptoassets. Our
own journey in token economics provided us with experience with third party
providers of token creation, digital token exchanges, token custodianship, and
navigating the complex legal and regulatory requirements for such an endeavor.
As we read through the study, a few highlights stood out in conrmation of our
own experiences. One notable observation was the growth patterns of full-time
equivalent (FTE) employees within the cryptoasset industry and at the rm level.
The industry saw overall slowdown in growth in employment; whereas individual
rms saw growth in FTEs indicating that while overall opportunities are shrinking,
the existing players are gaining traction and prominence within their area of
expertise. This is something we have witnessed ourselves rsthand through
the disappearance or consolidation of industry consortiums as the cryptoasset
industry sees increased participation from institutional investors and traditional
players in the nancial sector.
Invesco is proud to provide sponsorship to enable the Cambridge Centre of
Alternative Finance to continue their research in alternative nance industry
including this cryptoasset benchmarking report. We’d like to thank all of the
contributors in the research team who made this report possible through
collecting and analyzing data. These ongoing reports provide valuable insights
for benchmarking in emerging nancial markets and trends and scenarios that we
monitor to enable our own growth and the growth of alternative nance models in
general around the globe.
Dave Dowsett
Global Head of Technology Strategy, Emerging Technology,
and Intentional Innovation
Invesco
3rd Global Cryptoasset Benchmarking Study
6
RESEARCH TEAM
AUTHORS
Apolline Blandin: is a Research Associate at the Cambridge Centre for Alternative Finance and leads the
Centre’s cryptoasset research programme.
a.blandin@jbs.cam.ac.uk @ApollineBlandin
Gina Pieters (Dr.) is a Lecturer at the Kenneth C. Grifn Department of Economics, University of
Chicago, and a Research Fellow at the Cambridge Centre for Alternative Finance.
gcpieters@uchicago.edu @ProfPieters
Yue Wu is a Data Scientist at the Cambridge Center for Alternative Finance and helps to manage the
database and develop digital tools at the CCAF.
y.wu@jbs.cam.ac.uk @ClaireYueWu1
Thomas Eisermann is a Cryptocurrency and Blockchain Research Administrator at the Cambridge
Centre for Alternative Finance and assists with the data collection process.
t.eisermann@jbs.cam.ac.uk
Anton Dek (Dr.) is a Research Associate at the Cambridge Centre for Alternative Finance and leads
on the development of digital tools at the CCAF, such as the Global Alternative Finance Benchmark
Dashboard.
a.dek@jbs.cam.ac.uk @dektox
Sean Taylor is a Research Intern at the Cambridge Centre for Alternative Finance and is currently
pursuing a B.A in Entrepreneurship and Innovation at Lund University School of Economics and
Management.
Damaris Njoki is a Research Intern at the Cambridge Centre for Alternative Finance and a PhD student
at the University of Hudderseld.
d.njoki@jbs.cam.ac.uk
Other contributors include Felipe Ferri de Camargo Paes, Hatim Hussain, Steven Edwards,
Karim Nabil, Sean Stein Smith, Michel Rauchs, Bryan Zheng Zhang, and Nikos Yerolemou.
3rd Global Cryptoasset Benchmarking Study
7
ACKNOWLEDGEMENTS
The Cambridge Centre for Alternative (CCAF) would like to thank Invesco for sponsoring the research
study, and specically Dave Dowsett, Kevin Lyman, Henning Stein, Bradley Bell, and Heather Wied for
their continuous support throughout the research process.
We would also like to extend our gratitude to our research partners from the different regions. Without
the help of these industry associations, our survey dissemination would not have been possible. These
research partners were: Association pour le Developpement des Actifs Numeriques (ADAN), Asociación
Bitcoin Chile, Asociación FinTech Paraguay, Asociación FinTech Uruguay, Associação Brasileira de
Criptoeconomia (ABCripto), Association of Cryptocurrency Enterprises and Startups Singapore
(ACCESS Singapore), Bitcoin Argentina, Blockchain Nigeria User Group, Blockchain Ukraine Association,
Coin Center, Colombia Fintech, Crypto Valley Association, Fintech Mexico, Global Digital Finance (GDF),
German Blockchain Association, MinerUpdate, National Association of Blockchain and
Cryptotechnologies, SA Crypto, Thai Fintech Association, The Bitcoin Association of Hong Kong.
We greatly appreciate the help of the following media organisations, whose role was instrumental to
successfully disseminate our surveys and research: 8btc, ChainNews, CoinDesk, MinerUpdate and the
CryptoTool podcast.
We are also grateful to CryptoCompare for supplementing our survey data with additional data that
underpin their annual Exchange Benchmarking Report. We acknowledge their contribution wherever
applicable throughout the report.
We would also like to thank the entire CCAF team, especially Robert Wardrop, Raghavendra Rau, Hunter
Sims and Herman Smith, for their support. Special thanks go to Louise Smith for her fantastic design
work, as well as Kate Belger, Yvona Duncan, and Neil Jessiman for their hard work behind the scenes.
C
3rd Global Cryptoasset Benchmarking Study
8
In addition, we wish to thank Philippa Coney and Charles Goldsmith from the Cambridge Judge Business
School for their assistance in producing and publishing the report.
Finally, we would like to express our utmost gratitude to all survey respondents from across the globe
who participated in the surveys. Their contribution is core to the realisation of this study.
Note: some survey respondents prefer not to publicly disclose their participation.
сhilkoot
RGB
3rd Global Cryptoasset Benchmarking Study
9
MININGSTORE
3rd Global Cryptoasset Benchmarking Study
10
3rd Global Cryptoasset Benchmarking Study
11
EXECUTIVE SUMMARY
Over the past three years, the Cambridge Centre for Alternative Finance (CCAF) at the University of
CambridgeJudge Business School, has tracked and analysed the development of the global cryptoasset
industry. Since the publication of the 2nd Global Cryptoasset Benchmarking Study in December 2018,
the industry has undergone signicant changes: the 2017-2018 initial coin offering (ICO) bubble
has sparked closer scrutiny from regulators resulting in greater efforts with regards to regulatory
compliance, while new professional infrastructure and services have emerged to serve the increased
interest from institutional investors. Mining analysts, for their part, have suggested that nancial
engineering is underway in the mining sector.
This report reviews some of these market trends and provides insights into the state of the cryptoasset
industry. For the 3rd edition of its Global Cryptoasset Benchmarking Study, the CCAF gathered data
from 280 entities from 59 countries and across four main market segments, namely exchange, payments,
custody, and mining. The sample consists of 175 service providers, 75 mining companies and 30
individual miners. Data was collected between March and May 2020.
The key ndings from this global cryptoasset benchmarking study are as follows:
Analysing growth indicators of the cryptoasset industry
Full-time equivalent (FTE) employee growth slowed considerably following the late-2017 market
frenzy. Respondents across all market segments, reported year-on-year growth of 21% in 2019, down
from 57% in 2018.
Industry-wide, the growth in FTE employment declined by 36 percentage points between 2017 and
2019, whereas the median rm reported a 75-percentage point downward change in employment
growth. The difference in the industry-level and the rm-level employment growth gures reects the
rise of large rms within each industry group that dominates in the aggregate change in employment and
suggests that a few large players are dominating the industry.
However, not all rms are equal: individual rm employment data shows that a notable proportion
of companies (26%) have sustained an annualised growth in employment level above 10% over the
past three years. Using established denition and criteria, which sets the size threshold of a rm at
the beginning of the growth period at 10 FTE employees, these companies qualify as “high-growth
companies”.1
A deep dive into mining: from its environmental impact and the nancialisation of the sector to
centralisation concerns
The survey ndings estimate that on average 39% of proof-of-work mining is powered by renewable
energy, primarily hydroelectric energy. Understanding the energy source of mining is important because
electricity costs account for the majority of hashers’ operational expenditures - albeit with some
variability across world regions - and hashers have long competed on accessing the cheapest energy
source.
Faced with increased competition and tight prot margins, miners with access to sophisticated
nancial products, such as hashrate or cryptoasset derivatives, have begun using them to hedge their
risks (between 12% to 14% of all miners). This is paving the way for the nancialisation of mining.
1 Eurostat Glossary (2014) Glossary: High-growth enterprise. European Commission. Available from: https://ec.europa.eu/eurostat/
statistics-explained/index.php/Glossary:High-growth_enterprise [Last accessed: 20 August 2020].
3rd Global Cryptoasset Benchmarking Study
12
Signicant differences arise when comparing the cost structure of American and Chinese hashers:
capital expenditures - primarily constituted of mining equipment purchase - take up to 56% of
American hashers’ costs, compared to 31% for Chinese ones. This suggests that Chinese hashers
have a competitive edge in the acquisition of mining machines explained mostly by the concentration
of hardware manufacturers in China, leading to a shorter supply chain, easier business conduct (e.g.
language, working culture, local connections), and absence of international shipping fees and import
tariffs. Aligned with this, the study found that 52% of ASIC manufacturers total sales go to Chinese
hashers in 2019.
23% of surveyed hashers report receiving support from governments, primarily in the form of locally-
focused support, such as electricity subsidy for users within a region. Nearly two-fths of those receiving
local governmental support are based in China.
Off the chain story: understanding service providers’ internal ows
Aligned with 2018 ndings, new survey data shows that off-chain transactions, both in terms of
volumes and numbers, continue to be dominated by at-cryptoasset trades (and vice-versa), meaning
that users primarily interact with ‘gateway’ service providers, such as exchanges, to enter and leave the
cryptoasset ecosystem.
Usage seems to vary with the geographic location of the service provider. For instance, exchanges
based out of APAC record the highest share of cryptoasset-cryptoasset trades (40%) and most
transactions initiated at these exchanges are directed to the open-market (i.e. exchange’s order-book).
This suggests that APAC exchanges are primarily used for trading purposes.
Stablecoins are becoming increasingly available. The share of service providers supporting Tether grew
from 4% to 32% between 2018 and 2020, compared to 11% to 55% support growth for non-Tether
stablecoins. This is aligned with the rising value of transactions denominated in stablecoins.
Demographics of service providers’ customer base
An updated estimate of the number of cryptoasset users indicates a total of up to 101 million unique
users across 191 million accounts opened at service providers in Q3 2020. In 2018, the 2nd Global
Cryptoasset Benchmarking Study estimated the number of identity-veried cryptoasset users at about
35 million globally.
While rms continue to serve users from their region of operations, North American, Middle Eastern
and African companies appear to have a more geographically diversied clientele. Service providers
in both regions report that 42% of their customers are from other regions - primarily in Europe for MEA
rms and Latin America for North American ones.
Service providers operationally headquartered in North America and Europe indicate that business
and institutional clients make up 30% of their customers. This gure is much lower for APAC and Latin
American rms at 16% and 10% respectively.
The composition of business and institutional clientele differs from region to region. While North
American and European rms primarily serve cryptoasset hedge funds and traditional institutional
investors, Middle Eastern and African service providers that cater to non-retail clients focus on online
merchants (50%). Meanwhile, a notable share of APAC service providers deals with miners (41%), in part
explained by the high level of mining activities in the region, especially in China.
Regulatory and compliance standards across the industry and geographies
Just over two out of ve surveyed rms are licensed or in the process of obtaining a license; these
rms are primarily located in Europe. However, the remaining 58% should not be perceived as the share
of entities conducting unregulated activities or evading regulations: some surveyed service providers are
engaged in activities that do not yet warrant any authorisation process (e.g. non-custodial functions) or
are operating in jurisdiction(s) where no regulatory framework or guidance has been put forth.
3rd Global Cryptoasset Benchmarking Study
13
Compliance with KYC/AML obligations is heterogeneous across regions. Nearly all customer accounts
at European and North American service providers have been KYC’ed, whereas this is the case for only
one out of two accounts at MEA-based service providers.
The share of cryptoasset-only companies that did not conduct any KYC checks at all dropped from
48% to 13% between 2018 and 2020, most likely resulting from the progressive harmonisation of
KYC/AML standards across jurisdictions, such as initiated by the Financial Action Task Force (FATF).
The inclusion of rms exclusively supporting cryptoassets featured in FATF’s updated standards and
recommendations is believed to have spurred greater compliance among this group of rms. However,
this should not be interpreted as these companies becoming fully KYC compliant as some KYC checks
are only applied to a subset of consumers.
54% of surveyed custodial service providers indicated that they performed an externally-led audit of
their cryptoasset reserves over the past 12 months. This is a 24-percentage points decline compared to
our 2018 sample. Firms that have undergone an independent audit are most likely to be operating out of
Europe or the APAC region.
The development of best industry practices for IT security, security audits, and insurance coverage
plans
Regardless of their location or size, the vast majority of surveyed cryptoasset service providers keep
cryptoasset funds in cold storage (90%). To a lesser extent, they make use of multi-signature approaches
to secure their cold (81%) and hot (70%) storage systems.
Nevertheless, enhanced IT security measures do not automatically come alongside robust insurance
plans: 46% of service providers report not being insured against any risks. Those who do have
insurance plans are primarily insured against cybercrimes, professional errors, hazards, and loss or theft
of private keys.
The median non-custodial service provider usually spends a greater share of its resources, both
nancial and human, on IT security, between 11% to 20% compared to 6% to 10% for custodians.
This is partially explained by the fact that non-custodial systems are generally associated with greater
development costs and timeline.
Future outlook: striking the balance between integration and innovation?
A decoupling of duties, such as between custody, clearing and settlement responsibilities, appears to
be underway and may lead to greater resemblance with traditional nancial market infrastructure.
For instance, 45% of respondents indicate using a third-party, primarily crypto-native custodians, as part
of their cold storage system.
However, further intertwining with the traditional nancial system and greater institutional adoption
are conditional on enhanced compliance with international standards, such as those laid out by the
FATF. Survey data found that cryptoasset service providers legally incorporated in a jurisdiction member
of FATF are more likely to serve traditional institutional investors.
More risky and experimental innovations, such as in the realm of decentralised nance (“DeFi”), might
also come to fruition in the near future. Service providers, particularly large ones, expect that future
developments in the DeFi space will have considerable impact on their business operations and model in
the next 12 months.
3rd Global Cryptoasset Benchmarking Study
14
METHODOLOGY
For the third edition of the Global Cryptoasset Benchmarking Study, four market segments were
surveyed: (i) mining, (ii) payment, (iii) custody, and (iv) exchange. Two separate surveys were constructed
and distributed to respondents between March and May 2020, via secure web-based questionnaires.
1. The Cryptoasset Service Providers Survey was sent to entities active in one or more of the
payments, custody and exchange segments. The breakdown of respondents per market segment
is shown in Figure 1.
Figure 1: The exchange market segment is the most represented in the survey sample with more than two in three
surveyed respondents offering exchange services
2. The Cryptoasset Mining Survey was sent to individuals and organisations involved in the mining
industry, such as hashers, hardware manufacturers, and pool operators. The breakdown of mining
respondents is shown in Figure 2.
Figure 2: Nearly one in two respondents mine as part of a pool
Storage services
Exchange services
Respondents Breakdown by Market Segment
Share of service providers
Payment services 10%
29%
15%
17%
6%
18%
5%
Respondents Breakdown by Mining Activity
Share of mining actors
Mining as part of a pool
Mining hardware manufacturer and/or reseller
Remote hosting services (central administrator)
Solo mining
Cloud mining services (central administrator)
Mining pool operator (central administrator)
Other 14%
20%
21%
24%
32%
34%
44%
3rd Global Cryptoasset Benchmarking Study
15
The surveys were globally distributed to ensure a representative sample of geographic dispersion
across market segments. Both surveys were made available in English, Spanish, Portuguese, Chinese,
Japanese, Russian, and Korean. In addition, the Cryptoasset Mining Survey was translated into Arabic.
Respondents were able to choose their preferred language for the web-based questionnaire.
Over 500 invitations to complete the surveys were disseminated by email to known industry contacts
and to other participants whose email addresses were obtained through desktop research. Members
of relevant industry groups on different messaging platforms, such as Telegram and WeChat, were sent
invitation links to complete the surveys. Information and open invitations to complete the surveys were
posted on social media platforms, including Twitter, LinkedIn, Reddit, and BitcoinTalk. News outlets (e.g.
Coindesk, 8btc, ChainNews) assisted with distribution of the survey. Finally, the research team worked
and partnered with 26 national cryptoasset associations to ensure local and national distribution of the
surveys, thereby increasing wider global participation.
This study also saw the contribution of a third-party data provider: CryptoCompare provided the CCAF
with selected data underpinning their Annual Exchange Benchmarking Report to supplement data
collected via our own online surveys.2
All collected data was encrypted, safely stored and made accessible only to the CCAF research team
responsible for the production of this study. The privacy of all individual and company respondents was
ensured by anonymising all the data gathered from the surveys. In addition, the data was only analysed in
aggregate, using a range of categories that include industry segment, organisation size, supported assets,
custody types, and region.
Data was collected from 280 entities globally across 59 countries
In some instances, the survey data was supplemented by desktop research. This included web scraping
using manual techniques as well as Python scripts which were then veried and augmented through
a manual search process. Data from company websites, research reports, media outlets, and other
public sources was used to gather additional complementary data. If survey responses required
clarication, follow-up phone calls were made, or emails were sent to respondents. Where required, and
if feasible, additional checks were made by comparing survey results with other publicly available data
and responses on prior surveys. All responses were anonymised before the data was processed and
analysed.
280 entities across 59 countries and ve continents contributed to the surveys. 175 rms participated
in the Cryptoasset Service Providers Survey and 105 entities (75 organisations and 30 individuals)
completed the Cryptoasset Mining Survey.
Figure 3 provides a breakdown of survey participants by geographic region. European countries make up
more than a third of the Service Provider Survey sample. Compared to the previous study, respondents
from the Middle East and Africa (MEA) now make up 12% of global respondents, doubling the number
from last year. Latin America and the Caribbean (LAC) has also seen an increase from respondents, up
from 9% to 15% this year. This change in geographical distribution of respondents helps to provide a
more balanced dataset globally.
2 CryptoCompare Research (2020) Exchange Benchmark Report - July 2020. Available from: https://www.cryptocompare.com/
media/37072188/cryptocompare-exchange-benchmark-july-2020.pdf [Last accessed: 24 August 2020].
3rd Global Cryptoasset Benchmarking Study
16
Figure 3: APAC and European respondents constitute the majority of surveyed entities
Asia-Pacic (APAC) respondents still dominate the Mining Survey sample. The proportions of
respondents from Europe and North America remain approximately the same as 2018, with a 3%
increase from the MEA region and a 4% decrease from LAC. The three most represented countries in
our mining sample were China (24%), USA (16%), and Russia (8%).
The distribution of respondents in terms of age is relatively similar across both samples (Figure 4). Half
of surveyed rms have been in operation for up to 3 years. The other half of surveyed service providers
have been operating for between 3 and 10 years, compared to 3 to 8 years for the second half of
surveyed mining actors. Further, we observed a signicant difference in age distribution across regions.
In MEA, the majority of rms are young (less than 2 years old), whereas in Europe and North America
almost a fth of rms are almost as old as the industry itself (7+ years old).
Figure 4: For both samples, the median rm has been operating for three years
Geographic Breakdown of Respondents
Share of service providers
12%
12%
15%
38%
23%
Share of mining actors
5%
8%
24%
25%
38%
Asia-Pacic Europe Latin Ameri ca and the Caribbea n Middle Ea st and Africa North A merica
Age Distribution of Respondents
Service providersMining actors
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Outlier: 13 years
Outlier: 7 years
Outlier: 8 years
First quartile: 2 years
First quartile: 2 years
Median: 3 years
Median: 3 years
Third quartile: 6 years
Third quartile: 4 years
3rd Global Cryptoasset Benchmarking Study
17
SECTION 1: INDUSTRY GROWTH
INDICATORS
EMPLOYMENT FIGURES
The growth of the cryptoasset ecosystem has conventionally triggered a corresponding surge in job
opportunities. However, throughout 2018, employment growth decelerated as the effects of the late-
2017 market frenzy abated. The mining industry was particularly impacted by the slowdown: with a 37
percentage points (pp)3 decline in the segment aggregated employment level, compared to 36 pp for
service providers (Table 1).
The difference in the industry-level and the rm-level year-on-year (YoY) growth gures reects the rise
of large rms within each industry group that dominate in the aggregate. Larger companies appear to be
less affected than individual rms, which may suggest that a few large players dominate the industry.
3 “Percentage points (pp)” is the standard unit to express the difference between two percentages. For instance, at the aggregate
level year-on-year growth was 57% between 2017-2018 and 21% between 2018-2019: this is a decline of 57%-21%=36
percentage points (pp) decrease, but a 36/57 = 63% decrease in year-on-year growth .
4 Firm-level data is usually long-tailed in most industries (i.e. composed of many small rms and a few very large ones), which results
in signicant discrepancy between the mean and the median. From a rm-level perspective, using the median rather than the mean
is therefore more representative of the sample.
Table 1: Global patterns of employment levels at the industry- and rm-level
YoY 2017-2018 YoY 2018-2019
Industry aggregate Firm-level (median)4Industry aggregate Firm-level (median)
All market segments 57% 88% 21% 13%
Service providers 55% 100% 19% 0%
Mining 65% 60% 28% 0%
Cryptoasset industries in the different regions have not been impacted equally by this slowdown.
Employment growth in Europe decreased by 14 pp, from 32% between 2017-2018 to 18% between
2018-2019. This is roughly half as much as in Latin America (from 45% to 6%) and MEA (from 69%
to 35%). Figures from the APAC and North American cryptoasset industries reveal a sharp decline
in employment growth - from 73% to 21% for APAC and from 134% to 33% for North American,
amounting to a fall of 52 and 111 pp respectively (Table 2).
Firm-level gures point to other interesting patterns. In contrast to industry-wide gures, rm-level
data shows that individual rms in North America and Europe, alongside MEA companies, have been the
most impacted with a 110 and 55 pp negative change in employment growth. This difference between
the experience at the aggregate industry employment and the experience of the median rm is another
potential indication that large companies are growing ever larger as a share of the industry.
Table 2: Employment growth is uneven across region, both at the industry and rm levels
YoY 2017-2018 YoY 2018-2019
Industry-level Firm-level (median) Industry-level Firm-level (median)
APAC 73% 78% 21% 41%
Europe 32% 88% 18% 33%
LAC 45% 43% 6% 15%
MEA 69% 83% 35% 0%
North America 134% 100% 33% -10%
3rd Global Cryptoasset Benchmarking Study
18
Despite being one of the two most impacted regions by this decline, FTE gures for 2019 show that the
median APAC rm reports a larger workforce size, with a median of 40 FTE employees, than the median
companies from other regions. We also note greater variability in staff numbers for APAC companies,
which suggest that the regional industry is greatly diversied, with a mix of small and large entities.
5 Eurostat Glossary (2014) Glossary: High-growth enterprise. European Commission. Available from: https://ec.europa.eu/eurostat/
statistics-explained/index.php/Glossary:High-growth_enterprise [Last accessed: 20 August 2020].
6 Eurostat (2019) 1 in 10 enterprises in the EU classied as high-growth companies. European Commission. Available from: https://
ec.europa.eu/eurostat/web/products-eurostat-news/-/DDN-20191210-1 [Last accessed: 20 August 2020].
7 Based on the total of respondents that provided FTE gures for three consecutive years, from 2016 to 2019.
Figure 5: Larger rms tend to operate from APAC and North America
HIGH-GROWTH ENTERPRISES
The share of high-growth companies in an industry is often used as an indicator to assess the
development stage of the sector. A high-growth enterprise may be determined by its workforce size or
prots generated. From an employee gure perspective, an enterprise qualies as “high-growth” if the
average annualised growth in number of full-time employees (FTE) is greater than 10% p.a. over a three-
year period and having at least 10 employees at the beginning of the growth.5
According to our study sample and using established criteria, high-growth companies accounted for
more than one out of every four enterprises active in the cryptoasset ecosystem in 2019. This gure is
slightly above the share of high-growth rms in other industries. By comparison, in 2017, the European
Commission reported that high-growth rms represented respectively 17% and 13% of European
companies in the Information and Communication, and Professional, Scientic, and Technical Activities
sectors.6
The median YoY employment growth rate for high-growth
rms was 53% for the period 2017-2018 and 43% for 2018-2019
The share of high-growth rms as a proportion of the number of surveyed rms that provided FTE
gures for three consecutive years, is almost equal across the service providers and mining actors with
27% of surveyed service providers qualifying as high-growth rms compared to 25% of surveyed mining
actors.7 From a geographic standpoint, the majority of these rms are in APAC, where 39% of surveyed
Number of Full-time Equivalent (FTE) Employees
Capped to 210
200
180
160
140
120
100
80
60
40
20
0
Asia-Pacic Europe North America
Latin America and
the Caribbean
Middle East
and Africa
Number of FTE
3rd Global Cryptoasset Benchmarking Study
19
enterprises active since at least 2017 can be dened as high-growth rms (Figure 6). The lowest share
of high-growth enterprises was found in LAC, where only 13% of surveyed rms in that region met the
criteria.
Figure 6: In 2019, APAC recorded the highest share of high-growth enterprises
Higher growth rms account for a larger part of employment in the industry relative to small growth
rms. In 2019, more than two in three employees of surveyed enterprises that have been active since at
least 2017 were employed by a high-growth company. These rms, on average, more than doubled their
workforce over the three-year period. In 2017-2019, high-growth cryptoasset rms experienced on
average a positive growth in the number of employees from approximately 84 employees in 2017 to 200
employees in 2019. This minority of high growth rms appear to account for higher employment levels
across the industry.
A commonly shared view is that high-growth rms are typically young (but at least three years old).
Although survey data shows that 3-4 years-old rms represent 49% of high-growth enterprises, the
median high-growth rm in the cryptoasset industry is 6 years old (Figure 7).
Figure 7: High growth is primarily a young rm phenomenon
FINANCIAL PERFORMANCE OF SERVICE PROVIDERS
To perform a longitudinal assessment of the nancial performance of service providers over the years,
we collected data on the operating revenues and pre-tax prots. Majority of the surveyed entities that
have been active since 2017, indicate that they have generated operating revenues over the past three
years (Figure 8).
Share of surveyed enterprises active since at least 2017
High-Growth Enterprises by Region (2019)
World = 26%
Asia-Pacic
39%
Europe
27%
North America
23%
Middle East
and Africa
18%
Latin America and
the Caribbean
13%
Age Distribution of High-Growth Firms
Share of high-growth rms
49% 30% 19%
3 to 4 years old 5 to 6 years old 7 to 8 years old 9+ years old
2%
3rd Global Cryptoasset Benchmarking Study
20
Despite the growth in the number of rms reporting operating revenues between 2017 and 2019,
from 10% to 16%, the number of rms that realised pre-tax prot stagnated between 2017 and 2018.
2018 corresponds to the year when the market experienced a sharp drop in prices and total market
capitalisation, which may have had a debilitating effect on some service providers and their ability to
generate prots.
Figure 8: Surveyed service providers report increased protability in 2019 in comparison to the preceding years
Unsurprisingly, the older the company, the more likely it is to be protable. 80% of rms aged 7 years old
or older report having earned prots in 2019, compared to 60% for the 3-4 years old age group and 64%
for rms that are 5-6 years old.
Operating Revenues and Pre-Tax Prot
Share of surveyed service providers active since at least 2017
2018 2019
80%
89%
2017
75%
59% 57%
66%
Operating rev enues Pre-tax prot
3rd Global Cryptoasset Benchmarking Study
21
SECTION 2: MINING, THE BACKBONE
OF THE INDUSTRY
With mining operations reaching industrial scale, the ecosystem has morphed into a complex network
of interdependent actors,8 often opaque and hard to understand for external observers and the general
public. In particular, the crucial role of miners in the functioning and the security of proof-of-work (PoW)
systems, such as Bitcoin, is often underappreciated and overlooked. This section intends to address this
complexity by delving into the role of hashers, pool operators, and hardware manufacturers.
HASHING AS A BUSINESS
Hashers’ raison d’être is the existence of PoW, a consensus mechanism to produce a commonly-agreed
history of transactions without relying on a central coordinating authority.9 There are other consensus
mechanisms that exist, such as proof-of-stake, but given PoW’s predominance at the time of writing this
report, this section exclusively focuses on PoW mining.
COIN SELECTION
The entry cost into cryptocurrency mining has been on the rise since 2013, partly attributable to
increasing computational difculties that necessitate the utilisation of specic-purpose hardware.
Figure 9: Financial parameters primarily guide hashers’ choice of coins to mine
8 For an introduction to the mining industry please see Rauchs et al. (2018) 2nd Global Cryptoasset Benchmarking Study. Cambridge
Centre for Alternative Finance. Available from: https://www.jbs.cam.ac.uk/wp-content/uploads/2020/08/2019-09-ccaf-2nd-
global-cryptoasset-benchmarking.pdf [Last accessed: 24 August 2020].
9 In proof-of-work cryptocurrencies, “hashers” provide computing power and are commonly known as “miners”. For further
discussion on the role of hashers, see Rauchs et al. (2018) 2nd Global Cryptoasset Benchmarking Study. Cambridge Centre for
Alternative Finance. Available from: https://www.jbs.cam.ac.uk/wp-content/uploads/2020/08/2019-09-ccaf-2nd-global-
cryptoasset-benchmarking.pdf [Last accessed: 24 August 2020].
Coins Selection Criteria
Share of hashers
Daily reward amount
Price of cryptoasset
Ideology/personal affection
Market capitalisation
Large number of miners/mining pools
Energy requirement
Friends/colleagues recommendation
Proof system
Low number of other miners/mining pools
57%
30%
47%
15%
48%
4%
47%
8%
30%
4%
47%
11%
52%
30%
51%
26%
70%
13%
51%
32%
70% 68%
Reputation
Other
Large Small
3rd Global Cryptoasset Benchmarking Study
22
Hashers must therefore carefully select the coin(s) to mine against a set od criteria to break even. For the
majority of hashers that are driven by prot motives and returns, coin selection is generally guided by
nancial criteria, such as daily reward amount or cryptoasset prices (Figure 9).
Conversely, the remaining portion of hobbyist hashers, believed to be mostly located in Europe and
North America,10 are likely to be driven by more subjective criteria, such as ideology and personal
affection.
Interestingly, energy requirement seems to be a much more determining factor for small hashers than
large ones. This difference is also reected at the regional level: hashers operating out of Europe (56%)
and LAC (63%) more often base their coin choice on this criteria than those from APAC (37%) or North
America (35%). This might be an indication that hashers, and particularly large ones, from APAC and
North America are more condent in their ability to secure stable access to energy sources.
Bitcoin is the most popular coin mined, with 89% of surveyed hashers indicating that they mine it,
followed more distantly by Ethereum (35%) and Bitcoin Cash (30%). Interestingly, while Bitcoin mining
is predominant across all regions, other coins seem more popular in certain areas than others (Figure
10). For instance, Ethereum mining appears to be particularly popular among Latin American hashers,
whereas Bitcoin Cash is more popular in APAC and North America. The mining of privacy coins in
Western regions also differs from the global average: 28% and 19% of European and North American
hashers report mining ZCash, and as many North American hashers also engaged in Monero mining.
10 Genesis Mining (2020) State of Crypto Mining 2020. Genesis Mining. Available from: https://www.genesis-mining.com/state-of-
mining2020?download=conrm [Last accessed: 20 August 2020].
Figure 10: Beside global consensus on Bitcoin, the popularity of other PoW coins varies across world regions
COST OF HASHING
Hashers’ costs comprise capital expenditures (e.g. purchase of mining equipment, infrastructure
development and allied costs), which represent on average 45% of hashers’ total costs. The remaining
55% fund operational expenditures (e.g. electricity bills, maintenance, workforce), of which 75% is utilised
towards the payment of utilities. This gure varies slightly based on the type of coins mined. For instance,
for hashers exclusively focusing on cryptocurrencies that employ the SHA-256 mining algorithm,
Bitcoin (BTC)
Ethereum (ETH)
Bitcoin Cash (BCH)
Ethereum Classic (ETC) 16% 6% 14%
34% 22% 25% 38%
34% 22% 63% 33%
94% 100% 88% 86%
Asia-Pacic Europe Latin America
and the Caribbean North America
Coins Mined by Region
Share of hashers
25% 33% 25% 24%Litecoin (LTC)
Monero (XMR( 9% 6% 19%
Digital Cash (DASH) 13% 6% 13% 10%
16% 28% 25% 14%Other
ZCash (ZEC) 16% 28% 19%
3rd Global Cryptoasset Benchmarking Study
23
utility costs correspond to 79% of their operational expenditures. This is in part explained by diverging
production costs of the different PoW coins.
Utility costs represent on average 79% of SHA-256 hashers’ operational expenditures
Differences also arise at the regional level (Figure 11): hashers operating in LAC reported the lowest
share of utilities cost as part of their total cost structure. The clustering of respondents at the bottom
part of the distribution for North and Latin American hashers suggests that a select few hashers in these
regions are able to drastically minimise their utilities costs.
Importantly, Latin American hashers registered the highest share of capital expenditures - although
we note a relative variability across them - possibly explained by a lack of robust supply chains to ship
equipment to the region. In contrast, easy reach of hardware manufacturers is reected by a lower share
of capital expenditures as part of their total costs for APAC miners (37%).
11 For instance, since the introduction of new tariffs on Chinese imports, US hashers have to pay 28% tariffs on ASICs shipped to the
USA.
Figure 11: Most of North and Latin American hashers’ costs go to capital equipment
A focus on the two most represented countries in our study sample offers additional insights. Cost
structure data from Chinese and American hashers seems to conrm that Chinese hashers have a
competitive edge in the acquisition of mining machines (Figure 12). The concentration of hardware
manufacturers in China implies well-connected and shorter supply chains to Chinese hashers, simplied
business conduct (e.g. language, working culture), and absence of additional overseas shipping fees.11
Unexpectedly, the share of labour and maintenance costs do not seem to differ signicantly between
the two mining regions, despite common belief that cheaper workforce in China would necessarily lead
to labour contributing less to their overall expenditures. Cost of labour is indeed cheaper in China, but
Chinese hashers tend to rely on a larger workforce size to run their operations. In contrast, most North
American facilities have deployed sophisticated ASIC management software that reduces the need for
technicians.
Figure 12: Chinese hashers allocate more than half of their total expenditures to utilities
37%
43%
52%
52%
49%
41%
38%
21%
7%
9%
5% 5%
6% 5%
1%
7%
6%
16%
Asia-Pacic
Europe
Latin America and the Caribbean
North America
Cost Breakdown of Hashers per Region
Share of hashers’ cost structure
Capital equipment Utilities Maintenance Employees or contractors Other
31%
56%
55%
31%
7%
5%
1%
1%
6%
7%
China
United States
Cost Breakdown of Chinese and American Hashers
Share of hashers’ cost structure
Capital equipment Utilities Maintenance Employees or contractors Other
3rd Global Cryptoasset Benchmarking Study
24
ELECTRICITY PRICE
Utilities, primarily composed of electricity costs, take the lion’s share of hashers’ operational
expenditures. Contrary to the popular assumptions found in academic papers and mainstream media12,
the vast majority of hashers no longer pay residential electricity prices, but often access preferential/
industrial pricing by entering contractual agreements with power generators. The median electricity
price is comparatively higher in North America - albeit signicantly variable across hashers from the
region - and APAC at USD 0.05/kWh, whereas Latin American hashers report the lowest median
electricity price (USD 0.025/kWh) of all regions (Figure 13).
Globally, electricity price paid by miners averages USD 0.046 per kWh13
There is a notion that electricity surplus in some APAC areas, such as the province of Sichuan in China,
gives hashers who relocate their operations there during the rainy season a competitive advantage in
minimising their running costs. However, survey data demonstrates that this seasonal advantage appears
to be offset by less affordable electricity prices throughout the rest of the year when hashers migrate
back to other provinces, such as Xinjiang or Inner Mongolia in China.
12 See for instance, Malfuzi, A. et al. (2020) Economic viability of bitcoin mining using a renewable-based SOFC power system to supply the
electrical power demand. Energy. Available from: https://doi.org/10.1016/j.energy.2020.117843 [Last Accessed: 20 August 2020];
Benetton, B., Compiani, G. and Morse, A. (2019) Crypto Mining: Local Evidence from China and the US. University of Berkeley.
Available from: https://faculty.haas.berkeley.edu/morse/research/papers/BenettonCompianiMorse_CryptoMining.pdf
https://doi.org/10.1016/j.energy.2020.117843 [Last accessed: 21 August 2020].
13 The weighted average was calculated by combining survey data on electricity price and the estimated monthly share of total Bitcoin
hashrate per country for the period September 2019 to April 2020, according to the CBECI mining map. See Cambridge Bitcoin
Electricity Consumption Index. Cambridge Centre for Alternative Finance. Available from: https://cbeci.org/mining_map
[Last accessed: 21 August 2020].
Figure 13: The median Asian and North American hasher pay the same electricity price
GOVERNMENT SUPPORT
Hashers’ operational costs may be reduced through government support, which may take the form
of subsidies or tax exemptions. Government interventions through subsidies and tax exemptions are,
however, still relatively uncommon in most regions. Only 23% of the surveyed hashers report receiving
support from governments (Figure 14). This aid primarily takes the form of locally-focused support, such
as electricity price subsidy for users within a region. 38% of surveyed hashers who receive government
support operate in China, followed by Kazakh (19%) and Canadian (12%) hashers.
Electricity Price per Region
9
8
7
6
5
4
3
2
1
0
North America
Latin America and
the Caribbean
Europe
USD cent/kWh
Asia-Pacic
3rd Global Cryptoasset Benchmarking Study
25
Figure 14: A select few hashers benet from governmental subsidies
14 For an overview of regulatory developments in 2020, see TokenInsight (2020) 2020 Q2 Cryptocurrency Mining Industry Report.
Available from: https://tokeninsight.com/report/1189?lang=en&title=2020-Q2-Cryptocurrency-Mining-Industry-Report [Last
accessed: 21 August 2020].
15 ‘Texas State Securities Board vs Ultra BTC Mining LLC’ (2020) Emergency Cease and Desist Order. Texas State Securities Board.
Available from: https://www.ssb.texas.gov/sites/default/les/ENF_20_CDO_1801.pdf [Last accessed: 21 August 2020].
Mining activities have attracted greater regulatory scrutiny
CCAF’s regulatory landscape of cryptoasset activities study released in April 2019
revealed that very few jurisdictions have included or explicitly mentioned mining in their
regulatory guidance on cryptoasset activities. Mining has, however, sparked greater
regulatory attention recently, and in some instances led to the development of bespoke
legal frameworks for mining activities, such as in Kazakhstan.14 In other jurisdictions,
enforcement actions have provided further clarity on the regulatory treatment of mining-
related products. For instance, in a ruling dated April 2020, the Texan regulator indicated
that a cloud mining platform breached US securities law by offering unlicensed securities.15
Miners from the same region greatly diverge in their opinion on their immediate
regulatory environment. This heterogeneity in opinion suggests either that miners have
limited awareness of existing regulation, or that regulation is confusing and inconsistent.
Nonetheless, regulatory changes seem unlikely to induce a change in the geographic
location of operations as reported by surveyed miners; only 23% indicated having opened a
new mining facility following a change in local regulation.
75% Locality-focused support
31% Business-focused support
19% Unique agreement
28%
Benet from
governmental
subsidies
Government Support to Hashers
Share of hashers
3rd Global Cryptoasset Benchmarking Study
26
WHERE ARE WE ON POW’S ENERGY CONSUMPTION?
Despite increasing transparency and research on the environmental impact of PoW mining,16 the topic is
still typically misrepresented in most sources and on both sides of the debate. Similar to 2018, this year’s
survey data shows that a signicant majority of hashers (76%) use renewable energies as part of their
energy mix (Figure 15). However, the share of renewables in hashers’ total energy consumption remains
at 39%.
Hydropower is listed as the number one source of energy, with 62% of surveyed hashers indicating that
their mining operations are powered by hydroelectric energy (Figure 16). Other types of clean energies
(e.g. wind and solar) rank further down, behind coal and natural gas, which respectively account for 38%
and 36% of respondents’ power sources.
39% of miners’ total energy consumption comes from renewables
The data does not allow us to infer what share of natural gas usage corresponds to stranded gas, i.e.
represents energy that would otherwise be wasted or unused. Stranded gas either takes the form of gas
that would be ared at oil or gas wells due to limited pipeline capacity or gas coming from non-exploited
wells due to logistical or economic reasons. Despite reported challenging logistics, certain areas in the
USA have witnessed the installation of a few mining sites powered by stranded gas, such as in Texas or
North Dakota.
16 Stoll C., Klaaßen U. and Gallersdörfer, U. (2019) The Carbon Footprint of Bitcoin. Joule. Available from: https://www.cell.com/joule/
pdf/S2542-4351(19)30255-7.pdf [Last accessed: 21 August 2020], Bendikson, C. and Gibbons, S. (2019) The Bitcoin Mining
Network - Trends, Composition, Average Creation Cost, Electricity Consumption & Sources. CoinShares Research. Available from: https://
coinshares.com/assets/resources/Research/bitcoin-mining-network-december-2019.pdf [Last accessed: 21 August 2020], and
Cambridge Bitcoin Electricity Consumption Index. Cambridge Centre for Alternative Finance. Available from: https://cbeci.org/mining_
map [Last accessed: 21 August 2020].
Figure 15: PoW mining is primarily powered by non-renewable energy sources
The spread shape of regional distributions for the share of renewables suggests that there is extreme
variability across miners from the same region, particularly in APAC, Europe, and North America. The
median percentage of renewables in Europe and North America is relatively high at about 70% and 66%
respectively, while the median is much lower in APAC, at 25%.
Renewables as Part of Hashing Energy Mix
Share of hashers Share of total energy consumption
76%
Use renewable
energies as part
of their mix
39%
Of hashing’s
total energy
consumption comes
from renewables
3rd Global Cryptoasset Benchmarking Study
27
Figure 16: Hydroelectricity, coal, and natural gas are the most reported power sources
17 Hydropower accounts for approximately a quarter of the total power capacity in China. See, Youmei (2020) Hydropower and
Sustainable Development in China. Department of Economic and Social Affairs. Available from: https://www.un.org/esa/sustdev/
sdissues/energy/op/hydro_luyoumei.pdf [Last accessed: 21 August 2020].
18 Consumption of hydropower for mining operations has sometimes been encouraged by Chinese government ofcials,
including most recently by the Municipal Economic and Information Bureau and the Municipal Development and Reform
Commission of Ya’an District. See, Zamundzinski, A. (2020) Chinese Ofcials Support Renewable Energy-Powered Cryptocurrency
Mining. Cointelegraph. Available from: https://cointelegraph.com/news/chinese-ofcials-support-renewable-energy-powered-
cryptocurrency-mining [Last accessed: 21 August 2020].
19 Of note, most North American hashers connect their operations to the grid, which naturally combine a mix of power sources.
Regional breakdown of energy
sources reveals that APAC hashers
equally rely on coal and hydropower
(Figure 17). Coal-based mining is
principally adopted in regions such as
the Chinese provinces of Xinjiang and
Inner Mongolia, and in Kazakhstan,
whereas hydroelectric energy is
mainly generated in South-Western
regions of China (Sichuan and
Yunnan).
China’s oversupply of hydroelectric
energy during the rainy season
has often been used as evidence in
claims that a vast majority of mining
is powered by environment-friendly
power sources. While it is true that
the Chinese government’s strategy to
ensure energy self-sufciency has led to the development of massive hydropower capacity,17 18 the same
strategy has driven public investments in the construction of large-scale coal mines. Like hydroelectric
power plants, these coal power plants often generate surpluses. It should not come as a surprise then
that a signicant share of hashers in the region equally report using both hydropower and coal energy to
power their operations.
Figure 17: North American hashers appear to use a wider range of energy sources19
Power Sources of Hashing Facilities
Share of hashers
Hydroelectric
Coal
Nuclear
Natural gas
Geothermal
Wind
Other
Oil
15%
15%
17%
2%
8%
12%
Solar
36%
38%
62%
12% 13% 17% 17%
12% 33%7% 22%
23% 7% 22%
38% 33% 17% 44%
65% 20% 28%
65% 60% 67% 61%
Asia-Pacic Europe Latin America
and the Caribbean North America
Energy Sources per Region
Share of hashers
6%
8% 17%
12% 7% 22%
Hydroelectric
Coal
Nuclear
Natural gas
Geothermal
Wind
Other
Oil
Solar
3rd Global Cryptoasset Benchmarking Study
28
In July 2019, CCAF launched a real-time estimate of Bitcoin’s electricity consumption, the Cambridge
Bitcoin Electricity Consumption Index (CBECI), which was followed by the release in May 2020 of an
interactive map of the geographic distribution of Bitcoin hashpower from September 2019 to April
2020.20 The ndings from the map combined with survey data for SHA-256 hashers offer an alternative
methodological approach to estimate the energy mix of Bitcoin mining. The results of this top-down
calculation are displayed in Table 3 and indicates that about 29% of Bitcoin mining is powered by
renewables.
20 Data underpinning the map was provided by three pools, BTC.com, Poolin, and ViaBTC, that together represent 37% of Bitcoin’s
total hashrate.
21 Mining rewards comprise both the block subsidy and transaction fees. While in a PPS model, pools only redistribute the block
subsidy, the FPPS and PPS+ schemes also include the distribution of transaction fees.
Table 3: Aggregate share of renewables in Bitcoin mining energy sources
Region Regional average
share of renewables
Regional share of
Bitcoin hashpower
Regional weighted share of
renewables in Bitcoin mining
Asia-Pacic 26% 77% 20%
Europe 30% 10% 3%
Latin America and the Caribbean 20% 1% 0%
Middle East and Africa NA 4% NA
North America 63% 8% 5%
Global 100% 29%
Source: CBECI mining m ap, survey data. As of April 2020
MINING POOLS
PAYMENT METHODS
Pool operators coordinate the work of thousands of hashers to increase the likelihood of producing a
valid PoW. They are also responsible for compensating hashers based on the expected value of their
contribution and according to a chosen payment method (see call-out box). There are more than a dozen
reward systems for pools to choose from. However, one seems to have prevailed to date: the pay-per-
share (PPS) model and its associated subalterns full-pay-per-share (FPPS) and pay-per-share + (PPS+).21
The study of a separate data set focusing on the top-15 Bitcoin mining pools and their respective share
of total hashrate shows that the full-pay-per-share (FPPS) model dominates (Figure 18). This distribution
may vary for other coins, however, in particular those whose transaction fees are insignicant for miners
payouts.
Figure 18: FPPS is the most widely supported across Bitcoin top mining pools
Top-15 Bitcoin Mining Pools' Earning Models
Weighted share
73%
4%
23%
FPPS PPS+ PPLNS Source: BTC .com, Pools’
individual website
3rd Global Cryptoasset Benchmarking Study
29
The reward systems of mining pools
A reward system corresponds to the payment method used by pool operators to split
mining rewards among hashers contributing to their pool. One key difference between
existing reward systems lies in the valuation of hashers’ work. In the PPS model, a hasher is
immediately rewarded upon completion of work, even though the pool is not guaranteed to
nd a block. The amount paid is calculated based on the expected value of a hasher’s work.
As a result, PPS payoffs are deterministic for mining pool participants as the value of the
payoff is known in advance.
Conversely, in the pay-per-last-n-share (PPLNS) system, mining pools dene a time window
and pay out rewards to miners only after the pool has found a block. The actual value of the
payoff is based on the share of work produced by the hasher during this time window.
Each model comes with its own set of trade-offs; while the regularity of PPS payments
reduces the variance or “luck factor” on miners’ side, the revenue miners can expect
from a PPS pool is slightly lower, in the short term, than in a PPLNS model. In the long
run, however, a miner is expected to earn a higher revenue with the PPS model, all else
equal. This is because in a PPS setup, pools will continue to pay even under unfavourable
circumstances (e.g. orphan blocks, blockwitholding attacks). Conversely, in the PPLNS
mode, miners bear the “luck” risk, but usually pay a lower fee to the pool and might receive
higher earnings, in the short term, depending on how lucky the pool is in nding blocks.
Most reward systems are constructed to prevent “pool-hopping”, whereby hashers regularly switch
between pools as their protability changes. Hopping-proof reward systems (e.g. PPLNS, score-based)
disincentivise hashers from doing so by offering better rewards to “loyal hashers”. A growing body of
academic literature has emerged to study pool-hopping behaviours and hashers’ migration patterns
between pools.22 23 For instance, Belotti et al.’s (2018) analysis has shown that although pool-hopping
might be more protable, the practice is not necessarily widespread among hashers.
Pools are also developing novel techniques to better reward hashers and win their loyalty. For instance,
at the time of writing, pools are beginning to offer prot-switching algorithms between PoW coins using
the same hashing algorithm (e.g. Bitcoin, Bitcoin Cash, Bitcoin SV, Bitcoin Diamond, all use the SHA-
256 algorithm). Instead of performing work for a single coin, this service lets the pool direct hashers’
aggregate hashpower to the most protable coin of those that use the same hashing algorithm. In turn,
pools are expected to buy hashrate at a signicantly higher price from miners.24
This is also interesting in light of updated survey data on the distribution of miners’ hashpower
contribution to pools, which, per 2019, suggests that hashpower contributed by the most active miners
follows a power-law distribution.25 Figure 19 shows that the top-1% of active miners are responsible
for two-thirds of the pool’s total hashpower at the median. If pool-hopping was common practice, pools’
overreliance on a small number of customers would pose risk to their operations. Furthermore, the
change in the median contribution of hashers across all three groups between 2018 and 2020 data
suggests greater concentration at the top.
22 Belotti M., Kirati, S. and Secci, S. (2018) Bitcoin Pool-Hopping Detection. Proc. of 2018 IEEE 4th International Forum on Research and
Technology for Society and Industry. Available from: https://www-phare.lip6.fr/~secci/papers/BeKiSe-RTSI18.pdf [Last accessed:
21 August 2020].
23 Liu, K. and Ohsawa, Y. (2019) Auction based Rewards Distribution Method in Pool Mining. Proc. of 2019 IECC International Electronics
Communication Conference. Available from: https://dl.acm.org/doi/pdf/10.1145/3343147.3343162 [Last accessed: 21 August
2020].
24 For a detailed explanation of prot-switching see for instance, Luxor Tech (2020) Introducing Luxor Switch. Available from: https://
medium.com/luxor/introducing-luxor-switch-6c65401d3d71 [Last accessed: 21 August 2020].
25 Active participation is dened as contributing hashpower at least once a week to the pool. The median share of active members is
80%, with noticeable variability across pools as a signicant number of them report lower activity gures.
3rd Global Cryptoasset Benchmarking Study
30
Figure 19: The median hasher in the top-10% of active contributors produces 84% of the pool’s total hashrate
26 See page 95, Rauchs et al. (2018) 2nd Global Cryptoasset Benchmarking Study. Cambridge Centre for Alternative Finance. Available
from: https://www.jbs.cam.ac.uk/wp-content/uploads/2020/08/2019-09-ccaf-2nd-global-cryptoasset-benchmarking.pdf [Last
accessed: 24 August 2020].
27 In practice however, it is commonly believed that pools’ heavy involvement in the mining industry and entire dependence on miners
willing to connect to their pool may have disincentivised more than one to behave dishonestly.
A NOTE ON MINING CENTRALISATION AND POOL GOVERNANCE
As discussed in the 2nd Global Cryptoasset Benchmarking Study, centralisation risks are mainly
perceived to occur at three levels of the mining value chain: geographic location (and/or ownership) of
hashpower, hardware production, and pool operation.26 Specically, pool operations have been of great
concern given the censorship power that they have had to date over the work performed by hashers.
Given this concern, it is important to understand pools’ governance model and their role in the mining
process.
From a governance standpoint, survey data reveals that no clear-cut model stands out: one third of
pools use a mix of approaches in their decision-making process, while another third acknowledged that
decision-making is a prerogative of pool administrators.
In their coordination role, mining pools retain great leverage over the work done by miners and leave
them with relatively limited bargaining power.27 If incentivised to, pools could choose to exploit their
inuence in multiple ways, e.g. to dishonestly mine, blacklist transactions or addresses, or redirect
miners’ hashpower to support another chain. Similar scenarios could also materialise if a pool was to be
attacked and controlled by malicious actors. In response to this centralisation risk, different solutions
have been developed to ensure greater decentralisation in the mining process at the pool level (see
callout box on stratum v2).
Top Miners Contribution to Pool Hashrate
Share of total pool hashrate
Top-5% Top-10%
74%
84%
Top-1%
65%
20%
55%
70%
3rd Global Cryptoasset Benchmarking Study
31
Stratum v2: handling control back to hashers?
Hashers participating in pooled mining rely on a protocol stack to communicate with pool
services, contribute work, and receive rewards. Since the creation of the rst Bitcoin
mining pool in 2011, the dominant mining protocol in use has been stratum. Concerns
raised by multiple developers over centralisation risk and pools’ censorship power spurred
the research and development of an alternative mining protocol. Several actors have
laid the groundwork in putting forth several proposals. One such example is stratum v2.
Simply put, stratum v2 introduces an extra-step in the pooled mining process consisting
in a negotiation phase between hashers and pools. During the negotiations, hashers have
the possibility to choose what they will work on (i.e. transaction set) instead of letting the
pool unilaterally decide on their behalf. Ultimately this decouples the block building and
propagation to the network phases from payouts to miners.
For this approach to take off, alternatives to the original stratum protocol must be widely
supported and implemented by hashers and pools. From our survey data, a large majority
of surveyed pools and hashers report being undecided regarding the implementation of
Stratum v2. A fourth have reported planning to implement stratum v2, but they have yet to
follow through (Figure 20).
Figure 20: In APAC, one-fourth of hashers and pools are unsure about stratum v2 implementation
Implementation of Stratum V2
Share of hashers and pool operators
Asia-Pacic
24%
Undecided
North
America
18%
Planned
27%
Latin America and the Caribbean
6%
Other
5%
Unplanned
4%
Europe
16%
64%
Are undecided
about Stratum v2
implementation
3rd Global Cryptoasset Benchmarking Study
32
MINING HARDWARE MANUFACTURING
The mining hardware industry is another constituent of the mining industry that is often perceived as a
complete black box. Its relatively concentrated and secretive nature may conceal power dynamics and
relationships at play. Firstly, it is important to emphasize that manufacturing industries greatly differ
based on the type of hardware in use and the coin mined. For instance, the hardware space for Bitcoin
mining (and other SHA-256 coins) is dominated by ASIC producers, whereas Ethereum mining has had
a long-lasting loyalty to GPU-mining. Undeniably, the former has received most press coverage to date.
This subsection primarily focuses on the ASIC manufacturing industry.
The introduction of ASICs in 2013 hastened the professionalisation of the cryptoasset mining ecosystem
as a whole. Over the years, ASIC manufacturers have been acknowledged for their role in considerably
improving the efciency of equipment and lengthening the obsolescence period of hardware.28
However, the ASIC manufacturing industry is one that is still trying to nd its feet. ASIC manufacturers
are particularly dependent on their partnership with third-party foundries that supply them with
advanced integrated circuits technology necessary to build ASIC chips. Given the concentration of the
foundry market, ASIC manufacturers’ operations are heavily reliant on a few fabrication plants, and a rise
in price, a reduction in the foundry capacity allocation, or simply a deterioration of relationships could be
detrimental to manufacturers’ business.
In this context, it is interesting to examine how these challenges may have impacted the market
distribution of ASIC manufacturers’ equipment. There is, however, a severe shortage of reliable
information on the topic. In an attempt to gauge the scale of the ASIC primary market, several reports
based on publicly disclosed sales information attributed the majority of the market share to a single
manufacturer until 2018.29 After 2018, these reports found that ercer competition from other
manufacturers has led to a shift in dominance and to a more diversied landscape.
28 See for instance, Elmandjra, Y. and Hsue, D. (2019) Bitcoin Mining - The Evolution of A Multibillion Dollar Industry. Ark Invest. Available
from: https://research.ark-invest.com/bitcoin-mining-white-paper [Last accessed: 21 August 2020].
29 These reports estimate the market presence of each ASIC manufacturer, either based on the number of machines sold or the
amount terahash per second (Th/s) sold. See for instance, BitMEX Research (2020) Battle For ASIC Supremacy. BitMEX. Available
from: https://blog.bitmex.com/battle-for-asic-supremacy/ [Last accessed: 21 August 2020].
Figure 21: The vast majority of SHA-256 hashers report using Bitmain’s Antminers
Market Distribution of SHA-256 Equipment
Share of SHA-256 hashers
3rd Global Cryptoasset Benchmarking Study
33
Survey data partially corroborates these ndings. For instance, data collected from SHA-256 hashers
reveals that, as of April 2020, Bitmain Antminer machines appear to dominate the market (Figure
21). In particular, Bitmain S9 model, which was released in May 2016, was reportedly used by 32% of
surveyed SHA-256 hashers.30 Interestingly, alternative techniques using network data to estimate the
amount of hashpower provided by certain types of hardware suggested that Antminer S9 machines were
responsible for 32% of Bitcoin’s hashpower.31 The next most cited manufacturer was MicroBT, whose
Whatsminer M20S model appears to be particularly popular among hashers (16%).
Figure 22 shows that China accounts for a substantial portion of manufacturers’ total sales (52%),
dwarng other world regions, including the USA (12%) and Canada (9%). This is consistent with the fact
that China is the main hub of mining activities as revealed by pools’ data displayed on the CBECI. The
CBECI mining map also shows that Kazakhstan and Russia occupy a notable share of mining activities
albeit, they each account for only 4% of manufacturers’ total sales. Possibly, hashers from these countries
may primarily be supplied by equipment sold on the secondary market.
30 It should be noted however that data was collected prior to the Bitcoin halving of May 2020. Hence, the market share of each model
may have signicantly changed since then, as some machines became unprotable and obsolete.
31 Helmy, K. (2020) The Half-Time Show: The State of Bitcoin Network Security After the Halving. Coin Metrics. Available from: https://
coinmetrics.substack.com/p/coin-metrics-state-of-the-network-fcf [Last accessed: 21 August 2020].
32 Hashrate Index (2020) SHA-256 Rig Index. Hashrate Index. Available from: https://www.hashrateindex.com/machines/sha256-rig-
index [Last accessed: 21 August 2020].
Figure 22: In 2019, one in two ASICs produced is distributed to Chinese hashers
Accurate gures on the ASIC secondary market are even scarcer than information on the primary
market. Suspected to be particularly dynamic, the scale of the secondary market is hard to come by as
most trades take place over-the-counter often via informal channels (e.g. Telegram groups). Existing
data sets on deals happening on the secondary market suggest that the market share of each SHA-256
hardware manufacturer is relatively similar to their market share of primary sales.32
Figure 23: Secondary market trades may have help resellers sustain their sales volumes
Source: CCA F survey data, Ebang IPO lings
52% 19% 12 % 9% 4%4%
Sales Destination of ASIC Manufacturers (2019)
Share of total sales (weighted average)
China Rest of the world USA Canada Russia Kazakhstan
2018-2019
2017-2018
-26%
Sales Volumes of ASIC Manufacturers and Resellers
Year-on-year growth
172%
162%
148%
Manufac turers Resellers
3rd Global Cryptoasset Benchmarking Study
34
Data obtained from surveyed manufacturers and resellers (including the sale of second-hand machines)
reveals that their volumes have moved in opposite directions after 2018 (Figure 23): while both grew
signicantly between 2017 and 2018 - at +172% and +162% respectively - only resellers sustained this
growth between 2018 and 2019, at +148%, albeit at a slower pace. Meanwhile, manufacturers reported
a 26% decline in the number of machines sold.
Preferred distribution channels for manufacturers and resellers have been direct sales to mining
companies (80%), closely followed by online stores which open equipment purchase to all (67%). These
gures are fairly consistent across all manufacturers and resellers, regardless of their geographic location
or sale destination. A marginal number of resellers and manufacturers indicate relying on other, often
more informal, channels, such as messaging and social media platforms (e.g. Telegram, Twitter).
THE FINANCIALISATION OF MINING
Hashrate: a new commodity for derivatives markets?
Hashrate refers to the amount of computing power used to generate a valid PoW. Mining
pools have long been described as a simple aggregator of hashrate. In reality, pools and
their operators do more than pooling hashrate from miners. Practically speaking, pools
purchase hashrate from hashers, contributing to the commoditisation of this computing
power. As for any other commodity, the development of a spot market has led to the
introduction of derivative contracts. At the time of writing, a small number of companies
have started issuing a suite of nancial products based on hashrate. Miners may see this
development as an hedging opportunity to better manage their risks and improve their
cashow situation.
Increased competition among mining players and tighter prot margins have led them to explore various
strategies to hedge risks and generate additional cashows. In particular, the recent development of new
nancial instruments targeted at miners, such as hashrate forwards and difculty futures,33 has spurred
active discussion in the industry and made the headlines in cryptoasset-native outlets.34
What is a block subsidy halving?
Pioneered by Bitcoin, a “halving” corresponds to a periodic decrease (generally by 50%,
hence “halving”) of the block subsidy distributed to miners for every newly mined block
as determined by the supply issuance schedule. Several PoW coins have adopted these
halving events; however, it appears primarily consequential for PoW coins with a high
production cost, such as Bitcoin. Block subsidy is the main component of miners’ revenues
as transaction fees – which constitute the second element of the block reward – remain
marginal for the majority of cryptocurrencies other than Bitcoin and Ethereum. As such,
the scheduled reduction of block subsidy directly inuences miners’ protability.
33 Future contracts trading against Bitcoin’s future mining difculty.
34 See for instance Zhao, W. (2020) New York Power Plant Sells 30% of Its Bitcoin Mining Hashrate to Institutional Buyers. Coindesk.
Available from: https://www.coindesk.com/new-york-power-plant-greenidge-has-sold-up-to-30-of-its-bitcoin-mining-hash-rate
[Last accessed: 20 August 2020].
3rd Global Cryptoasset Benchmarking Study
35
Figure 24, however, shows that miners’ hedging strategies remain relatively elementary, and primarily
consist of holding cryptoassets (58%) or at reserves (41%). Only a handful of miners make use of
sophisticated nancial instruments, such as cryptoasset (12%) or hashrate (14%) derivatives, or choose
to collateralise their coins (15%).35
35 The reader should note, however, that this survey response data predates the decentralised nance (“DeFi”) explosion which may
have increased the use of collateralisation by miners. For a discussion on decentralised nance, please refer to Section 7.
Figure 24: The use of complex nancial products is limited to a handful of actors
Geographic distribution reveals that North American mining actors are twice as likely to use hashrate
derivatives than APAC actors and six times more likely than European actors (Figure 25). However,
miners from either APAC or North America are equally likely to enter cryptoasset derivative contracts.
Factors, such as availability of these nancial products and regulatory clarity, may explain these regional
discrepancies.
Figure 25: Holding cryptoasset reserves remain the main strategy used by miners across regions
Hedging Strategies Employed by Miners
Share of mining actors
Holding cryptoasset reserves
Holding fiat national currency reserves
Buying cryptoassets on demand
Collateralising cryptoasset(s)
Use of hashrate derivatives
Use or cryptoasset derivatives 12%
14%
15%
17%
41%
58%
Hedging Strategies by Region
Share of mining actors
Holding cryptoasset reserves
Holding fiat national currency reserves
Buying cryptoassets on demand
Use of hashrate derivatives
Use or cryptoasset derivatives
11% 24% 33% 10%
15% 5% 17% 14%
15% 5% 29%
19% 10% 17% 24%
37% 38% 17% 57%
56% 48% 83% 67%
Collateralising cryptoasset(s)
Asia-Pacic Europe Latin America
and the Caribbean North America
3rd Global Cryptoasset Benchmarking Study
36
SECTION 3: THE OFF-CHAIN STORY
ON-CHAIN AND OFF-CHAIN STORIES
Most analyses of cryptoasset usage and activity are based on data generated by on-chain activity
(“on-chain data”). On-chain activity refers to transactions that clear and settle on the corresponding
blockchain base layer (e.g. Bitcoin, Ethereum). Tracking tools then turn this raw blockchain data into
readable information to produce valuable data insights (e.g. blockchain explorers). Further investigation,
such as those performed by blockchain forensic rms, associates wallet addresses with real-world
entities to examine on-chain transaction ows between known actors on the network.
On-chain data analysis has been helpful in understanding the share of illicit activity using
cryptocurrencies, which is estimated to amount to less than 1.1% of total volumes transacting on 25
chains.36 On-chain analysis is also useful to deduce the value being moved on-chain between real-world
entities, demonstrating for instance that exchanges account for 90% of all funds sent by cryptoasset
services.37
However, on-chain data only tells us part of the story since it highlights what happens between entities
that use the blockchain base layer to settle their transaction, but does not capture transactions between
entities using an intermediary to settle their transaction outside of the blockchain layer, for example two
traders on an exchange’s internal order-book. The latter can only be studied using off-chain data.38
A typical example of off-chain data is trading volumes or market data reported by individual cryptoasset
exchanges, where there have been numerous controversies surrounding faked trade volumes. Another
often cited resource for off-chain data has been volumes displayed on peer-to-peer exchanges, such as
LocalBitcoins. The information reported on peer-to-peer trading platforms similarly depicts a somewhat
skewed picture of cryptocurrency usage: the platform is primarily used to identify reliable brokers, and
subsequently a signicant share of trade is taking place outside the platform and left unrecorded.
Another unknown in the cryptoasset usage realm is the amount of trades taking place over-the-counter
(OTC). These trades cannot purely be captured either by on-chain or existing off-chain data. Previous
estimates place OTC trades at two to three times larger than global exchange volumes. More recent
anecdotes estimate it at around USD 600 million a day.39
In this section, we use data collected from exchanges, which provide a nancial market for cryptoassets,
and payment service providers, who facilitate the use of cryptoassets for payments of goods and
services (for example, merchant services, bill payment service, etc.40) to offer a glimpse into the use of
cryptoassets off-chain.
36 Chainalysis (2020). The 2020 State of Crypto Crime. Chainalysis. Available from: https://go.chainalysis.com/rs/503-FAP-074/
images/2020-Crypto-Crime-Report.pdf [Last accessed: 21 August 2020].
37 Chainalysis (2020). Who’s Who On The Blockchains? The Chainalysis Guide to Cryptocurrency Typologies. Chainalysis. Available from:
https://go.chainalysis.com/rs/503-FAP-074/images/Typologies-Report-nal.pdf [Last accessed: 21 August 2020].
38 In our previous benchmarking study, we made the distinction between two types of off-chain transactions: “trusted” transactions
that are recorded by, and reliant upon, service providers for internal clearing and settlement, and “trust-minimised” transactions
that are based on payment channels using the blockchain exclusively for settlement (e.g. the Lightning Network).
39 Chaparro, F. (2019) Inside B2C2: The crypto market making rm that almost closed shop in 2018, and is now growing market share across
the globe. The Block. Available from: https://www.theblockcrypto.com/daily/47750/inside-b2c2-the-crypto-market-making-rm-
that-almost-closed-shop-in-2018-and-is-now-growing-market-share-across-the-globe [Last accessed: 24 August 2020].
40 For a full description, see Rauchs et al. (2018) 2nd Global Cryptoasset Benchmarking Study. Cambridge Centre for Alternative
Finance. Available from: https://www.jbs.cam.ac.uk/wp-content/uploads/2020/08/2019-09-ccaf-2nd-global-cryptoasset-
benchmarking.pdf [Last accessed: 24 August 2020].
3rd Global Cryptoasset Benchmarking Study
37
OFF-CHAIN CRYPTOASSETS AND FIAT CURRENCY SUPPORT
41 The current generation of stablecoins are digital tokens that offer a xed conversion rate to a specic asset or commodity reserves
(e.g. at-collateralised such as Tether, USDC, or the Gemini Dollar).
42 Maddrey, N. (2020) The Rise of Stablecoins. Coin Metrics. Available from: https://coinmetrics.substack.com/p/coin-metrics-state-of-
the-network-f0a [Last accessed: 24 August 2020].
Figure 26: Service providers have rallied to support stablecoins
Bitcoin continues to be the most popular cryptoasset on exchanges, payments, and storage service
providers, which is unsurprising given its high convertibility into sovereign at currencies and other
cryptoassets even though its support has declined slightly over time from 98% of service providers in
2017 to 90% in 2020 (Figure 26). Ether has seen rapid gains in its availability since 2017 and is now
the second most common token, reecting the extent to which smart contracts and decentralised
applications rely on the Ethereum blockchain. The growth in popularity of ERC-20 coins also reects
this shift. Litecoin, Bitcoin Cash, and Ripple are available at about half of service providers. Despite
increasingly strict regulations and concerns over their use for dark market activities, privacy coins Zcash
and Monero are still becoming increasingly more available, and are supported at 24% and 17% of service
providers respectively.
Stablecoins,41 both asset-backed and algorithmic, are also becoming more available, with Tether support
growing from 4% to 32% of service providers and all non-Tether stablecoins growing from 11% to 55%.
This increase is not simply from service providers holding stablecoins diversifying their holdings, but
rather more service providers offering stablecoins. Among exchanges alone, the number of exchanges
offering at least one stablecoin increased from 11% to almost half (48%) of the same. In June 2020 more
value was transacted using stablecoins than Bitcoin for the rst time.42
Bitcoin (BTC)
Ethereum (ETH)
90%
74%
55%
93%
76%
56%
98%
33%
26%
Supported Cryptoassets
Share of service providers
Litecoin (LTC)
55%
11%
5%
Non-Tether stablecoins
32%
4%
Tether (USDT)
Ripple (XRP) 47%
38%
13%
24%
17%
19%
27%
26%
58%
19%
11%
10%
40%
8%
24%
Monero (XMR)
Bitcoin SV (BSV
ERC-20
ZCash (ZEC)
Etherum Classic (ETC)
Other cryptoassets
2020 2018 2017
Bitcoin Cash (BCH) 51%
48%
3rd Global Cryptoasset Benchmarking Study
38
Stablecoin issuers promise a xed, or windowed, conversion rate between their token and corresponding
underlying asset (similar to an exchange rate peg). In response to price deviations from the peg, an
investor has an incentive to buy (sell) the token from the issuer at a one-for-one rate and sell (buy) the
token in the secondary market when that price trades above (below) parity.
Historically, traders primarily used stablecoins to facilitate quick at-denominated transfers between
cryptoasset exchanges for arbitrage. Albeit less commonly, stablecoins have also been used as an
alternative to highly volatile cryptoassets for temporarily storing wealth.43 Following the price crash of
cryptoassets in March 2020, the tokens saw a surge in demand as investors sought to meet liquidity
needs and avoid exposure to the highly volatile markets. This resulted in several stablecoins trading at
a premium (trading at a higher price than their peg).44 However, deviations in stablecoin parities are not
one sided; collateral concerns (in the case of reserve-backed stablecoins) or mechanism concerns (in the
case of “two-coin” systems) have caused tokens to trade at a discount relative to their peg.
As of August 2020, the largest and most successful stablecoin, Tether, had a market capitalisation of
USD 10 billion, representing 80% of the total stablecoin market cap weathering various controversies.
For example, in April 2019, Tether Limited ofcials conrmed that only 74% of Tether was backed by cash
and other securities, and not 100% backed as had been understood.45
The scalability of these tokens is hindered by the ever-changing regulation in the cryptoasset space: in
July 2020 the FATF released a report on stablecoins, emphasising the need for stablecoin issuers to
comply with global anti-money laundering (AML) and counter-terrorist nancing (CFT) standards.46
Though monitoring and centralisation may threaten the immediate widespread adoption of stablecoins,
such necessary regulatory infrastructure bolsters the legitimacy of both issuers and stablecoins.
43 Lyons, R. and Viswanath-Natraj, G. (2019) What Keeps Stable Coins Stable?. SSRN Electronic Journal. Available from: https://jbs.cam.
ac.uk/wp-content/uploads/2020/08/2020-conference-paper-lyons-viswanath-natraj.pdf [Last accessed: 24 August 2020].
44 Coin Metrics Research (2020). The Rise of Stablecoins. Coin Metrics. Available from: https://f.hubspotusercontent00.net/
hubfs/5264302/The%20Rise%20of%20Stablecoins.pdf [Last accessed: 24 August 2020].
45 De, N. (2019). Tether Lawyer Admits Stablecoin Now 74% Backed by Cash and Equivalents. CoinDesk. https://www.coindesk.com/
tether-lawyer-conrms-stablecoin-74-percent-backed-by-cash-and-equivalents [Last accessed: 24 August 2020].
46 FATF (2020) FATF Report to G20 on So-Called Stablecoins. Financial Action Task Force. Available from: www.fatf-ga.org/
publications/virtualassets/documents/report-g20-so-called-stablecoins-june-2020.html [Last accessed: 24 August 2020].
Figure 27: All national at currencies are becoming more widely available
USD
EUR
59%
51%
30%
62%
47%
45%
29%
CNY 12%
11%
JPY 21%
9%
34%
Supported National Fiat Currencies
Share of service providers
GBP
Other national fiat currencies
16%
7%
KRW
2020 2018
3rd Global Cryptoasset Benchmarking Study
39
The growth in popularity of stablecoins has not prevented providers from increasing their supported
at currencies (Figure 27): US Dollar (USD) support has grown from 47% to 59% of service providers,
while the Japanese Yen (JPY) saw the largest increase from 9% to 21%. Non-major sovereign currencies
are also increasingly offered, with an increase from one third to two thirds of all providers from 2018 to
2020 offering a national at currency that was not USD, Euro (EUR), British pound (GBP), Chinese Yuan
(CNY), JPY, or Korean Won (KRW).
The growing regulatory clarity may have helped this increase in support for national at currencies, as
service providers may have previously avoided at currencies to avoid nancial regulations. With many
regulations now updated to include cryptoasset service providers even if they do not incorporate at
currency, the gains from not listing a sovereign at currency is diminished.
OFF-CHAIN ACTIVITY PROVIDERS
EXCHANGES
A key component in the off-chain story are exchanges. Exchanges’ internal ows reveal that these
platforms are primarily used as at on-off-ramps, i.e. when a user seeks to enter the cryptoasset market
by converting its at currencies into cryptoassets, or leave and convert cryptoassets into at. Fiat-
cryptoasset transactions make up most of exchanges’ trades, both in terms of trading volumes and
transaction numbers, while at-at trades are a small share of trades.47
Once on-boarded onto an exchange platform, users may choose to settle their transactions off-chain (for
example, engage in an open market buy/sell order on the exchange’s order book, or an internal transfer
on the exchange’s recordkeeping system to another user account within the exchange). The user may
also direct their on-boarded exchange balance to a wallet external to the exchange, a transaction that
normally necessitates an on-chain transaction.
47 So-called stablecoins are classied as “cryptoassets” for the purpose of this section.
Figure 28: APAC, and to a lower extent European exchanges appear to be primarily used for trading purposes
Transaction Type Breakdown
Share of transaction
Transaction destinationExchange location
3rd Global Cryptoasset Benchmarking Study
40
Large exchanges have a higher average of transactions on their open market than small exchanges (70%
compared to 42%), with users of smaller exchanges almost twice as likely to send transactions to an
external wallet (14% on large exchanges and 33% on small exchanges).48 This is consistent with small
exchanges being used more intensively as on-off-ramps, while larger exchanges are used for trading.
The unequal distribution of small and large exchanges also leads to regional patterns in exchanges
(Figure 28). The presence of large exchanges in APAC means that very little of APAC exchanges’ volumes
leave their exchange platform, consistent with the observation that half of APAC exchanges are large
exchanges.49 The opposite is true for the predominantly small exchanges in MEA, where up to 47% of
transacted volumes is directed to external wallets.
In 2019, over 70% of transaction volume for exchanges headquartered in Europe, LAC, and North
America were at-cryptoasset transactions (Figure 29) compared to 54% of APAC transaction volume.
42% of APAC volume stemmed from cryptoasset-cryptoasset transactions, although the APAC
cross-exchange variation for cryptoasset-cryptoasset trades ranges from almost 0% to almost 100%.
The LAC region has a similarly large variation in crypto-crypto trade shares across exchanges. MEA
platforms stand out for the popularity of at-at transactions, which account for 25% of trade volumes,
considerably higher than other regions. The at-at transaction population in MEA exchange could
reect the larger geographic distribution, presented in Section 4.
48 We dene a large exchange to be one with more than 40 full-time equivalent employees. In most regions small exchanges are 70%
or more of all exchanges, though small exchanges represent 95% of all exchanges in MEA. APAC breaks with the 70% rule of thumb
and has approximately the same number of large and small exchanges.
49 The data however does not provide further detail about genuine public trading and proprietary trading.
Figure 29: In APAC and LAC, cryptoasset-cryptoasset trades take up more than one third of total transaction
volumes
A possible contributing factor for the regional differences in cryptoasset-cryptoasset trade volume is
that many exchanges are used to onboard users into the cryptoasset ecosystem before they migrate
to APAC exchanges for trading purposes. We can provide three pieces of supporting evidence for this
assertion:
the majority of APAC exchange transactions are directed to the open market within the exchange
(i.e. exchange’s internal order-book) consistent with trading behaviour. Additionally, our survey
sample reveals 83% of APAC platforms have an internal order book, compared to only 40% for
MEA exchanges,
APAC platforms usually support a wider range of coins. The average APAC exchange supports
40+ coins, compared to 13 in Europe or 10 in North America, and
41%
19%
13%
14%
35%
54%
74%
62%
83%
65%
5%
7%
25%
3%
Asia-Pacic
Europe
Latin America and the Caribbean
Middle East and Africa
North America
Exchanges Internal Currency Mix (2019)
Share of internal transactions
Cryptoasset <-> Cryptoasset Fiat <-> Cryptoasset Fiat <-> Fiat
3rd Global Cryptoasset Benchmarking Study
41
APAC exchanges offer considerably greater leverage to users allowing for the chance of greater
gain (or loss) for speculative investors.50 In our survey, 55% of surveyed exchanges offering
leverage to users are headquartered in APAC, followed by 30% out of Europe. These APAC
exchanges are well known in the industry for their high leverage multiples, with a median at 15x
and some outliers offering up to 110x leverage (Figure 30).
50 Under a leverage model, the exchange offers a loan to users to use for buying or selling cryptoassets. There are various different
models for leverage, for example margin trading.
Figure 30: Margin trading is more widely available on APAC exchanges
PAYMENT PLATFORMS
The payment service provider landscape continues to be dominated, both in transaction volume and
transaction number, by at-cryptoasset transactions, which make up nearly two-thirds of all volumes
(65%) and transactions (63%). This dominance of at-cryptoasset features in both large and small
payment service providers, though large providers have over twice the transaction share of cryptoasset-
cryptoasset transactions as small providers (35% to 16%) as shown in Figure 31.
Further geographic breakdown reveals patterns similar to exchanges, with payment service providers
in APAC reporting a greater share of cryptoasset-cryptoasset volumes compared to European, North
American, or MEA actors.
Figure 31: Cryptoasset-at trades are more frequent on small payments service providers
There is a difference in the payment values for domestic and cross-border transactions . Low-value
Exchanges’ Highest Leverage Multiple per Region
100
90
80
70
60
50
40
30
20
10
0
North America
Middle East
and Africa
EuropeAsia-Pacic
Highest leverage multiple
54%
68%
35%
16%
11%
16%
Large
Small
Payments Internal Currency Mix (2019)
Share of internal transaction volume
Cryptoasset <-> Cryptoasset Fiat <-> Cryptoasset Fiat <-> Fiat
3rd Global Cryptoasset Benchmarking Study
42
domestic transactions (below USD 100) account for 44% of all domestic transactions, while high-value
domestic transactions (over USD 1,000) are less than one third (31%) of domestic transactions. In
contrast, low-value transactions account for only 30% of the total cross-border transactions, whereas
higher-value transactions account for 45%. This contradicts anecdotes of individuals using cryptoasset
payment service providers to facilitate personal international payments that would otherwise be too
small to be economically feasible to transmit internationally via established entities.
There are also notable differences between large and small payment providers regarding the payment
value mix in cross-border transactions but relative uniformity in the value mix for domestic transactions.
Large payment providers predominantly offer 65% high-value cross border transactions against 8% low-
value cross border transactions, whereas small payment providers offer 41% high-value cross border
transactions against 35% low-value cross border transactions. This indicates that these different service
providers are respectively serving disparate market niches. An examination of the transaction types
conrms this inference: small payment providers’ transactions are twice as likely to be peer-to-peer in
comparison to large payment providers’ transactions (33% to 17%), with both consumer-business (36%
to 45%) and business-to-business (31% to 38%) transactions more common across large providers than
small providers (Figure 32).
Figure 32: Transactions are divided relatively equally across the three payment groups for small payments service
providers
For lower and upper middle-income countries, payment service providers mostly facilitate small
payments, whether domestic or international, while in high income countries most payments are large.
This income-based pattern is repeated at the regional level, with Europe and North America engaging
in mostly large payments domestically and internationally, LAC and MEA engaging in mostly small
payments domestically and internationally, and APAC mixing, with a higher share of smaller payments
domestically and larger payments internationally (Figure 33).
38%
31%
45%
36%
17%
33%
Large
Small
Type of Payments Transfers
Share of internal transaction volume
Business-to-business payments Consumer-to-business payments Consumer-to-consumer transfers
3rd Global Cryptoasset Benchmarking Study
43
Figure 33: Low-value payments account for the largest proportion of transaction volumes in LAC and MEA regions
Domestic and Cross-border Payment Size per Region
Share of internal transaction volume
USD 0 - USD 100 USD 100 - USD 1,000 > USD 1,000
Domestic Cross-border
Asia-Pacic
42%
27%
31%
Europe
32%
25%
43%
Latin America
and the
Caribbean
64%
22%
14%
Middle East
and Africa
57%
30%
13%
North
America
24%
16%
60%
Asia-Pacic
27%
20%
53%
Europe
19%
28%
53%
Latin America
and the
Caribbean
57%
23%
20%
Middle East
and Africa
49%
25%
26%
North
America
7%
22%
71%
3rd Global Cryptoasset Benchmarking Study
44
SECTION 4: PROFILING
CRYPTOASSET USERS
Descriptive data on cryptoasset holders are crucial for industry participants and regulators alike.
Individual service providers often conduct consumer surveys to better tailor their services to their
user prole. Meanwhile, in some jurisdictions, regulatory authorities have undertaken similar studies
to grasp the size of the cryptoasset market in their jurisdiction and understand consumer attitudes
toward cryptoassets to better assess what part of the population is most at risk. This section offers
complementary insights into the composition of cryptoasset holders.
USER NUMBER AND ACTIVITY
In 2018, the 2nd Global Cryptoasset Benchmarking Study estimated the number of identity-veried
cryptoasset users at about 35 million globally.51 Applying the same methodology, an update of this
estimate indicates a total of up to 101 million unique cryptoasset users across 191 million accounts
opened at service providers in Q3 2020 (Figure 34).52 This 189% increase in users may be explained by
both a rise in the number of accounts (which increased by 37%), as well as a greater share of accounts
being systematically linked to an individual’s identity, allowing us to increase our estimate of minimum
user numbers associated with accounts on each service provider.
51 The methodology is detailed p.33 in Rauchs et al. (2018) 2nd Global Cryptoasset Benchmarking Study. Cambridge Centre for
Alternative Finance. Available from: https://www.jbs.cam.ac.uk/wp-content/uploads/2020/08/2019-09-ccaf-2nd-global-
cryptoasset-benchmarking.pdf [Last accessed: 24 August 2020].
52 It should be noted that this gure does not include self-hosted wallets.
Figure 34: The total number of cryptoasset accounts held at service providers has experienced a fourfold increase
over four years
Despite clear limitations
to this methodology
(see Section 2 p.33 of the
2nd Global Cryptoasset
Benchmarking Study for a
more thorough discussion),
there are reasons
to believe that this
estimate offers a reliable
approximate gure of
the total number of
cryptoasset holders
globally. Other recent
consumer research also
highlights an increase in
cryptoasset ownership.
A study commissioned
by the UK nancial
regulator estimated
an increase of 78%
compared to 2019
Lower Bound Estimate of Total Cryptoasset Users and Accounts
200
180
160
140
120
100
80
60
40
20
0
Millions
2018 2020 Q3
139
191
2016
45
5
35
101
Total number of accounts Total number of users
3rd Global Cryptoasset Benchmarking Study
45
estimates.53 Finally, large service providers have concomitantly publicly reported a rise in the entrance of
new users, especially in the rst quarter of 2020.54
However, consumers vary widely in how they engage with cryptoassets. User activity as reported by
service providers is one useful metric to monitor users’ interaction with the cryptoasset ecosystem.
Service providers operating from North America and Europe generally report higher user activity, with
the median rm indicating that 40% of total users are considered active. However, as the spread of the
distribution shows, reported gures on user activity vary signicantly between actors from the same
region. This heterogeneity is particularly pronounced for North American, European, MEA companies.
On average, small service providers experience higher level of user activity
Disparity in user activity may also be explained by inconsistent denitions used by service providers to
monitor activity levels. While 54% of service providers dene as “active” a user that logs in or interacts
with the service at least once a month, 33% do so using a weekly timeframe. Perhaps unsurprisingly,
exchanges that voluntarily dene user activity using a shorter timeframe (e.g. weekly) also report greater
user activity, on average, than those using a monthly-based denition.
USER GEOGRAPHY
53 English, R., Tomova, G. and Levene, J. (2020) Research Note: Cryptoasset consumer research. Financial Conduct Authority (UK).
Available from:https://www.fca.org.uk/publications/research/research-note-cryptoasset-consumer-research [Last accessed: 24
August 2020].
54 See for instance, Partz, H. (2019) Coinbase Added 8 Million New Users in the Past Year. Cointelegraph. Available from: https://
cointelegraph.com/news/coinbase-added-8-million-new-users-in-the-past-year [Last accessed: 24 August 2020], Binance (2020)
Binance 2019: Year in Review. Binance. Available from: https://www.binance.com/en/blog/418708327988203520/Binance-
2019-Year-in-Review [Last accessed: 24 August 2020], and Krekotin, V. (2020) Sharing Thoughts on Security, OKEx’s Jay Hao Says
Customers Come First. Cointelegraph. Available from: https://cointelegraph.com/news/sharing-thoughts-on-security-okexs-jay-hao-
says-customers-come-rst [Last accessed: 24 August 2020].
Figure 35: Firms continue to majorly serve users from their region of operations
Geographic Distribution of Customer Base
Share of total users
Firm location User location
3rd Global Cryptoasset Benchmarking Study
46
Cryptoasset users span the globe, but rms continue to primarily serve customers based in their region
of operation (Figure 35), in line with our 2018 ndings. This is particularly the case for companies
headquartered in LAC, but less so for those in MEA and North America. North American rms’ presence
in other regions seems to conrm the success of the internationalisation strategy adopted by these
companies. Finally, the slightly more geographically distributed customer base of MEA companies -
APAC, Europe, and North America each take 10% on average of MEA companies’ customer base - may
reect the presence of diaspora from MEA into those three regions.
USER TYPES
Several studies have reported a growing interest from institutional investors in cryptoasset markets. For
instance, a blind survey of American and European institutional investors conducted by Fidelity Digital
Assets reveals that 36% of respondents have invested in cryptoassets and that three in ve believe that
cryptoassets should form part of their portfolios.55 Interestingly, a growing share of US institutional
investors have exposure to cryptoassets via the derivatives market.
The past four years have indeed seen a rapid increase in the number of nancial instruments available
to market investors. These instruments, such as perpetual swaps, options and futures, were initially
launched by unregulated offshore entities, slowly followed by regulated incumbents. Since 2016,
unregulated products have been dominating the market in terms of volume and aggregated open
interest, though the share of regulated products has progressively increased. Due to counterparty risk
and duciary responsibility, institutional engagement is expected to grow in tandem with the expansion
of regulated nancial products and regulatory clarity.56
Despite the considerable development of institutional-grade nancial instruments and infrastructure,
our data suggests that cryptoasset service providers’ customer base is still primarily retail-driven,
showing that despite growing institutional interest, the conversion rate (from expression of interest to
investment) remains limited. As further explored in Section 7, there are several hurdles that the industry
needs to overcome to bolster engagement of traditional institutional investors (e.g. asset managers,
family ofces), such as concerns on market manipulation and price volatility.
55 Bhutoria, R. (2020) Institutional Digital Asset Survey Report - 2020 Review. Fidelity Digital Assets. Available from: https://www.
delitydigitalassets.com/bin-public/060_www_delity_com/documents/FDAS/institutional-investors-digital-asset-survey.pdf [Last
accessed: 24 August 2020]. Note: The survey included cryptoasset hedge funds, which may skew the results since their investment
portfolios naturally comprise a greater share of cryptoassets.
56 For a comprehensive overview of the cryptoasset derivatives market, please refer to, Todd, R. (2020) Institutional Digital Asset
Derivatives Markets. The Block. Available from: https://www.theblockcrypto.com/genesis/65421/institutional-digital-asset-
derivative-markets [Last accessed: 24 August 2020].
Figure 36: Retails take the lion’s share of service providers’ customer base
75%
63%
75%
70%
82%
16%
30%
20%
30%
10%
9%
7%
5%
8%
Asia-Pacic
Europe
Latin America and the Caribbean
Middle East and Africa
North America
Customer Base Breakdown by Type
Share of total users
Retail clients Business and institutional clients Unknown
3rd Global Cryptoasset Benchmarking Study
47
There are noticeable differences between companies from different regions (Figure 36). While the
majority of rms’ customer base is composed of individual clients, North American and European rms
report that an average of 30% of their customers are business and institutional clients. In contrast, this
gure for APAC, LAC, and MEA is 16%, 10%, and 20% respectively.
There is also signicant disparity among service providers within North America and Europe, suggesting
that these regions have a mix of retail-focused and institutional-focused companies, whereas companies
in regions like LAC are primarily targeting retail cryptoasset users.
A deeper analysis of the type of business and institutional clients reveals that, globally, cryptoasset
service providers primarily serve cryptoasset hedge funds (37%), online merchants (30%), and miners
(27%). Interestingly, company size mix is fairly stable across institutional and businesses clients (60% to
40% respectively for small and large service providers) except for traditional hedge funds, which equally
deal with large- and small-scale rms.
Figure 37 reveals noticeable regional differences in the type of business and institutional clients served:
beyond cryptoasset hedge funds, North American, APAC, and European companies appear to engage
more with traditional investors (hedge funds, venture capitalists, and other institutional investors),
whereas LAC and MEA companies primarily focus on online merchants and cryptoasset companies
(other than miners).
Figure 37: The composition of business and institutional clientele differ from region to region
A considerable share of APAC companies serves miners (41%), in part explained by the high level of
mining activities in the region, especially in China. Miners use their services to liquidate their coin
inventory for national at currencies and cover at-based expenditures. Evidence from the study of
miners’ hedging strategies (see Figure 24 in Section 2) also reveal that a growing number of miners rely
on service providers to collateralise their coins (i.e. through loans) and unlock additional funds.
Business and Institutional Clients per Region
Share of service providers
Crypto hedge funds
Family Offices 5%
46%
Banks
Asset Managers
Asia-Pacic Europe Latin America and the
Caribbean Middle East
and Africa North America
24%Venture Captial (VC) rms 19% 12% 5% 30%
9% 4% 5%
4%
9% 5%
Other 10% 12% 4% 15% 15%
Miners 41% 27% 31% 5% 25%
Other institutional investors 20% 25% 12% 5% 15%
Online merchants 39% 25% 23% 50% 20%
Traditional hedge funds 17% 19% 15% 30%
36% 35% 15% 50%
Other cryptoasset companies 20% 25% 8% 30% 20%
Brick and mortar merchants 22% 18% 27% 20% 20%
3rd Global Cryptoasset Benchmarking Study
48
Alternative sources using on-chain data analysis conrm that mining actors, in particular mining pools,
are active users of Asian exchange platforms.57 For instance, in 2019, 28% of bitcoins owing into
exchanges originated from mining pools, though this distribution is not even, with the vast majority of
bitcoins (77%) sent to one of the top 10 exchanges.58
57 Chainalysis (2020) Mining Pool Market Power. Chainalysis. Available from: https://go.chainalysis.com/2020-markets-case-study-
mining-pools.html [Last accessed: 24 August 2020].
58 This analysis excludes bitcoins that were transferred from exchanges.
3rd Global Cryptoasset Benchmarking Study
49
SECTION 5: REGULATORY AND
COMPLIANCE
COMPLIANCE BENCHMARKS
The cryptoasset industry has been on the radar of regulators from as early as 2011. The industry has,
however, experienced a steady appreciation in the levels of regulatory scrutiny since 2013 when popular
darknet market Silk Road was shut down and the market witnessed its largest bubble since the inception
of Bitcoin. Since then, authorities have issued guidance, retrotted their existing regulation, or even
developed bespoke regulatory frameworks to bring cryptoasset-related activities within their scope.59
These changes in the regulatory environment often translate into internal adjustments in how rms go
about meeting mandated requirements. Increasing regulatory burden is ranked as the second highest
operational risk by rms, regardless of their size and location (see Appendix).
RESOURCES ALLOCATED TO COMPLIANCE
Across all surveyed geographic regions, 75% of the respondents reported having an in-house compliance
team. The median share of a company’s total headcount and costs allocated to compliance is relatively
consistent across regions, with the exception of Europe (Figure 38). Half of European service providers
report compliance headcount and costs equal or greater than 13% (compared to 8% in 2018). The top
25% of European respondents report compliance headcount and cost greater than 18% and up to 40%.
59 For a more detailed analysis of the cryptoasset regulatory landscape, please refer to, Blandin et al. (2019) The Global Cryptoasset
Regulatory Landscape Study. Cambridge Centre for Alternative Finance. Available from: https://www.jbs.cam.ac.uk/leadmin/user_
upload/research/centres/alternative-nance/downloads/2019-04-ccaf-global-cryptoasset-regulatory-landscape-study.pdf [Last
accessed: 24 August 2020].
Figure 38: European service providers allocate most resources to compliance
Compliance Headcount and Cost Allocation
Median share
North America
8%
Latin America and
the Caribbean
8%
Europe
13%
Middle East
and Africa
8%
Asia-Pacific
8%
3rd Global Cryptoasset Benchmarking Study
50
In 2019, EU member states were expected to transpose the 5th AML Directive, whose scope was
broadened to include cryptoasset exchanges and custodial service providers, into their national
regulatory frameworks. This has increased the compliance requirements for companies and
consequently some European rms have already announced their closure in the face of increased
regulatory burden.60
While the median rm in the four other regions allocate roughly the same share of human and nancial
resources to compliance, an analysis of North American rms’ distribution shows signicant variability
amongst rms from the region. This may be a reection of diversity in regulatory approaches amongst
different US states because entities operating in the USA have to navigate a patchwork of state-level
regulations, in addition to federal ones. Consequently, companies operating in all US states would have
more resources dedicated to compliance to cope with the heterogenous state-level regulations. US
companies drawn to crypto-friendly states (e.g. Wyoming) and avoiding states with tighter regulations
(e.g. New York City) are likely to have relatively lower compliance cost and headcount overall.
From a regulatory perspective, 2019 was a particularly active year for the US cryptoasset industry,
with both federal and state regulators, releasing new statements and guidance. At the federal level, for
instance, FinCEN issued a guidance in May 2019 regarding the applicability of the Bank Secrecy Act
(BSA) to cryptoasset businesses. According to the guidance, with the exemption of non-custodial wallets,
decentralised exchanges that do not settle trades, and certain infrastructure providers (e.g. DApp
developers, cloud miners), most cryptoasset businesses qualify as money transmitters and must comply
with AML/KYC regulations.61
One common element of these new regulatory developments worldwide is the absence of distinction
between entities exclusively supporting cryptoassets and those supporting both at currencies and
cryptoassets. This equal treatment of cryptoasset-only and cryptoasset-and-at entities seems to have
led to an increase in at support among cryptoasset service providers. When comparing 2018 and 2020
samples, we noticed that more than one in three companies that exclusively offered cryptoasset in 2018
now support at currencies.
Over 30% of companies that exclusively supported cryptoassets in 2018 have added at support since then.
From a data analysis perspective, this development matters because the extension of regulatory
authorities’ supervisory mandate to the cryptoasset realm had the effect to erase minimal discrepancies
in compliance headcount and costs observed in 2018 between cryptoasset-only and cryptoasset-and-
at companies. While these discrepancies were already limited in 2018, greater homogeneity is observed
across both types of rm groups in 2020. For instance, the median compliance headcount is similar for
cryptoasset-and-at and cryptoasset-only entities, while the median for compliance cost is slightly lower
for the latter. We observe greater dispersion in the compliance cost of cryptoasset-and-at companies,
which may account for regional variability.
AUDIT OF CRYPTOASSET RESERVES
Regulatory frameworks governing traditional nancial institutions mandate banks to hold minimum
reserve requirements and to perform independent audits. Similarly, cryptoasset reserve audits are
good industry practices to provide assurance that a rm maintains equivalent reserves of its customers
funds, and play a pivotal role in upholding stakeholders’ condence. They ensure rms are operating
transparently and adhering to set performance, security, and compliance standards. The audits could
either be on-chain proof-of-reserves or more traditional audits performed by independent third parties.
60 See for instance, BitKassa Team (2020) BitKassa closing down. BitKassa. Available form: https://www.bitkassa.nl/en/bitkassa-closing-
down [Last accessed: 24 August 2020].
61 Similarly, in 2020, Canadian regulators issued a notice indicating that any entity dealing in “virtual currency” will be considered as a
money service business (MSB) and must register with the regulators.
3rd Global Cryptoasset Benchmarking Study
51
Proof-of-reserves and proof-of-solvency
Cryptographic features of public blockchains enable new forms of public accountability,
sometimes referred to as “proof-of-reserves”.62 There are several techniques to perform
proof-of-reserves audits available to custodial service providers, but in its simplest form,
proof-of-reserves audit entail signing a transaction with the entirety of their on-chain
customer funds and publishing a signed message emanating from the associated address
where funds are held.
Combined with the disclosure of the total customer liabilities of a service provider (i.e.
how much a service provider owes to its users), proof-of-reserves would help prove that a
custodial service provider has sufcient cryptoassets in reserve to meet (at a minimum) its
liabilities.
Although this approach has its own limitations (e.g. unaccounted or omitted liabilities,
impracticality), regular on-chain audits combined with more traditional externally-led
audits might play a pivotal role in enhancing trust and transparency in the conduct of
operations by custodial service providers.
The need to perform independent cryptoasset reserves audits is more imperative in the absence of a
widespread practice of public proof-of-reserves programmes. 59% of rms indicate that they had their
cryptoasset reserves audited by an independent comptroller over the past 12 months, primarily based
out of Europe (35%) and APAC (31%).
Two out of ve service providers did not conduct an independent audit of their cryptoasset reserves in the past 12
months
Surprisingly, a 2018-2020 comparison of custodial service providers performing externally-led
cryptoasset reserve audits indicates that only 54% of custodial service providers had their reserves
audited in the past 12 months. This is a 24 percentage points decline from 2018, which may suggest that
rms feel a decrease in scrutiny relative to 2018 when it was revealed that the stablecoin Tether did
not have 100% reserves as previously thought. Additionally, large rms are nearly twice likely to have
undergone independent audit relative to small rms. Similarly, 22% more of companies incorporated in
a FATF country conduct externally-led audits than those incorporated in a jurisdiction that is not part of
the inter-governmental body.
62 Bitcoin Improvement Proposal 127 “Simple Proof-of-Reserves Transactions” (2019). Available from: https://github.com/bitcoin/
bips/blob/master/bip-0127.mediawiki [Last accessed: 24 August 2020].
Figure 39: Independent cryptoasset reserves audits are more prevalent among APAC and North America
companies
Externally-led Audit of Cryptoasset Reserves
Share of service providers
Asia -Pacic
North America
Europe
Latin America and the Caribbean
Middle East and Africa 38%
58%
75%
80%
40%
3rd Global Cryptoasset Benchmarking Study
52
User’s legal compliance
As the industry matures, individual cryptoasset holders also come under scrutiny by
regulatory and government bodies, in particular tax authorities. A signicant majority
of surveyed service providers (67%) indicate that they provide users with compliance
documentation, such as tax receipts. However, this might not be sufcient to ensure users
accurately report liabilities to authorities. Inconsistencies in transaction reporting, absence
of uniformity in tax forms issued by various service providers, or lack thereof, may render
cryptoasset holders’ compliance processes more difcult.
AUTHORISATION OF SERVICE PROVIDERS
Relevant authorisation regimes, involving either a licensing or registration process, for cryptoasset
service providers are jurisdiction-specic and depend on the type of services provided, as well as the
type of assets supported. 37% of the surveyed service providers are licensed or regulated, whereas
11% have an outstanding application (3% are both license holders and prospective applicants) as shown
in Figure 40.63 The research team has been unable to identify any license or registration (granted or
outstanding) for 55% of survey respondents. It should, however, be noted that some service providers
are engaged in activities that may not warrant or exempt licensing/registration processes (e.g. hardware
wallet manufacturing, non-custodial wallet software provision).
63 The gures presented in this section only consider national and federal licenses and registrations.
Figure 40: Just over two out of ve rms are licensed or in the process of obtaining a license
The median number of licenses/registrations held by a single rm is one, but some companies hold
as many as ve. As shown in Figure 41, license holders primarily hold a crypto-specic license (42%),
followed by payment or e-money licenses (29%), and money business licenses (28%). The existence of a
crypto-specic licensing regime arises either from the introduction of a bespoke regulatory framework,
which specically regulates cryptoasset-related activities as a standalone activity (e.g. Gibraltar’s DLT
Provider licensing regime), or from the retrotting of an existing law or regulation to include activities
dealing with cryptoassets (e.g. Japan’s amendment of its Payment Service Act). Several other types of
license exist, including, inter alia, qualied custodian (5%) and banking (3%).
License and Registration Status
Share of service providers
55% 34%
11%
No identiable
license or registration License and/or registered
Outstanding
3rd Global Cryptoasset Benchmarking Study
53
Figure 41: Firms may seek alternative licenses when no crypto-specic licensing regime exists
Note: This chart is based on public data collecte d by CCAF in combination of data provided by Cr yptoCompare
Registration or license seekers typically have one outstanding application and primarily pursue a crypto-
specic (58%) or payment institution (21%) license.
Of the licensed and registered entities, licenses and registration were primarily issued by British (23%)
and American (23%) regulatory authorities. Switzerland and Estonia rank both at the third place (17%)
of most cited jurisdictions for license and registration. Interestingly, only a small share of licenses held
were issued by jurisdictions with a bespoke regulatory regime, such as Gibraltar (4%). Entities with an
outstanding application were primarily seeking approval from the US (26%) and Singaporean (26%)
authorities. The other main jurisdictions for outstanding applications were Hong Kong, the UK, South
Korea, Japan, Switzerland, and Thailand.
72% of license holders or prospective applicants obtained or are seeking a
license/registration from their home country (i.e. operational HQ)
With 48% and 58% of rms being registered/licensed, European and North American rms are roughly
twice more likely to be regulated than companies from LAC or APAC, which both report that 23% of
rms are registered or license holders. Meanwhile, slightly less than one in three companies operationally
headquartered in the MEA region holds a license or a registration.
Of all identied licenses and registrations, 36% were granted in 2019 after the rise of regulatory scrutiny
following the boom of 2017-2018 (Figure 42). As the cryptoasset regulatory landscape continues to
evolve and the number of pending applications continues to grow, more rms are expected to be licensed
and registered in the coming years.
Geographic Distribution of Licensed Firms and Type of License Held
Share of licensed and registered service providers
North America
20%
Latin America and
the Caribbean
9%
Europe
48%
Middle East
and Africa
9%
Asia-Pacific
14%
42%
Crypto-specic
29%
Payment or
e-money institution
28%
Money services
business
13%
Trading facility
13%
Broker-dealer
3rd Global Cryptoasset Benchmarking Study
54
Figure 42: The number of issued licenses almost doubled between 2018 and 2019
Note: This chart is based on public data collecte d by CCAF in combination of data provided by Cr yptoCompare
AML AND KYC PROCEDURES
The cryptoasset industry is progressively integrating with the global nancial system and, by the same
token, required to abide by the same standards. Around the world, AML/KYC standards are being
harmonised to regulate an industry that is global by its very nature. The charge has rst been led by the
FATF to ensure consistency in AML/KYC requirements across its member states.
These updated requirements seem to have spurred industry actors to enhance their due diligence
measures: in 2020, the vast majority (77%) of surveyed service providers perform AML/KYC checks for
every single account, with only 20% using specic criteria (e.g. withdrawal/deposit thresholds, frequency
of activity, location of account’s owner). This gure is lower for platforms with an exclusive focus on
cryptoassets (Figure 43): while 82% of entities supporting both at and cryptoassets verify user’s
identity for every account, only 48% of cryptoasset-only platforms do so, and 39% use other criteria.
Only a small minority of service providers (3%) do not perform KYC checks at all
However, it should be noted that user identity verication is usually a multi-tier process. Often, the
stringency of the verication process increases with the amount a user is willing to deposit, withdraw,
or trade, but also based on the type of assets for platforms supporting both at and cryptoassets. By
registering and undergoing an identity check of all surveyed respondents, the CCAF research team
found that 72% of them performed KYC checks when dealing with at, but only 46% did so when
customers exclusively used cryptoassets.
It is also relevant to note that, regardless of their size, location, or assets they support, nearly 90% of
surveyed service providers have a policy in place about which parties can access sensitive customer
information, such as identication documents and bank account information.
Granted Licenses per Year
License count
2010
1
2013
2
2015
2
2017
9
2014
1
2016
4
2019
32
2018
17
Unidentied
11
3rd Global Cryptoasset Benchmarking Study
55
FATF Recommendation 16: Travel Rule
The FATF is an inter-governmental body that coordinates member countries’ efforts on
AML and counter-terrorism nancing (CFT). It issues non-binding recommendations to its
members, which it monitors and reviews on a regular basis.
In June 2019, FATF released an interpretive guidance (revised FATF standards) including
virtual assets and virtual asset service providers (VASPs), which sets out several
requirements and recommendations that apply to VASPs, such as licensing, registration,
and Customer Due Diligence (CDD) requirements. Recommendation 16, also referred
to as the “travel rule”, has proven to be particularly challenging for VASPs to implement
technically. The travel rule requires VASPs to obtain, hold and transmit accurate (veried)
originator information, information about intended beneciary, and securely transmit
required customer information. From a technical standpoint, the travel rule necessitates
the establishment of common technical standards (e.g. messaging) to streamline and
standardise information transmission from the originator to its beneciaries, as well as the
development of technological solutions for VASPs to comply with the travel rule globally.
Figure 43: The share of cryptoasset-only companies that do not conduct any KYC checks dropped from 48% to 13%
between 2018 and 2020
The share of accounts’ owners whose identity was veried differs from region to region: it is consistently
high across European and North American rms, approaching 100%, whilst service providers in other
regions report lower and more disparate numbers. The median gure is particularly low in MEA-based
companies, with approximately 50% of account owners having their identity veried.
The median rm in Europe, North America, or Middle East and Africa
reported that 8% of its KYC checks led to account closure
Companies incorporated in a FATF member country more frequently conduct KYC checks on all
accounts, than entities legally headquartered in non-FATF countries (Figure 44). Furthermore, at
the median only 3% of KYC checks lead to the closure or refusal to open an account for companies
incorporated in non-FATF countries, against 8% for FATF-incorporated entities. These trends might be
read as a testament to the importance and inuence of commonly established global standards.
82%
48%
17%
39%
1%
13%
Fiat and Cryptoasset
Cryptoasset-only
Know-Your-Customer Checks Criteria
Share of service providers
All accounts Other criteria No KYC checks
3rd Global Cryptoasset Benchmarking Study
56
Figure 44: At the median, the identity of all account holders has been checked for companies incorporated in FATF
countries
64 It is also worth noting that jurisdictions that adopted a bespoke regulatory framework are often those known for their relatively
exible business regulations, see Blandin et al. (2019) The Global Cryptoasset Regulatory Landscape Study. Cambridge Centre for
Alternative Finance. Available from: https://www.jbs.cam.ac.uk/leadmin/user_upload/research/centres/alternative-nance/
downloads/2019-04-ccaf-global-cryptoasset-regulatory-landscape-study.pdf [Last accessed: 24 August 2020].
REGULATION IMPACT: REDEFINING GEOGRAPHIES
As the previous subsection has extensively shown, regulatory changes have had a direct impact on
companies’ internal compliance standards. The impact of regulation may also be observed from a
geographical standpoint by reshaping the geographic boundaries of the cryptoasset ecosystem.
REGULATORY ARBITRAGE
In the context of cryptoassets, regulatory arbitrage often occurs when companies choose to settle
in jurisdiction(s) offering greater regulatory certainty due to the existence of a bespoke regulatory
framework to supervise cryptoasset activities.64 In this context, the concept of “regulatory arbitrage”
slightly departs from its traditional meaning. It is understood as seeking maximum regulatory certainty and
the most benign environment, rather than to exploit legal and regulatory loopholes (i.e. absence of regulation) as
is often discussed in other industries. Another form of regulatory arbitrage is the exploitation of different
tax rules; a practice which is not specic to cryptoasset businesses but also observed in traditional
nance and other industries.
Nearly one in three companies sought regulatory approval from outside its main jurisdiction of
operations. Another approach to investigate geographic relocation of entities to amenable jurisdictions
is to compare respondents’ operational headquarters with their country of incorporation. On this front,
of surveyed entities, 22% have been incorporated in a country different from where their operational
headquarters are based, and up to 15% in a different region. For these companies, top countries for
incorporation include Switzerland, British Virgin Islands, the UK, and the Republic of Seychelles.
100
90
80
70
60
50
40
30
20
10
0
FATF incorporatedNon-FATF incorporated
Identity Verification of Account Users
Share of total accounts
% of KYC’ed accounts
3rd Global Cryptoasset Benchmarking Study
57
Supervisory duty: whose responsibility?
While numerous regulators take varied approaches to govern service providers in the
industry, several jurisdictional issues arise when attributing organisation’s liabilities.
Recently, the organisational structure of large cryptoasset companies have come to
resemble those of large corporations from other industries. The growing complexity of
corporate structures combined with the inherent global nature of the cryptoasset industry
have rendered the supervisory duty of regulators ever more challenging to execute. It is
often unclear which authorities should oversee an organisation with no physical presence
operating in a given jurisdiction and domiciled overseas. Some authorities and international
bodies, such as FATF, have suggested that cryptoasset service providers should be licensed
and regulated in the jurisdiction where they are ‘created’, obliging regulatory authorities
from the jurisdiction of incorporation to supervise and regulate the entity.
This approach may, however, prove to be insufcient in the event of a company
incorporated in one jurisdiction and domiciled in another through its subsidiaries, or in
cases where the service provider is decentralised and has no home jurisdiction. While a
coordinated risk based response to use of cryptoassets in criminal activities is desirable,
consensus across nations on this front is still a work in process. A possible alternative
approach could also be to implement the Place of Effective Management (POEM) or
Permanent Establishment (PE) model prevalent in few jurisdictions to determine residency
for taxation purposes, where individual jurisdictions dene a threshold/nexus above which
economic activity in that jurisdiction may be subjected to regulatory supervision. However,
while the use of POEM may solve many concerns around round tripping, it may not always
result in a clear determination of domicile or reect outcomes which accord with policy
intentions, given the highly decentralised nature of operations of the industry. Jurisdictions
globally should, therefore, give due regard to substance over form in determining liabilities
and preventing abuse.
CLOSURE AND OPENING OF A LOCATION
The introduction of a new regulation, enforcement actions, or any other regulatory developments are
believed to considerably inuence a rms’ decision to open or close an ofce in a jurisdiction. However, as
Figure 45 shows, changes in the regulatory environment affect location closure and opening to differing
degrees: 35% of survey respondents indicate that regulatory changes have led them to open a new
location, compared to 18% for location closure.65
Though the data does not show in which jurisdictions facility opening and/or closure takes place, this
trend might be a direct consequence of the so-called “chilling effect of regulation”. This concept refers to
the vagueness of existing applicable rules, creating more confusion and uncertainty than the complete
absence of regulations. Lack of explicit regulatory requirements for cryptoasset businesses in certain
jurisdictions may have led some to seek greater regulatory certainty by relocating part of their activities
in jurisdictions with bespoke regulatory framework, or at least unambiguous applicability of existing
regulations.
65 The kind of changes inducing these decisions was not specied in the data. As such, regulatory change may refer to increased
regulation, as well as relaxation in a regulatory regime.
3rd Global Cryptoasset Benchmarking Study
58
Figure 45: Changes in regulation are more likely to drive location opening than closure
66 21shares (2020) Crypto Exchanges Database by 21Shares-apr21,20. Available from: https://docs.google.com/spreadsheets/d/1vhM-
IYVzWx1xytLxbL9FVeZdnceNJKVuvwrNdtQ7qmo/edit#gid=0 [Last accessed: 24 August 2020].
USER BANS
Another well-documented impact of regulation is the geographic restrictions imposed by service
providers to deny users from certain geographies access to their services. Unsurprisingly, these
geographic restrictions replicate those that exist in other industries, namely countries subject to
international sanctions, such as Iran, North Korea, etc (Figure 46). It has also been widely reported that
US customers, mostly from New York City, were banned from these platforms due to the high cost of
compliance to serve users in these locations.
Figure 46: Users from the USA are as often banned from exchanges than Iraqi users
Source : 21shares database66
Impact of Regulatory Change on Location Opening and Closure
Share of service providers
18%
74%
8% 6%
35%
59%
Yes No Prefer no t to answer
18%
Of respond ents
closed a locatio n
due to regulatory
change
35%
Of respond ents
opened a loc ation
due to regulatory
change
Iran
North Korea
Syria
Yemen
35%
35%
36%
36%
37
40%
40%
34%
Top 8 Restricted Locations
Share of exchanges
Sudan
Iraq
Cuba
USA
3rd Global Cryptoasset Benchmarking Study
59
SECTION 6: IT SECURITY
THE DEVELOPMENT OF BEST MARKET PRACTICES
Greater retail adoption post-2017 turned cryptoasset service providers into prime targets for hackers.
Blockchain analytics rm Chainalysis reports that the number of attacks on service providers has been
on the rise with the total value of stolen funds peaking at nearly USD 900 million in 2018 (Figure 47).67
Companies have consequently ramped up their security measures, e.g. increased use of cold storage,
additional verication layers for withdrawal, and a more stringent monitoring of transactions to detect
suspicious activities.
67 Chainalysis (2020) The 2020 State of Crypto Crime. Chainalysis. Available from: https://go.chainalysis.com/rs/503-FAP-074/
images/2020-Crypto-Crime-Report.pdf [Last accessed: 24 August 2020]
Figure 47: Improved security measures by service providers have helped shrink down the amount of lost value
between 2018 and 2019
Source: Chainalysis, The 2 020 State of Crypto Crime
Attackers on the other hand have adapted to these enhanced security measures by designing more
sophisticated phishing attacks, or using enhancing-privacy methods (e.g. CoinJoin, mixers). The
adherence to best market practices for IT security might be a signal that companies across the spectrum
are keeping up with more advanced attacks.
Number of Hacks and Total Value Stolen per Year
Currencies included: ADA, BCH, BTC, ETH, EOS, LTC, NANQ, NEM, USDT, XRP and others
2011
$0.04
2012
$0.70
2014
$483.10
2016
$75.50
2019
$282.60
2013
$3.20
2015
$8.10
2018
$875.50
2017
$86.00
Cryptocurrency stolen in millions of USD
$0
$250
$500
$750
$1,000
1
3 3 3
5
4 4
6
11
Number of hac ks
3rd Global Cryptoasset Benchmarking Study
60
Figure 48: The use of cold storage and multisig is close to becoming an industry standard
“Hot” and “cold” storage: misnomers for cryptoasset custody
The notions of “hot” and “cold” have been borrowed from the data storage world to
describe the level of security in the system used to safekeep users’ private keys. Often, in
the cryptoasset space, the low latency of ‘hot’ storage is associated with lower security
level compared to ‘cold’ storage systems that are composed of several security layers to
unlock funds, hence necessarily more time-consuming. However, these adjectives, which
originally relate to the latency of different tiered storage plans, should not be conated
with the degree of security of a storage system. In fact several companies have started
designing key storage systems that combine the security level of so-called ‘cold’ storage
with the low latency of ‘hot’ storage.
At the median, custodial platforms usually keep a slightly lower share of cryptoasset funds in cold storage
(85%) compared to non-custodial ones (90%). This should not come as a surprise; platforms offering
services generally associated with high user activity (e.g. trading platforms) often happen to be custodial
ones. For instance, 79% of exchanges that have their own order-book have control over user funds (i.e.
hold users’ private keys). Therefore, to allow users to swiftly access their funds when trading, the service
provider must hold a limited amount of customer funds in cold storage. This trend is also a reection of
users’ perpetual trade-off between convenience and security.
RESOURCES ALLOCATION TO IT SECURITY
Non-custodial service providers generally spend a greater share of their resources, both nancial and
human, on IT security at between 11-20% compared to 6-10% for custodians (Figure 49). Development
costs and timeline are believed to be generally higher for non-custodial systems, requiring greater
allocation of resources over a longer period.
It is worth noting that service providers that allow both the user and the service provider to each hold
a private key are classied as “non-custodians” for the purpose of this report. This form of co-managed
custody is often complex and associated with several layers of security and iterative processes to unlock
customers’ funds, which would require additional stafng.
Share of service providers
Storage System
90%
Use cold
storage
70%
Use multisig for
hot storage
81%
Use multisig for
cold storage
3rd Global Cryptoasset Benchmarking Study
61
Figure 49: The median non-custodial service providers allocate a greater share of its resources to IT security than
the median custodian
Small service providers are more likely to report higher gures for IT security headcount with at least
50% of small service providers indicating that 11-20% of their workforce is orientated towards IT
security against 6-10% for large ones. However, both groups report a median IT security cost between
11-20% of total costs.
From a regional perspective, apart from LAC service providers, respondents from all regions reported
the same median share of IT security headcount (6-10%). However, the median company in APAC
and LAC reports a higher share of costs associated with IT security than the median companies
headquartered elsewhere. This does not, however, mean that companies from other regions spend less
overall on IT security. As discussed in Section 5, European and North American companies are more
likely to spend a higher share of their nancial resources on compliance; this variable cost in turn takes up
some of the share of IT security resources. Similarly, costs associated with other activities, such as human
resources, marketing, R&D, may also result in a lower share of resources being allocated to IT security
overall. As such, IT security may be considered a large xed cost component.
SECURITY AUDITS
The 2nd edition of the Global Cryptoasset Benchmarking Study found that a signicant number of survey
respondents were reluctant or unable to disclose information about both internal and external security
auditing processes. It was speculated that this could be interpreted as either a general lack of awareness
around formal security standards, or that entities do not follow security auditing best practices, or in
some cases both. The results of this year’s survey conrm the hypothesis from the previous study that
the adoption of best practice standards and formal security processes varies considerably across regions
and jurisdictions.
Unsurprisingly, internal audits were undertaken more frequently than external audits across all ve
regions; internal audits are generally easier to carry out and do not require the disclosure of sensitive
information to third parties, or additional external cost. For example, in APAC, North America and
MEA (31%), a signicant proportion of internal audits were carried out monthly. In Europe and LAC,
the highest proportion of internal audits were carried out on a weekly basis. Whereas, in all regions, the
highest proportion of respondents, carried out external audits on an annual basis.
Up to 12% of respondents from all regions stated that they never engaged in internal audits. As for
external audit, this gure is particularly high for respondents from MEA: 31% indicated they had never
engaged in external audits. This gure is 20% for LAC, 18% for North America, 14% for Europe, and at a
noticeably low 3% for APAC respondents.
IT Security Cost and Headcount
Share of service providers
Custodian
Non-custodian
3% 4%
3%
5%
7%
18%
17%
25%
18%
31%
16%
35%
18%
25% 10%
25% 15%
17%
16%
9%
17%
9%8%
16%
7%
6%
Custodian
Non-custodian 19%
0% 1-5% 6-10% 11-20% 21-40% 41-60% 60+%
1%
CostHeadcount
3rd Global Cryptoasset Benchmarking Study
62
Figure 50: On average, non-custodians conduct external and internal security audits more frequently
68 Evertas (2020) Huge lack of capacity in insurance sector for cryptoassets. Evertas. Available from: https://www.evertas.com/in-the-
news/huge-lack-of-capacity [Last accessed: 24 August 2020].
INSURANCE
The safety of cryptoasset storage should not be left to well-designed IT security systems alone. The
insurance of funds is a key component of a sound service offering. Nevertheless, 46% of surveyed
service providers indicated not being insured against any risk. Companies with insurance plans are
primarily insured against cybercrime, professional errors (including directors and ofcers liability
insurance), loss or theft of private keys, and hazards. These observations hold true for both custodial and
non-custodial service providers, although a slightly largest proportion of custodians is insured against the
loss or theft of private keys (Fig ure 51).
Nearly half of service providers are left fully uninsured
The insurance sector for cryptoasset services is undeniably underdeveloped. Other studies have shown
that less than 0.5% of the insurance market goes to cryptoasset coverage.68 The relatively small size of
the cryptoasset industry and the reluctance of insurance carriers to enter this volatile market are the
most probable explanatory factors for this nding.
Figure 51: Little difference appear between the insurance plans of custodian and non-custodians
Frequency of Internal and External Security Audits
Share of service providers
Custodian
Non-custodian
45%
38%
17%
13%
15%
16%
8%
10%
8%
9%
21%
13%
3%
7%
13%
30%
7%
18%
4%19%
8%
22%
12%
18%
8%
Custodian
Non-custodian 17%
Annually Bi-annually Quarterly Monthly Weekly or more often Never Decline to respond
InternalExternal
Insurance Plan Coverage
Share of service providers
None
Cybercrime
Other
Professional services error
Loss or theft of private keys
Hazard (e.g. re)
12% 11%
24% 23%
27% 20%
29% 30%
31%
4%
31%
7%
42% 40%
Kidnap and ransom
Custodian Non-custodian
3rd Global Cryptoasset Benchmarking Study
63
Although service providers often do not have formal insurance policies in place - nor access to such -
certain service providers have earmarked funds allocated to cover specic insurable events for the users.
Regional breakdown unearthed some interesting disparities (Figure 52). For instance, a great proportion
of European companies are not insured at all (43%), so are companies in Latin America (46%) or MEA
(52%). In comparison, 24% and 37% of North American and APAC entities respectively indicate not
having any insurance coverage. For insured companies, the proportion of the types of insurance
coverage are somewhat similar across regions.
Figure 52: Companies in APAC tend to be insured against a wider range of risks
Insurance Plan Coverage per Region
Share of service providers
None
Cybercrime (e.g. hacking)
Other
Professional services error
Loss or theft of private keys
Hazard (e.g. re)
22%
24%
27%
29%
29%
10%
37%
Kidnap and ransom
Asia-Pacic Europe Latin America and the
Caribbean Middle East
and Africa North America
10% 8% 10%
24% 12% 24% 24%
22% 23% 14% 33%
28% 31% 19% 38%
31% 35% 24% 33%
4% 5%
43% 46% 52% 24%
3rd Global Cryptoasset Benchmarking Study
64
SECTION 7: BALANCING BETWEEN
INTEGRATION AND INNOVATION
As the industry enters a new decade, it faces several challenges to further expansion, including
integrating with traditional market infrastructures and maintaining a sustained pace of innovation. To
full both objectives, industry participants may have to invest more time and resources into compliance
and industry restructuring, while bringing continuous improvements to innovative solutions, such as
decentralised nance.
ENHANCED TRANSPARENCY AND COMPLIANCE
The 12-month review report released by FATF in June 2020 reveals that a growing number of
jurisdictions have implemented the revised FATF standards on cryptoasset service providers.69 As
discussed in Section 5, wider implementation of AML/KYC rules would help bolster legitimacy and trust
in the industry. Previous investor surveys often quote the lack of transparency and regulatory certainty
around cryptoassets as a concern for institutional investors, along with other concerns, such as volatility
and absence of reliable valuation models.
Survey data suggests that compliance and institutional investor adoption are indeed closely related. For
instance, Figure 53 indicates that service providers incorporated in a member jurisdiction of FATF tend
to serve investors from the traditional nancial system more often that rms from non-FATF countries:
the share of business and institutional clients for the median service provider legally headquartered in a
FATF country is two times higher than that of enterprises incorporated in a jurisdiction that is not part
of the inter-governmental body. Although both groups equally deal with traditional hedge funds (17%),
companies incorporated in a FATF member state are nearly twice as likely to also serve other types of
institutional investors, such as family ofces and asset managers.
69 FATF (2020) 12-month Review of the Revised FATF Standards on Virtual Assets and Virtual Asset Service Providers. Available from:
https://www.fatf-ga.org/publications/fatfrecommendations/documents/12-month-review-virtual-assets-vasps.html [Last
accessed: 30 August 2020].
Figure 53: Cryptoasset service providers incorporated in a jurisdiction member of FATF are more likely to serve
traditional institutional investors
Business and Institutional Clients for FATF and Non-FATF Incorporation
Share of service providers
33% 29%
26% 29%
33% 39%
Non-FATF incorporated FATF incorporated
Crypto hedge funds
Online merchants
Miners
11% 21%Other institutional investors
17% 22%Brick and mortar merchants
11% 11%Other
17% 23%Other cryptoasset companies
13% 21%Venture Captial (VC) rms
17% 17%Traditional hedge funds
3rd Global Cryptoasset Benchmarking Study
65
A DECOUPLING OF FUNCTIONS ACROSS THE VALUE CHAIN
Another possible driver of institutional
adoption is the evolution of cryptoasset
market structure towards a more
traditional setup. As the cryptoasset
industry congures to resemble traditional
nancial market infrastructure, institutional
investors might be in a better position to
engage with the industry. An emerging
trend in this evolving conguration is the
decoupling of activities amongst different
rms, as well as within rms (via the
setup of subsidiaries). To a certain extent,
the segregation of functions across the
value chain similar to traditional market
infrastructure is already happening. For
instance, 45% of respondents indicate
using a third-party as part of their cold
storage system (Figure 54). Cryptoasset-
native custodians have dominated the
market (64%) and been the go-to for
service providers willing when outsourcing
custodial duty to a third-party. The
relatively limited custodial role played
by banks and other more traditional
custodians may be simply explained by the persisting reluctance of these actors to enter the cryptoasset
space due to regulatory barriers. This gure is likely to grow as the regulatory landscape evolves for
traditional banking actors.70
THE GROWTH OF ‘DECENTRALISED FINANCE’
DeFi is an umbrella term that refers to an emerging nancial software stack that consists of several
protocols, platforms and applications built on top of public blockchains.
DeFi development and usage has gained most traction on the Ethereum blockchain, but decentralised
nance applications may also emerge on other smart contract blockchains that offer different trade-offs
around scalability, transaction speed and degree of decentralisation.
The DeFi space is still largely experimental and, in general, most DeFi applications cannot be considered
meaningfully decentralised by any measure. The majority of these applications are still dependent on kill
switches, centralised oracles, or some other centralised support or maintenance. A stated core objective
for many developer teams is to focus on increasing decentralisation over time. The emergence of
governance tokens and new incentive mechanisms are examples of experimental approaches designed to
make DeFi protocols less dependent on centralised control. Depending on the region, DeFi is in fact the
second or third most cited future development that may be a game changer for service providers
(Table 4).
70 For instance, in July 2020, the US Ofce of the Comptroller of the Currency issued an interpretative letter allowing US national
banks and savings associations to offer services for the custody of cryptoassets. See, Gould, J. (2020). Interpretive Letter #1170.
Ofce of the Comptroller of the Currency. Available from: https://www.occ.gov/topics/charters-and-licensing/interpretations-and-
actions/2020/int1170.pdf [Last accessed: 24 August 2020].
45%
Make use of third-
party custody
3rd party No 3rd party
55%
Other
6%
Banks
10%
Crypto-native
custodians
29%
Usage of Third-party Custody
Share of service providers
Figure 54: More than one in three service providers outsource
its custody responsibility to a cryptoasset native custodian
3rd Global Cryptoasset Benchmarking Study
66
Notably, decentralised nance protocols are subject to several risks such as:
smart contract risk: as the amount of money locked into decentralised nancial systems grows,
they inevitably become targets for hackers who look to exploit vulnerabilities in smart-contract
code security. Stacking and composability of smart-contracts also pose a risk. Should an
underlying smart-contract break then the stack may fall like a house of cards,
oracle risk: some decentralised protocols rely on so-called oracles to access off-chain data (e.g.
cryptoasset exchange rate against at currencies). Oracles, either hardware or software, funnel
real world data to the smart contract. As several attacks targeted at decentralised protocols
have shown, oracles are a possible source of systemic risk and their data feeding role is prone to
manipulation,71
nancial risk: many DeFi protocols are dependent upon underlying cryptoasset collateral that is
subject to volatility. Hence, a large and sudden drop in cryptoasset price presents a liquidation risk
to users of the protocol,
regulatory risk: regulation regarding the DeFi space is unclear and untested in many jurisdictions.
As the space grows, the response of regulators to decentralised nancial applications is a
regulatory risk that needs greater study and understanding.
71 See for instance the double attack on the bZx ethereum-based ending project in February 2020. Available from: https://etherscan.
io/address/0x360f85f0b74326cddff33a812b05353bc537747b#internaltx [Last accessed: 24 August 2020].
Table 4: Service providers’ sentiment over future developments per respondent’s region of operations
“Other” include s “tokenised rights”, “FATF guidance”, “ lending”
Service providers assess impact of future developments (per region)
Asia-Pacic Europe Latin America
and the Caribbean
Middle East
and Africa North America
Stablecoins 4.0 3.7 3.5 4.1 4.3
Staking 3.8 3.3 2.8 3.3 3.1
Decentralised finance (DeFi) 3.6 3.2 3.4 3.4 3.2
Layer-2 solutions (e.g. Lightning network) 3.0 3.1 2.9 3.3 3.0
Central Bank Digital Currencies (CBDCs) 3.3 2.9 2.7 3.5 3.1
Security tokens issued on a public blockchain 3.2 2.9 3.1 3.4 2.7
Sidechain 2.9 2.8 2.5 3.3 2.9
Privacy enhancing overlays 3.1 2.7 2.8 3.5 3.1
Non-fungible Tokens (e.g. ERC-721) 2.8 2.5 2.9 2.3 2.5
Other developments 1.8 3.1 4.0 3.5 N/A
Responde nts scored these c ategories on a 1-5 scale:
1: Not impor tant at all 2: Not importa nt 3: Ne utral 4 : Somewhat impor tant 5 : Very important
Lowest aver age score Highest ave rage score
3rd Global Cryptoasset Benchmarking Study
67
APPENDIX
MINERS’ INFLUENCE
Figure 55: Large miners are less likely to think that they have very high inuence on protocol governance than small
and individual miners
OPERATIONAL RISKS FOR MINERS
Table 5: Miners rank their concerns over operational risks (per region)
Miner Inuence on Protocol Governance
Share of mining actors
2017
Large miners Small miners
(incl. individuals)
10%
20%
31%
41%
41%
27%
37%
27%
25%
35%
24%
23%
30%
16%
7%
23%
40%
27%
12%
15% 24%
5%
18% 17%
2018 2017 20202018
8% 7%
Very low Low Medium High Very high
3% 3%
2020
4%
Miners ranks concerns over operational risks (per region)
Asia-Pacic Europe Latin America and
the Caribbean
Middle East
and Africa North America
Government Seizure or shutdown of
your mining-supporting facilities 3.3 2.4 3.1 2.5 1.8
Halving 3.3 2.9 3.0 3.5 2.9
Regulations creating barriers to mining 3.3 2.8 3.7 2.5 3.2
Sudden increases in energy prices 3.2 3.0 3.5 3.0 2.6
Increased taxation of mining prots 3.0 2.6 3.4 1.5 2.8
Unplanned protocol change 2.9 2.5 2.6 3.5 2.0
Intensive competition among miners
of the same cryptoasset 2.9 2.7 3.0 3.0 3.0
Cyber attacks (e.g. DDoS) 2.8 2.4 3.7 3.0 2.1
Declining popularity of the cryptoasset
you mine 2.7 2.2 3.3 3.0 2.2
Lack of immediate availability of
state-of-the art hardware 2.7 2.5 3.7 2.5 2.9
Planned protocol change 2.5 2.7 2.6 3.5 1.7
Responde nts scored these c ategories on a 1-5 scale:
1: Not impor tant at all 2: Not importa nt 3: Ne utral 4 : Somewhat impor tant 5 : Very important
Lowest aver age score Highest ave rage score
3rd Global Cryptoasset Benchmarking Study
68
Table 6: Miners rank their concerns over operational risks (per size)
ADDITIONAL RISKS FOR MINERS
Table 7: Miners rank their concerns over additional risks (per region)
Miners ranks concerns over operational risks (per size)
Small Miners (incl. individuals) Large Miners
2017 2018 2020 2017 2018 2020
Halving N/A N/A 3.0 N/A N/A 3.2
Regulations creating barriers to mining N/A 3.3 3.1 N/A 2.9 3.2
Sudden increase in energy prices N/A 3.1 3.0 N/A 3.5 3.1
Government seizure or shutdown of your mining facilities N/A 3.6 2.4 N/A 2.4 3.0
Intensive competition among miners of the same cryptoasset 3.2 3.3 2.9 3.3 3.2 2.9
Increased taxation of mining prots N/A 3.2 2.8 N /A 2.9 2.9
Cyber attacks (e.g. DDoS) 2.8 3.1 2.5 3.0 3.3 2.7
Lack of immediate availability of state-of-the art hardware 2.9 3.4 2.9 2.4 2.5 2.6
Unplanned protocol change 2.5 3.0 2.5 1.6 3.4 2.6
Declining popularity of the cryptoasset you mine N /A 3.0 2.5 N/A 3.0 2.5
Planned protocol change N/A N/A 2.4 N/A N/A 2.4
Responde nts scored these c ategories on a 1-5 scale:
1: Not impor tant at all 2: Not importa nt 3: Ne utral 4 : Somewhat impor tant 5 : Very important
Lowest aver age score Highest ave rage score
Miners ranks concerns over additional risks (per region)
Asia-Pacic Europe Latin America and
the Caribbean
Middle East
and Africa
North
America
Unfavourable global regulation related to cryptoasset mining 4.0 3.4 3.6 4.5 3.4
Unfavourable global regulation related to cryptoassets 3.9 3.1 3.4 2.0 3.2
Centralisation of hashing power in the hands of a few 3.5 4.3 3.7 2.0 4.4
Criminal use of cryptoassets 3.1 2.7 3.5 4.5 2.3
Underdeveloped fee market 3.1 3.6 3.4 3.0 2.9
Risk of state-sponsored attack on a cryptoasset system 3.0 3.3 2.8 3.0 2.6
Lack of open-source mining software 2.9 3.5 2.6 2.0 2.5
Centralisation of hashing power in a particular geographic area 2.8 3.5 3.3 1.5 3.8
Centralisation of mining equipment production in a particular
geographic area 2.7 3.7 3.4 3.5 3.7
Popularity of pre-mined/'mining-less' cryptoassets 2.7 3.0 3.0 4.0 2.4
Too many cryptoassets in the market 2.6 2.5 2.9 5.0 2.2
Emergence of Central Bank Digital Currencies (CBDCs) 2.3 2.7 3.1 5.0 2.6
Responde nts scored these c ategories on a 1-5 scale:
1: Not impor tant at all 2: Not importa nt 3: Ne utral 4 : Somewhat impor tant 5 : Very important
Lowest aver age score Highest ave rage score
3rd Global Cryptoasset Benchmarking Study
69
Table 8: Miners rank their concerns over additional risks (per size)
OPERATIONAL RISKS FOR SERVICE PROVIDERS
Table 9: Service providers rank their concerns over operational risks (per size)
Miners ranks concerns over additional risks (per size)
Small Miners (incl. individuals) Large Miners
2017 2018 2020 2017 2018 2020
Unfavourable global regulation related to cryptoasset mining N/A 3.3 3.5 N/A 3.0 4.0
Unfavourable global regulation related to cryptoassets N/A 3.3 3.2 N /A 3.2 3.8
Centralisation of hashing power in the hands of a few 3.9 4.4 4.2 3.3 4.0 3.4
Criminal use of cryptoassets N/A 3.2 2.7 N/A 2.8 3.1
Underdeveloped fee market N/A N/A 3.2 N/A N/A 3.1
Risk of state-sponsored attack on a cryptoasset system N/A 3.4 3.1 N/A 2.9 2.8
Centralisation of hashing power in a particular geographic area 3.7 3.9 3.5 3.1 3.7 2.7