Small Bus Econ (2021) 57:353–406

https://doi.org/10.1007/s11187-019-00309-8

Global drivers of cryptocurrency infrastructure adoption

Ed Saiedi & Anders Broström & Felipe Ruiz

Accepted: 10 December 2019 / Published online: 2 March 2020

# The Author(s) 2020

Abstract A vast digital ecosystem of entrepreneurship occurrence of country-level inflation crises. On the other
and exchange has sprung up with Bitcoin’s digital infra- hand, our findings also suggest that active support for
structure at its core. We explore the worldwide spread of Bitcoin is higher in locations with well-developed bank-
infrastructure necessary to maintain and grow Bitcoin as ing services. Finally, we find support for the view that
a system (Bitcoin nodes) and infrastructure enabling the bitcoin adoption is also partly driven by cryptocurrencies’
use of bitcoins for everyday economic transactions usefulness in engaging in illicit trade.
(Bitcoin merchants). Specifically, we investigate the role
of legal, criminal, financial, and social determinants of Keywords Bitcoin network . Digital currencies .
the adoption of Bitcoin infrastructure. We offer some Cryptocurrencies . Financial technology (Fintech) .
support for the view that the adoption of cryptocurrency Bitcoin nodes . Bitcoin merchants
infrastructure is driven by perceived failings of traditional
financial systems, in that the spread of Bitcoin infrastruc- JEL classifications O3 . P40 . O57 . L86 . L17 . D84 .
ture is associated with low trust in banks and the financial L26
system among inhabitants of a region, and with the

1 Introduction
Electronic supplementary material The online version of this
article (https://doi.org/10.1007/s11187-019-00309-8) contains
supplementary material, which is available to authorized users. Cryptocurrencies are proliferating. A decade on since their
dawn with the invention of Bitcoin, the value of all
E. Saiedi (*) : A. Broström cryptocurrencies reached $0.25 trillion. To put that in
Department of Industrial Economics and Management, KTH
Royal Institute of Technology, Lindstedtsvägen 30, 114
perspective, there is $1.7 trillion USD and $1.4 trillion
28 Stockholm, Sweden Euros in circulation today (European Central Bank 2019;
e-mail: [email protected] U.S. Federal Reserve Board 2019). As of November 2019,
bitcoin is the world’s sixth largest currency in circulation.1
A. Broström
The average daily trading of cryptocurrencies has
e-mail: [email protected] surpassed 1% of trading in foreign exchange markets, the
world’s largest market by trading volume.2 Bitcoin
E. Saiedi : F. Ruiz
Department of Business Administration and Statistics, School of
Industrial Engineering (ETSII), Universidad Politécnica de
Madrid, 28006 Madrid, Spain 1
https://howmuch.net/articles/how-much-currencies-are-worth
2
In October 2019, daily trading in all cryptocurrencies varied between
$41.7B and $15 6.3B (source: https://coinmarketcap.com/charts/),
F. Ruiz whereas daily trading in global foreign exchange was approximately
e-mail: [email protected] $6.6 T in April 2019 (Bank of International Settlements 2019).
354 E. Saiedi et al.

transactions and unique accounts alone have grown at information dissemination and transaction verification,
nearly 60% per annum over the past 5 years. In short, Bitnodes, as well as infrastructure necessary for the
cryptocurrencies are being adopted rapidly and broadly. adoption of bitcoin as a means of payment in regular
While theoretical papers are emerging discussing why retail. We believe these types of Bitcoin infrastructure3 to
cryptocurrencies, or digital currencies in general, may be provide an informative lens and context for exploring the
adopted by individuals or businesses (e.g., Cohen 2017; drivers behind the recent growth of the cryptocurrency.
Dierksmeier and Seele 2018; Dodgson et al. 2015), there is In particular, we believe that that by studying patterns of
a scarcity of global empirical studies on drivers of their adoption of these two different types of infrastructure,
adoption. we obtain complimentary perspectives of the drivers
The emergence of cryptocurrencies has often been behind the growth of Bitcoin as a system.
viewed as driven by the opportunity for radical innova- We chose to focus on Bitcoin since it is presently the
tion and entrepreneurship in financial solutions as cre- most well-known and widespread cryptocurrency.4
ated through the spread of new Internet-based technol- Most cryptocurrencies are recorded as clones or as var-
ogy (Iyidogan 2018; Teo 2015). However, recent studies iations of the Bitcoin technology (e.g., Litecoin), build
have emphasized that in order to understand the histor- their blockchain as a fork of the Bitcoin blockchain
ical growth and future prospect of fintech innovations, (e.g., Bitcoin cash) or on nearly identical transaction
we must also understand the nature of the needs ad- ledgers (Gandal and Halaburda 2014; Wang et al.
dressed by such innovations (Cohen 2017; Huang et al. 2019). Bitcoin has the highest cryptocurrency trading
2019; Saiedi et al. 2017). The development and spread volume, constitutes more than half of cryptocurrencies’
of technology is in this sense a prerequisite—but not a market capitalization, provides much-needed liquidity
sufficient factor—for the spread of cryptocurrencies. In to other cryptocurrencies, and enables a secondary mar-
this paper, we focus on socio-economic and institutional ket for the cryptocurrency ecosystem to thrive on.
factors related to potential drivers for interest in Bitcoin’s decade-long dominating position is demon-
cryptocurrency development. Thereby, we seek to offer strated by it being the only cryptocurrency whose price
a novel and broader account for the growth of this type has causality effects on alternative cryptocurrencies or
of financial technology. altcoins5 (Svetlana et al. 2017).
Digital entrepreneurial ecosystems have two founda- Bitcoin has also been shown to play an important role
tional pillars of digital infrastructure and users (Sussan for the emergence of new digital entrepreneurial ecosys-
and Acs 2017) and infrastructure facilitating connectivity, tems, consisting of Initial Coin Offering (ICO) issuers,
e.g., digital infrastructure, are found to particularly enhance payment processors, exchanges, wallets, financial ser-
startup activity (Audretsch et al. 2015). Digital infrastruc- vices, mining hardware, and developers (Sameeh 2018).
tures enable innovation (Henfridsson and Bygstad 2013; Bitcoin, or emulations of Bitcoin, play a central role in
Sussan and Acs 2017), anchor open entrepreneurship this development (The Economist 2008). For example,
(Ingram Bogusz and Morisse 2018), allow for fintech the first ICO was carried out in July 2013 by
platforms to grow (Yermack 2018), and create
3
decentralized work organizations (Tilson et al. 2010). A review of these two types of technical infrastructure and their role
While recent studies have shed light on determinants of for bitcoin as a currency is provided in the Appendix.
4
See https://www.blockchain.com/en/charts/n-transactions-
fintech startup activity or fundraising using total?timespan=all for bitcoin transactions and https://www.quandl.
cryptocurrencies (Fisch 2019; Haddad and Hornuf 2019; com/data/BCHAIN/NADDU-Bitcoin-Number-of-Unique-Bitcoin-
Huang et al. 2019), there are, as of yet, no empirical studies Addresses-Used for unique accounts.
5
Bitcoin constitutes ~ 20–35% of all cryptocurrency trading volumes,
exploring determinants of adoption of the infrastructure
and litecoin or bitcoin cash, which utilize Bitcoin’s blockchain, are
supporting cryptocurrencies. This study fills this gap. respectively the 4th and 7th largest cryptocurrency by volume as of
We explore drivers behind the global uptake of digital January 3, 2019 (coinmarketcap.com/charts). Wei (2018) calculates
Amihud-illiquidity ratios for 456 cryptocurrencies, listing Bitcoin as
infrastructure enabling the use of the most prominent
the most liquid. Hu et al. (2018) document average daily and monthly
digital currency to date; Bitcoin (spelled with capital B bitcoin price cross-correlation of respectively 0.174 and 0.21 with all
when referring to the system, and with lower-case b altcoins with market values exceeding $1 M. Bitcoin’s price cross-
when referring to the unit of account, in keeping with correlation with ether, litecoin, and monero was 0.88 (Fisch 2019),
0.43 and 0.43 (Hu, Parlour, and Rajan 2018) respectively. ICO-bitcoin
convention in the computer science literature.) We con- correlations have increased since January 2018’s cryptocurrency price
sider infrastructure necessary for Bitcoin’s blockchain peaks (Fatás and Weder 2019).
Global drivers of cryptocurrency infrastructure adoption 355

Mastercoin, a cryptocurrency built on the Bitcoin digital currency support over time in a panel data set.
blockchain (Shin 2017). The central role of bitcoin is We have location data, covering the entire world, en-
illustrated by the findings of Masiak et al. (2019), who abling the use of geospatial data analysis to investigate
find that shocks in bitcoin prices positively influence the role of socio-economic and institutional factors, in
ether—the cryptocurrency underlying most ICOs’ parallel to technological and economic factors, in driv-
blockchain platforms—but not vice versa. In this light, ing their adoption. Recent literature has emphasized the
the development of Bitcoin infrastructure might be ex- need to integrate the social, economic, and cultural
pected to play a role in enabling the expansion of digital elements when studying entrepreneurial ecosystems
financing and entrepreneurship rapidly emerging (Spigel 2017). As to the authors’ best knowledge, there
through cryptocurrencies. are no empirical studies yet that have simultaneously
Our results provide some support for the view that investigated the relevance of both types of factors for
bitcoin adoption is driven by perceived failings of tradi- interest in Bitcoin on a global scale. Our unique context
tional financial systems (see, e.g., Cohen 2017; Shiller of cryptocurrency infrastructure allows us to do so.
2019; Vigna and Casey 2015). In particular, we find By increasing our understanding of what socio-
more adoption where distrust in banks and the financial economic and institutional factors that are associated
system are greater, as well as in countries experiencing with the adoption of the infrastructure behind virtual
inflation crises. Meanwhile, the spread of Bitcoin infra- currencies, the study offers important insights to
structure seems to be complimentary to existing finan- scholars seeking to understand the growth of
cial systems, as we observe less adoption where bank cryptocurrencies to date. These results are also of po-
rents and share of the unbanked are highest. In line with tential interest to monetary authorities as well as for
expectations that interest in bitcoin as a speculative developers and entrepreneurs in the virtual currency
investment is a partial driver of adoption of bitcoin ecosystem, including financial institutions, e-
(e.g., Baur et al. 2018b), we find more Bitcoin infra- commerce payment system providers or technology
structure where the willingness to take risks are higher. companies, which are exploring or planning to issue or
We furthermore show that bitcoin adoption is greatest accept virtual currencies.
where the risk of narcotics-related money laundering is The remainder of this article is organized as fol-
greatest, as well as where perceptions of the rule of law lows. In section 2, we discuss a conceptual frame-
is strongest. This latter finding may hint at a shift to work for our analyses. Section 3 provides a thor-
pseudonymous online cryptocurrency crime where ough description of our novel data, starting with an
offline law enforcement is strong. overview of the Bitcoin infrastructure studied. Sec-
While a number of studies make strides in tions 4 describes our methodology, and Section 5
researching the adoption of bitcoins—qualitatively by explains our results. We conclude and discuss im-
surveying experts (Ermakova et al. 2017) or users plications of our findings in Section 6.
(Henry et al. 2018; Schuh and Shy 2016), anonymous
online marketplaces (Böhme et al. 2015), archival data
(Sadhya and Sadhya 2018), or quantitatively using the 2 Conceptual framework
small de-anonymized fraction of an online forum
(Athey et al. 2016) or within one continent solely In considering what socio-economic and institutional
(Yermack 2018), to the best of our knowledge, we factors may be related to the intensity of Bitcoin infra-
provide the first global empirical study on the Bitcoin structure adoption, we build on a view of decisions to set
phenomenon. Our study is also the first to empirically up and operate Bitcoin infrastructure as being related to
analyze the growth of Bitcoin infrastructure. We thereby a combination of extrinsic and intrinsic motivations.
contribute to a nascent body of research that delves into The launch of the Bitcoin system was embedded in
exploring the adoption of financial technologies (Xue idealistic notions of providing means to replace existing
et al. 2011, Rau 2017, Haddad and Hornuf 2019, Huang financial structures, and nurturing an alternative mone-
et al. 2019, Yermack 2018, Higgins 2018, Estrin et al. tary and financial system that would enable greater
2018). anonymity, privacy, and autonomy (Bashir et al. 2016;
We improve upon current research on adoption of Böhme et al. 2015; Dodd 2018; Shiller 2019). Individ-
digital currencies by empirically examining actual uals setting up Bitcoin nodes to grow the peer-to-peer
356 E. Saiedi et al.

network would often do so as an expression of support cryptographic technology, that enables pseudonymity
for Bitcoin as a system and financial phenomenon. in online transactions (Cohen 2017).6
Meanwhile, those running a node do so to verify that Preferences for remaining anonymous in financial
their own transactions are secure, while it may also be transactions may be associated with entirely ideological
related to bitcoin mining activities, where individuals views. In an empirical examination of the related phe-
seek to earn bitcoins. Merchants’ willingness to adopt nomenon of ICOs, Fisch et al. (2019) find such ideo-
technology necessary to accept bitcoins as payment for logical motives to be among the most important drivers
their goods is clearly to a large extent driven by business behind the early development of ICOs. However, pref-
motivations, while the customer preferences to which erences for anonymity also arise due to specific inten-
merchants react to, may considerably be driven by their tions to evade legal authorities. Böhme et al. (2015)
support for Bitcoin on idealistic grounds. ‘Support for document that while consumer payments and buy-and-
Bitcoin’ is thus in our study a central (albeit not directly hold purposes became important drivers of bitcoin
observed) factor. adoption in later stages, in the early use of bitcoin,
The growth of Bitcoin infrastructure can be expected online sale of narcotics, and gambling transfers played
to be directly linked not only to support and positive a very important role.
attitudes, but also to the actual use of bitcoin (e.g., in Against this background, it would seem valid to
terms of the number of people holding bitcoins). We expect that the global spread of bitcoin has been driven
expect that in the relatively early stage of development both by its potential role as a partial substitute role for
that we are studying, the use of bitcoin as a currency and traditional financial services and currencies, and by its
the support for the Bitcoin system are to be understood potential role as facilitator of illicit activity. In what
as interdependent entities. That is, we expect that posi- follows, we consider these two alternative accounts for
tive attitudes towards the cryptocurrency will translate what drives adoption of infrastructure for the supply and
into more use, and we expect greater use to spread the demand of bitcoins.
interest for (and general awareness of) Bitcoin, i.e., to
increase the number of people interested in supporting
the Bitcoin network. In developing hypotheses about the 2.2 Bitcoin as complement or substitute to established
adoption of Bitcoin infrastructure, we hence consider financial systems
what set of socio-economic and institutional factors may
be associated with greater support and use of the The nascent literature on the Bitcoin system and bitcoin
cryptocurrency. currency indicates their financial potential as (a) a new
exchange/payment system, (b) an investment, or (c) a
speculative trading instrument. Bitcoin’s novelty as a
2.1 Bitcoin’s differentiating technologies payment system is due to its relatively low foreign-
exchange transaction costs (Kim 2017), and its indepen-
Cryptocurrencies such as bitcoins have been hailed as dence from monetary authorities, governments or pro-
pioneering and potentially disruptive financial technol- cessing via third-party financial intermediaries. Mean-
ogies. Bitcoin’s blockchain technology allowed for a while, Glaser et al. (2014) find that uninformed bitcoin
novel way of solving the “double spending problem” users are primarily interested in bitcoin as an alternative
intrinsic to digital currencies, without relying on a cen- investment vehicle, rather than an alternative transaction
tral clearinghouse or trusted third party (Folkinshteyn 6
Such opportunities were in line with influential factors at the very
et al. 2015; Böhme et al. 2015). Active support for
inception of the blockchain within the cyberpunk movement. Bitcoin’s
Bitcoin may hence be understood as being fundamen- white paper outlines anonymity and decentralization as influential
tally related to opportunities enabled from these differ- factors shaping the evolution of the blockchain sector (Iansiti and
Lakhani 2017). While the developmental motives of the blockchain
entiating technologies; making it an attractive substitute
were anonymity and decentralization, the practical outcome can more
to traditional currencies for groups of users (Athey et al. exactly be described as it providing pseudonymity (Böhme et al. 2015)
2016; Ermakova et al. 2017). Opportunities arising due and disintermediation, not anonymity and decentralization. Eyal and
to the differentiating technology underlying Bitcoin Sirer (2014) state that “At this point, the currency is not decentralized
as originally envisioned.” In reality, blockchain has helped create a
stem from its ability to remove the need for a trusted disintermediated payment system, as well as a disintermediated method
third party or disintermediation, and due to its of raising venture financing via ICOs.
Global drivers of cryptocurrency infrastructure adoption 357

system. As an investment it increases opportunities for 2.2.1 Inflation

portfolio risk management owing to its very limited
correlation with other asset returns such as fiat curren- In its capacity as a global currency which is not tied to any
cies, stocks, bonds and commodities such as gold (Baur particular economy, bitcoin has the potential to act as a
et al. 2018a; Dyhrberg 2016) and hedging against eco- hedging opportunity against country-specific risk. In par-
nomic uncertainty (Bouri et al. 2017a; 2017b). ticular, buying bitcoins offers a novel opportunity to hedge
While previous academic studies have attributed less against (very) high inflation, in parallel to how gold and
(e.g., Blau 2018) or more weight (e.g., Baur et al. other assets have been known to function in the past
2018b) to speculative trading as the primary use of (Arnold and Auer 2015). Luther (2016) finds that currency
Bitcoin, it is clear that speculation is a factor behind transitions have often occurred during episodes of hyper-
the rising number of individuals holding bitcoin. To the inflation, exemplified by many Bolivians and Peruvians
extent that the adoption of Bitcoin infrastructure is re- who switched to US dollars, perceived to be safer, during
lated to the stock of bitcoin holders, geographical vari- such national episodes in the 1980s. People affected by
ation in Bitcoin infrastructure adoption may be related to very high inflation may therefore become more actively
geographical differences in the intensity of speculative interested in holding and using bitcoin, and in supporting
trading. In this analysis, however, we are primarily Bitcoin as an alternative financial system.
concerned with analyzing support for Bitcoin infrastruc- Cryptocurrencies have indeed been touted by advo-
ture in relation to the cryptocurrency’s potential as com- cates as a means to a less crisis-prone financial system
plement or substitute to more fundamental functions of (Maurer et al. 2013) and as a counterweight to (hyper)-
the established financial system. This is since node inflation (Dierksmeier and Seele 2018). There are also
infrastructure are primarily set up to support bitcoins reports that countries experiencing high inflation have
as an exchange and payment system. Merchants’ adop- seen surges in interest in bitcoins. This is visible in the
tion of Bitcoin infrastructure may also be understood as example of Venezuela, where inflation soared, trust in
driven by support for Bitcoin, both in regard to the the national government policy and currency
merchants’ own views and in regard to the views of plummeted and interest in bitcoins increased, evidenced
customer groups that merchants would want to appeal by the popularity of bitcoin mining (Kliber et al. 2019).
to. Another noted example is Cyprus during its 2012–2013
In the remainder of this section, we describe three financial crisis (Subramanian and Chino 2015). We
sets of factors that may drive support for Bitcoin as a hence expect that high inflation levels or inflation crises
financial exchange/payment system. Bitcoin may may systematically affect the adoption of the two types
substitute for (real or perceived) failings of of Bitcoin infrastructure (bitnodes, bitcoin merchants).
established financial systems, due to opportunities
of disintermediation and decentralization technolo- Hypothesis 1: The occurrence of inflation crises is
gies. We first consider bitcoin as a substitute for fiat associated with increased adoption of Bitcoin
currencies in inflation crises. We subsequently con- infrastructure.
sider less acute failings of established financial sys-
tems, in discussing whether Bitcoin can be consid- 2.2.2 Banking market development and competition
ered a viable (potential) substitute to poorly function-
ing national banking markets. Finally, we consider a There is a potential for financial technologies to substi-
more ideological perspective of interest in Bitcoin as tute for deficient provision of traditional banking ser-
being driven by the spread of general attitudes of vices, as evidenced, e.g., by the use of mobile money
distrust in banks and the established financial system. accounts to transfer money in Sub-Saharan Africa
When discussing banking market development, we (Demirguc-Kunt et al. 2018). Digital currencies have
also acknowledge that financial development may also been hailed as a promising means to reach businesses
be expected to drive the interest in cryptocurrencies. If and people in remote and marginalized areas (Lagarde
this mechanisms dominates, support for Bitcoin and the 2018). Around the world, most existing payment sys-
growth of Bitcoin infrastructure will be stronger—not tems (e.g., credit and debit cards) rely on transactors to
weaker—in countries with well-developed banking hold bank accounts. It is conceivable that digital curren-
markets. cies could serve as a payment system of choice for the
358 E. Saiedi et al.

unbanked (i.e., those without bank accounts). In fact, an In summary, we suggest the following set of
extensive survey of the Bitcoin community finds that hypotheses:
serving the unbanked is one of the foremost purported
promises of this currency (Vigna and Casey 2015; see Hypothesis 2a: The lower the population of finan-
Chapter 8 entitled “Unbanked”). cially included adults, the greater the adoption of
Whereas payment of a minor fee for access to services of Bitcoin infrastructure.
an online trading platform is sufficient for a digital currency
transaction (Dwyer 2015), banking systems impose fees on Hypothesis 2b: The lower the level of competition in
businesses and individuals, in the form of depository and banking markets, the greater the adoption of Bitcoin
transactional fees. Philippon (2016) finds that the unit cost infrastructure.
of financial intermediation has remained consistently and
surprisingly expensive from 1886 to 2012. Conventional On the other hand, it is also possible that financial
transactions impose costs ranging from a small percentage development is a prerequisite for interest in
(e.g., 1–3% for credit card purchases) up to highs of one fifth cryptocurrencies. Lacking experience with traditional In-
of transferred amounts (e.g., international remittances; see ternet banking services, people may not be prepared to
Beck and Martínez pería 2011). deal in cryptocurrency. A lack of familiarity with finan-
By removing the intermediary, the development of cial intermediaries and their services may also lead to
Bitcoin has the potential to do away with these costs. little interest in exploring their alternatives. Such a view
Transaction fees for bitcoins are in the 0–1% range would suggest that digital infrastructure aimed at
(Kaskaloglu 2014). Bitcoin offers a welcome alternative disrupting banking may develop most strongly in envi-
when high transaction costs of traditional transactions ronments with the most well-developed banking markets.
either disincentives the transaction altogether or diminish Financial literacy and sophistication is a pre-requisite
its benefits (Dierksmeier and Seele 2018). The ease of for taking advantage of financial innovations (Campbell
exchange via cryptocurrencies can extend much-needed 2006) and engaging in complex financial products, such
liquidity to recipients of micropayments or loans in devel- as stock market investments (van Rooij et al. 2011) or
oping and underdeveloped countries, offering the world’s retirement planning (van Rooij et al. 2012). Financial
‘unbanked’ millions an unprecedented degree of conve- literacy has been shown to underlie financial inclusion
nience and security (Vigna and Casey 2015). Lacking and increase the use of financial services (Grohmann
access to a financial institution or the needed documenta- et al. 2018) and hence, it is likely that a high level of
tion to use one, such individuals have to rely on storing financial literacy and inclusion is required for use of
cash, endangering themselves and limiting them to transact complex financial innovations such as bitcoins. The use
with those within their physical reach (Dierksmeier and of electronic payments has been found to be associated
Seele 2018). As long as access to a mobile phone with with financial inclusion (c.f. Thakor 2019). Emerging
SMS technologies or Internet connection exists on any research points to complementarities between banking
device, a whole world of transaction and investment pos- markets and fintech development (Hornuf et al. 2018;
sibilities becomes available (Raymaekers 2015). Klus et al. 2019). For example, Gazel and
Besides the penetration of banking, underdeveloped Schwienbacher (2018) find locations with more bank
competition in the banking market may be another factor headquarters and financial competition attract more
driving interest in Bitcoin as a payment system. Low fintech clusters. Such a development can generate ex-
competition within the financial system is expected to be ternalities in the form of greater know-how of financial
associated with high rents and limited innovation by in- technologies, and thereby also interest in
cumbents, aggravating the frictions of traditional banking cryptocurrencies such as bitcoin. It is also possible that
services in terms of costs, service availability, and service in a more vibrant banking market, where competition
scope. Therefore, low competition may in principle drive drives banks to innovate, banking costumers become
consumers to adopt alternative financial technologies for less risk-averse towards trying new electronic services.
reasons parallel to those developed above. Moreover, low Together, these arguments suggest that interest in
competition in the banking market stimulates investment bitcoin may be highest in countries with well-
in fintech (Thakor 2019). Interest in Bitcoin may follow in developed banking systems and high banking market
the wake of such activity. competition.
Global drivers of cryptocurrency infrastructure adoption 359

We consider the following two competing While to the best of our knowledge no systematic
hypotheses: investigation has been done regarding distrust to
banks and financial institutions as a driver of support
Hypothesis 2c: The greater the population of finan- for Bitcoin, the role of trust to banks has started to be
cially-included adults, the greater the adoption of explored in other fintech settings. Saiedi et al. (2017)
Bitcoin infrastructure. and Bertsch et al. (2017), show that a decline in trust
in banks and financial institutions increases partici-
Hypothesis 2d: The higher the level of competition in pation respectively, by lenders and borrowers, in
banking markets, the greater the adoption of Bitcoin online peer-to-peer (P2P) lending markets. Theoreti-
infrastructure. cal papers modeling trust-driven substitutions be-
tween intermediaries and fintech firms are also
emerging (e.g., Thakor and Merton 2018). We thus
2.2.3 Trust and social attitudes posit that lower trust in banks and the financial sys-
tem underlies the emergence and adoption of
A prerequisite for economic exchange, trust has been bitcoins. In our variable definitions, we use distrust
found to be positively associated with financial devel- to mean lack of trust for brevity.
opment and investments. Researchers have explored its Yermack (2015) reports that one of the major
role in online markets such as e-commerce (Ba and obstacles facing bitcoin in becoming an adopted cur-
Pavlou 2002), peer-to-peer lending (Duarte et al. rency is its extreme time-series volatility. This high
2012), or crowdfunding (Liang et al. 2019; Rau 2017). volatility is a product of the highly speculative nature
While the blockchain technology may be perceived as of the bitcoin market (Baek and Elbeck 2015).
reducing the need for direct trust in individuals, many Yermack (2015) further reports that it is widely un-
important uses of bitcoin would seem to be positively derstood that most bitcoin transfers involve transac-
associated with general trust. We therefore expect higher tions between speculative investors. Empirical stud-
levels of trust in others to increase interest in Bitcoin, ies corroborate the use of bitcoins as a speculative
e.g., in the form of increased trade in bitcoins for invest- investment (Baur et al. 2018b; Bouoiyour and Selmi
ment, speculation and online or in-store commerce. 2015) and that it exhibits speculative bubbles (Cheah
Cohen (2017) documents that bitcoins emerged as and Fry 2015; Fry 2018). It thus seems reasonable to
part of the 99% movement—initiated by the Occupy assume that the interest to buy and hold bitcoins for
Wall Street protests—and frustrations with banks that speculative purposes, and the willingness to accept
had become too big to fail.7 The launch of the Bitcoin risk when holding bitcoin to be used for transactions,
system was embedded in idealistic notions of providing would require a certain level of risk-willingness. Re-
means to replace existing financial structures, and nur- gional differences in average willingness to take fi-
turing an alternative monetary and financial system that nancial risks may therefore affect the use of bitcoin.
enables greater anonymity, privacy, and autonomy We expect that the higher the willingness to take risk
(Bashir et al. 2016; Böhme et al. 2015; Dodd 2018; in a region, the greater the adoption of bitcoins for
Shiller 2019). Via enabling cheap and automated verifi- speculative trading.
cation on distributed ledgers, the blockchain technology In summary, we expect that active support for Bitcoin
underlying cryptocurrencies allows for trust in an inter- is associated with a lack of trust in the established
mediary to be replaced by trust in the devised code and financial system. Such support would be manifested,
rules that define how the network reaches consensus e.g., through the running of Bitcoin nodes, and may also
(Goldfarb and Tucker 2019). Desire for this replacement affect merchants’ willingness to accept bitcoin. Greater
is likely greater where trust in financial intermediaries, trust in other people and higher risk-willingness is ex-
whom traditionally verified payments, is lower, and can pected to facilitate the use of bitcoin, and thereby the
thus motivate the uptake of cryptocurrencies. interest in adoption of Bitcoin infrastructure in a
region—in particular in regard to merchants’ adoption
7
Indeed, in the very first bitcoin mined, its founder, Satoshi Nakamoto, of bitcoin payments.
embedded the message “The Times 03/Jan/2009 Chancellor on brink
of second bailout for banks,” referring to the Times of London’s same- Hypothesis 3a: The greater the level of trust in others,
day headline (Elliott and Duncan 2009). the greater the adoption of Bitcoin infrastructure.
360 E. Saiedi et al.

Hypothesis 3b: The greater the level of distrust in means of payment, also allows greater personal security
banks and the financial system, the greater the adoption (threat of violence, etc.) than buying drugs directly from
of Bitcoin infrastructure. dealers or contacts (Barratt et al. 2016). Some sellers and
buyers of drugs are therefore actively supporting bitcoin
Hypothesis 3c: The higher the risk-willingness, the in view of its role in enabling on-line acquirement of
greater the adoption of Bitcoin infrastructure. drugs. It can also be expected that the use of bitcoin for
drug purchases may spill over to other types of use, and
2.3 Bitcoin as a facilitator of illicit activity thereby stimulate the general interest and familiarity
with bitcoin as a means of payment.
Bitcoin’s differentiating technology of pseudonymity Even though cryptocurrency markets create a poten-
makes the cryptocurrency interesting for actors pursuing tially global platform for drug distribution, they tend to
activities of illegal trade such as drug trafficking, be used for local trade as opposed to cross-border trade
weapons trade, and prescription drug trade (Foley (Demant et al. 2018). Hence, the adoption of Bitcoin
et al. 2019). Böhme et al. (2015) describe that the first infrastructure can be associated with the level of activity
notable adopters of bitcoin sought greater anonymity in on-line drug trade across locations. However, reliable
and a lack of regulation concerning what could be measures for such activities are hard to come by. We
purchased using the currency. Bohannon (2016) states suggest that the intensity of money-laundering activity
that bitcoin’s purported anonymity made it popular and the effectiveness of law enforcement – two factors
among criminals. A recent study finds that around that are being measured with some reliability for a
25% of all bitcoin users and 44% of all transactions sufficiently large number of countries for our purpose
relate to trading in illegal activities (Foley et al. 2019). – can be used as indicators for such demand.
The source of almost all bitcoins used for illicit pur- Money-laundering refers to processes whereby the
poses and laundered through exchange services are dark- proceeds from illicit trade are being transacted through
net marketplaces (Fanusie and Robinson 2018), i.e., mar- financial institutions so as to hide its origin in illicit
ketplaces on encrypted websites that do not appear activities. A substantial share of such activity is directly
through conventional Internet browsing. Bitcoin is the related to drug trade. While money laundering may take
primary means of payment on the dark-net (van Slobbe place across as well as within nations, locations with
2016) with estimates showing it accounts for nearly 95% high levels of trade in drugs would typically rate high on
of the transactions on some dark-net marketplaces (Smith measures of money-laundering activity.
2019). Moreover, trading in bitcoin is also advantageous The second factor that we consider, concerns to what
to sellers of illicit goods sold through such marketplaces. extent drug trade would require the pseudonymity pro-
By allowing for pseudonymous exchange, digital curren- vided by on-line purchasing using bitcoins. We focus on
cies such as bitcoin have been identified as ideal money the level of law enforcement in a country.
laundering vehicles (FATF 2014; NDIC 2008). More effective law enforcement may in principle
In view of these arguments, we expect the ceteris reduce bitcoin adoption for illicit intent. However, cur-
paribus interest in bitcoin to be higher in locations with rent research suggests that contemporary law enforce-
high activity in trade of illegally acquired goods and ment does not possess the capacity to control much
services, and in locations where financial services relat- Internet-related crime (Holt et al. 2015). In most coun-
ed to such trade take place. In particular, bitcoin has an tries, no particular Internet-mediated drug trafficking
important role to play in the trade of illegal drugs online investigations are carried out (Lavorgna 2014). Despite
(Soska and Christin 2015). Bitcoins and online drug law enforcement efforts, such as the closure of Silk
markets have made it possible to sell drugs to unknown Road and Silk Road 2.0, crypto-markets continue to
customers, whereas for decades retail drug markets proliferate and have proven to be extraordinarily resil-
remained localized, i.e., dealers sold primarily to known ient (Aldridge and Décary-Hétu 2016b; Soska and
customers (Aldridge and Décary-Hétu 2016a). Using Christin 2015). Drug buyers and sellers using the dark
cryptocurrencies to trade in drugs provides a means for net perceive the likelihood of arrest to be much lower
circumventing state prohibition. Furthermore, the op- than alternative means (Aldridge and Décary-Hétu
portunity to shift drug trade online, with anonymity 2016b; Barratt et al. 2016; Ormsby 2016; Van Hout
relatively well-preserved due to the use of bitcoins as and Bingham 2014).
Global drivers of cryptocurrency infrastructure adoption 361

We therefore would not expect that stricter law efforts becoming available to a broader audience (Albrecht
may hamper Internet-based drug trade, but it should be et al. 2019; van Wegberg et al. 2018).
expected to affect other forms of buying and selling Together, the arguments outlined above can be sum-
drugs (e.g., openly on the street). Conditional on the marized in the following two hypotheses:
level of demand for illegally acquired goods and ser-
vices, greater perceptions of the quality of police, and Hypothesis 4: The more money-laundering activities
courts could therefore drive illicit activity towards on- taking place, the greater the adoption of Bitcoin
line anonymous (or pseudonymous) activities.8 In other infrastructure.
words, we expect that the greater the likelihood of arrest
on the streets for traditional drug trafficking, the greater Hypothesis 5: The stronger the rule of law, the great-
the interest in selling drugs online using anonymity- er the adoption of Bitcoin infrastructure.
granting technologies. And as argued above, we expect
such interest to generate support for and interest in
Bitcoin. 3 Data
Therefore, we expect that the greater the enforcement
of rule of law, the greater the use of bitcoins for illicit 3.1 Dependent variables
activity (conditional on the level of activity of a local
drug market). We conclude that the level of drug trade We utilize two global digital infrastructure datasets
activity in a location can be expected to be related to the supporting the use of Bitcoins encompassing the five-
extent of money-laundering and to law enforcement year period from 2014 to 2018, one of computers that
effectiveness. On the basis of our expectation that inter- observably disseminate blockchain information, and
est in bitcoin is partially driven by on-line drug trading validate and verify transactions on the Bitcoin network
activity, we thus suggest that both money-laundering (known as Bitnodes; see Figure 1 for validation and
and law enforcement are related to the adoption of verification actions in a bitcoin transaction life cycle),
Bitcoin infrastructure. and one of merchants accepting bitcoins as payment.9
While we have emphasized above that the level of A Bitcoin node is any computer or data processor that
money-laundering constitutes a useful proxy for trades connects to the Bitcoin network, in order to receive and
in illicit goods in a country, it should be noted that some send information relating to Bitcoin’s blockchain (a
locations may feature as prominent sites for money growing list of records of bitcoin transactions). Each
laundering activity without necessarily being centers of bitcoin user represents a node on the network. A subset
illicit trade. Even where this may be the case (primarily of all nodes are Full Bitcoin nodes, which fully down-
in countries where financial regulation is conducive to load every block and transaction and check them against
money laundering), we expect money-laundering to be all rules of Bitcoin (e.g., no double-spending of
associated with interest in bitcoin. Consistent with this bitcoins). Any decentralized cryptocurrency depends
expectation, 97% of direct bitcoin payments from iden- on such full nodes to ensure its blockchain adheres to
tifiable criminal sources were received by exchanges in agreed-upon rules. Hence, Full Bitcoin nodes are essen-
countries with weak anti-money laundering regulations tial to ensure security of Bitcoin’s digital ecosystem and
(CipherTrace 2018). To the extent that they provide for full removal of a need for a trusted alternative node
anonymity, bitcoins and other cryptocurrencies can be or third party. A user becomes a Full Bitcoin node by
exploited for money-laundering purposes (Evans-Pughe running a software, called Bitcoin Core, on her comput-
et al. 2014; Moser et al. 2013), e.g., in the form of er. By default, Full Bitcoin nodes accept incoming con-
automated laundering operations (McGinn et al. 2016). nections (in technical terms, are reachable or listening)
With the rise of specialized bitcoin money laundering and upload updated blockchain information to other
services on the dark-net, such opportunities are peers or nodes on the network. These are termed
Bitnodes, which are reachable full nodes on the net-
work. Our data consists of all Bitnodes operating world-
8
In fact, our rule of law variable has a 0.49 correlation with dark-net wide between 2014 and 2018. While there is no reliable
market drug purchase statistics in 26 countries, reported by Winstock
9
et al. (2018). Available at an open data repository associated with this publication.
362 E. Saiedi et al.

Fig. 1 Validation and verification role of nodes in the Bitcoin transaction life cycle (see red boxes). Infographic modified from original by
Patrícia Estavão (printed with her permission under license CC BY-SA 4.0), available at patestevao.com/work/

method to determine if these Bitnodes are businesses, worldwide, as referred to in online Bitcoin forums, to find
individuals, exchanges, bitcoin miners, or merchants, businesses where bitcoin holders can spend them. Figure 3
there are indications that they are users or businesses depicts the global spread of merchants, whereas the
transacting in bitcoins or volunteers supporting the net- Appendix describes the Bitcoin merchants dataset we use
work (e.g., more nodes are online during US working and our dataset preparation in greater detail.
hours). The Appendix describes Bitcoin nodes, The accumulated stock of merchants recorded on
Bitnodes, and our dataset preparation in greater detail. CoinMap may not, however, be an entirely accurate repre-
Our dataset allows us to construct measures sentation of the situation in that year, as there are numerous
reflecting both the total intensity of Bitnode activity reported cases of merchants added to the database, who
(Bitnode intensity) per country or region and year and subsequently forego acceptance of bitcoins after months or
the average number of Unique bitnodes by country or years of low-demand (while there is no cost to offering this
region and year. Figure 2 provides a global depiction of service other than training cashiers/personnel). In view of
these variables. We also construct a measure of node this, we choose to focus on annual inflow to the database.
availability by calculating for how many hours there is Our main dependent variable constructed from this database
at least one Bitnode active in a country or region is hence the number of new bitcoin-accepting merchants in
(Bitnode hours). For further details, see Table 1. any given year in a country, or sub-country state, province or
Our data for merchants accepting payments in bitcoin is region. Our dataset does not suffer from survival bias, as
from CoinMap.org. CoinMap is a website where either even when merchants are removed from the map by users or
business owners or interested customers or users would the web admin, they are not dropped from our database.
add a business that accepts bitcoins (or bitcoins together Our dataset has the limitation that we only observe non-
with other cryptocurrencies) as payment in exchange for a hidden Full Bitcoin nodes and merchants. Bitnodes, being
service or good. Some of the businesses on these websites reachable full nodes, are a subset of Full Bitcoin nodes
only accept bitcoins on their online websites. Soon after keeping the network secure, as some nodes simply do not
launching in 2013, CoinMap became the primary website accept incoming connections, primarily due to installation
Global drivers of cryptocurrency infrastructure adoption 363

Fig. 2 a Global Map of Bitnode intensity (Monthly measure, Map of Unique bitnodes (Averaged for every hour over 2014–
averaged over 2014–2018; See Table 1 for variable definition). 2018; See Table 1 for variable definition). Top 15 countries are San
Top 15 countries are Monaco, Netherlands, Singapore, Lithuania, Marino, Austria, Sweden, Germany, Luxembourg, Croatia, Swit-
Luxembourg, Germany, Ireland, Switzerland, Sweden, Belize, zerland, Netherlands, Slovenia, Lithuania, Monaco, Australia,
Slovenia, Canada, Latvia, Finland, and Hong Kong. b. Global Bulgaria, Finland, and UK

of firewalls. Some estimates (e.g., Wang and Pustogarov cryptocurrency’s potential function as a facilitator of
2017) indicate as many as 6 to 8 times more nodes on the illicit activity.
network than our data includes. Similarly, Bitcoin mer-
chants do not include merchants who hide their presence 3.2.1 Financial characteristics
from CoinMap or only sell online. Hence, our study is
focused on digital infrastructure providers who are openly In constructing an indicator for inflation as a driver for use
and visibly supporting the Bitcoin network. of bitcoins, we utilize the concept of an Inflation Crisis,10
defined by Reinhart and Rogoff (2011) as depreciation in
3.2 Explanatory variables currency value of 20% per annum. For robustness tests, we
also define alternative variables measuring such phenom-
We construct a set of country-level variables related to ena called 15% Min Inflation and 25% Min Inflation with
Bitcoin’s adoption potential given existing technologi-
cal factors, its role as a potential substitute or comple- 10
It is notable that currency crashes and inflation crises go hand in
ment to established financial institutions or to the hand as shown by Reinhart and Rogoff (2011).
364 E. Saiedi et al.

Table 1 Variable description and data sources

Variable Description Source Source variable name

Panel A
Bitnode intensity An intensity measure of the number of Bitcoin node data are from Constructed from
active Bitnodes in a country multiplied bitnodes.com API-extracted data in JSON
by how many hours each was active in a format
month, standardized by the number of
hours in a month and averaged to
produce an annual measure (by taking
the mean of results for the first month of
each quarter). Standardized by dividing
it by the jurisdiction’s population of
Internet users in millions (see Internet
variable).
Unique bitnodes Number of unique Bitnodes in a country in Constructed from
a month. This measure is then made into API-extracted data in JSON
an annual measure, by averaging its format
value for the first month of each quarter
of a year. Standardized by dividing it by
jurisdiction’s population of Internet
users in millions.
Bitnode hours Number of hours in a month where at least Constructed from
one Bitnode from a country is active API-extracted data in JSON
format
Bitcoin merchants Number of new bitcoin merchants in a Bitcoin Merchant data are Constructed from
country in a year, as per timing it was from CoinMap.com API-extracted data
added on CoinMap’s map
Bitnode intensitypc Similar to Bitnode intensity, but Bitnodes.com
standardized by population in millions
Unique bitnodespc Similar to Unique bitnode, but Bitnodes.com
standardized by population in millions
Bitcoin merchantspc Similar to Bitnode Merchants, but CoinMap.com
standardized by population in millions
GDP per capita Log of annual GDP per capita in a country. World Bank WDI (2018) NY.GDP.MKTP.CD
GDP is at purchaser’s prices converted
into current US dollars
Restrictive regulation An indicator variable equal to one in years Hand-collected data from Law
in which hostile or contentious Library of Congress (2018),
regulation against the use of bitcoins is bitlegal.io, and Wikipedia
issued. Hostile regulation towards
bitcoin consists of regulatory authorities
imposing full prohibition of its use, or
partial prohibitions such as barring
financial institutions from dealing with
it. Contentious regulation towards
bitcoin consists of some legal
restrictions against use of bitcoin (incl.
Imposition of cumbersome compliance
requirements) or warnings against
bitcoin use by regulatory authorities
going beyond discouragement (incl.
Statements that bitcoin transactions may
cause violation of anti-money
laundering or terrorist financing rules).
Variable equals zero for countries that
have no regulatory framework or a
favorable regulatory framework for
bitcoins.
Global drivers of cryptocurrency infrastructure adoption 365

Table 1 (continued)

Variable Description Source Source variable name

Population Total population in a country in millions. World Bank WDI (2018) SP.POP.TOTL
Mid-year estimates.
Internet penetration Percentage of inhabitants using the Internet International
in a year Telecommunications Unit
Broadband penetration Percentage of inhabitants with fixed (ITU) (2018)
(wired)-broadband subscription in a
year
ICT market development A composite index of a nation’s Measuring the Information ID (ICT Development Index)
development in ICT. The index includes Society Reports, ITU
three aspects of digitalization: ICT (2015–2018)
capability (skills and knowledge),
access to ICT infrastructure, and use of
intensity of ICT.
Latest technology National average of response to “In your World Economic Forum, EOSQ067
country, to what extent are the latest Global Competitiveness
technologies available? [1 = not Report, Executive Opinion
available; 7 = widely available]” Surveys (2014–2018)
Internet servers Secure Internet servers per million people World Bank and https://www. IT.NET.SECR.P6
using encryption technology in Internet netcraft.com
transactions
Mobile subscriptions The variable measures the number of World
mobile telephone subscriptions per 100 Telecommunication/ICT
adults in the population. A subscription Development report and
refers to a public mobile telephone database
service that provides access to the
public-switched phone network using
cellular technology, including the count
of pre-paid SIM cards active in the final
three months of the year.
Electricity cost Electricity cost is the total median cost in World Bank Doing Business
percentage of income per capita Reports (2014–2018)
associated with completing procedures
to connect a warehouse to electricity.
Inflation crisis An indicator variable equal to one if in a Raw Data are from
country the annual decline in the International Monetary
average Consumer Price Index is greater Fund (IMF)‘s World
than 20% following Reinhart and Economic Outlook (2018)
Rogoff (2011)
Inflation The average annual change in Consumer
Price Index as a ratio.
Unbanked % aged 15+ without account ownership at WB Global Findex Database FX.OWN.TOTL.ZS
a financial institution and without a
mobile-money-service provider
Internet banking % aged 15+ who have used the Internet or Fin5.d.1
a mobile phone to access an account
Five-bank asset concentration Assets of five largest banks as a share of all World Bank’s Global GFDD.OI.06
banking assets. These include all Financial Development
earning assets, cash and due from banks, Index (GFDI 2018); Raw
foreclosed real estate, fixed assets, Data are from Bankscope
goodwill, other intangibles, current tax and Orbis Bank Focus,
assets, deferred tax, discontinued Bureau van Dijk
operations and other assets.
Three-bank asset Assets of three largest commercial banks GFDD.OI.01
concentration as a share of all commercial banking
assets.
Rule of law Captures perceptions of the extent to which RQ.EST
agents have confidence in and abide by
366 E. Saiedi et al.

Table 1 (continued)

Variable Description Source Source variable name

the rules of society, and in particular, the World Bank Worldwide

quality of contract enforcement, Governance Indicators
property rights, the police, and the (2018)
courts, as well as the likelihood of crime
and violence. This variable ranges from
approximately − 2.5 (weak) to 2.5
(strong) values for rule of law.
Organized crime National average of response to “In your World Economic Forum, EOSQ055
country, to what extent does organized Global Competitiveness
crime (mafia-oriented racketeering, Report, Executive Opinion
extortion) impose costs on businesses? Surveys (2014–2018)
[1 = not at all; 7 = to a great extent]”
Crime and violence costs National average of response to “In your EOSQ034
country, to what extent does organized
crime (mafia-oriented racketeering,
extortion) impose costs on businesses?
[1 = to a great extent—imposes huge
costs; 7 = not at all—imposes no costs]”
Money laundering Countries identified as heavily engaged in US Dept. of State Bureau for
“currency transactions involving International Narcotics and
significant amounts of proceeds from Law Enforcement Affairs
international narcotics trafficking.” (2018)
Panel B
Bitnode intensity An intensity measure of the number of Bitcoin node data are from Constructed from
active bitnodes in a region multiplied by bitnodes.com API-extracted data
how many hours each was active in a
month, divided by the number of hours
in a month and averaged to produce an
annual measure (by taking the mean of
results for the first month of each
quarter). Standardized by dividing it by
the jurisdiction’s population of Internet
users in millions (see Internet users
variable).
Unique bitnodes Number of unique bitnodes in a statistical Constructed from
region in a month made into an annual API-extracted data
measure, by averaging its value for the
first month of each quarter of a year.
Standardized by dividing it by
jurisdiction’s population of Internet
users in millions.
Bitnode hours Number of hours in a month in which at Constructed from
least one bitnode from a region is active. API-extracted data
Bitcoin merchants Number of new bitcoin merchants in a Bitcoin Merchant data are Constructed from
statistical region in a year, as per timing from CoinMap.com API-extracted data
it was added on CoinMap’s map
High risk willingness A region’s average of an indicator variable EBRD (2016) Life in Q4.28 in LiTS III; J2 in NFCS
for each LiTS III’s survey respondent, Transition Societies (LiTS)
equal to one if response to this question, Survey III; US Data are
is above 8: “Please, rate your complemented with same
willingness to take risks, in general, on a question in FINRA
scale from 1 to 10, where 1 means that (2015)‘s National Financial
you are not willing to take risks at all, Capabilities Study in 2015
and 10 and means that you are very
much willing to take risks.”
Trust in others
Global drivers of cryptocurrency infrastructure adoption 367

Table 1 (continued)

Variable Description Source Source variable name

A region’s average of an index EBRD (2016) LiTS Survey Q4.03 in LiTS III; TRUST in
standardized to run from −1 to 1, in III; US Data are GSS
answering to “Generally speaking, complemented using the
would you say that most people can be same question’s response
trusted, or that you cannot be too careful using NORC (2016)‘s
in dealing with people? Please answer General Social Survey
on a scale of 1 to 5, where 1 means that (GSS) Geo-sensitive data
you have complete distrust and 5 means
that you have complete trust.”
Distrust in banks & financial A statistical region’s average of an index Q4.04j in LiTS III;
system standardized to run from −1 to 1, in CONFINAN in GSS
answering to “To what extent do you
trust banks and the financial system?
Please answer on a scale of 1 to 5, where
1 means that you have complete distrust
and 5 means that you have complete
trust.”
Distrust in other institutions A region’s average of indices standardized Q4.04b, e, f, h, I, m & n in
to run from −1 to 1, in answering to, LiTS III; Trust in the same
similar to that above, on the following institutions in GSS
institutions, other than banks and the
financial system: The
Government/Cabinet of Ministers, the
Parliament, Courts, the Military, The
police, Unions and Religious
Institutions.
Bitcoin price Annually-averaged price of Bitcoin in www.cryptodatadownload. BTC/USD
thousands of USD, from quotes on four com/; Accessed on
large Bitcoin exchanges of 2014–2018, September 3, 2018
namely Bitstamp, Kraken, Coinbase and
Gdax

corresponding alternative thresholds. In further robustness each country. For use in robustness testing, we also use
tests, we utilize a continuous measure of Inflation along Three-bank asset concentration, which is the asset con-
with the square root of Inflation. Inflation data is collected centration of the largest three banks in each country.
from the International Monetary Fund.
Data on the proportion of unbanked adults is collect- 3.2.2 Criminality characteristics
ed from the World Bank’s Financial Inclusion Index.
Following Gross et al. (2012), we define the unbanked The US’ Bureau of International Narcotics and Law
as individuals older than 15 years without a checking, Enforcement Affairs performs an annual assessment of
saving or money market account. A subset of those the risk of money laundering in 200 countries and juris-
holding bank accounts, are those with access to the dictions. The bureau identifies “major money laundering
Internet and mobile banking, which can serve as another countries” as those whose “financial institutions engage
proxy of financial inclusion in a digital age. From the in transactions involving significant amounts of proceeds
same source, we obtain an Internet banking variable for from international narcotics trafficking”. The definition
robustness of results.
0
We extract data on concentration in the banking in- We are not aware of any multinational social surveys which is carried
dustry from the World Bank Global Financial Develop- out annually, and even if such a survey existed, social characteristics
typically do not change much 1 year to another (e.g., see Hoffman et al.
ment Index. We primarily use Five-bank asset concen- 2009 for an explanation of how general trust changes over intermediate
tration, the asset concentration of the largest five banks in time horizons).
368 E. Saiedi et al.

of “financial institutions” employed by the bureau in the spread of Bitcoin infrastructure. It is therefore important
period we study includes digital currencies and “other that our modeling exercise takes the level of technolog-
value transfer systems.” We introduce an indicator vari- ical development into account. In particular, as an
able (money laundering) that takes the value one if a Internet-enabled technology, bitcoin adoption is only
country is included in the bureau’s list in a given year. feasible for users who have regular access to the Inter-
As an indicator for the level of law enforcement, we net. We are, however, reluctant to introduce separate
obtain from the Worldwide Governance Indicators of control variables for such factors in order to avoid
the World Bank a rule of law variable. This variable overfitting our models, and in order to avoid introducing
captures the likelihood of crime and violence, and the multicollinearity issues (the level of technological de-
extent to which agents abide by and have confidence in velopment across countries is well-known to co-vary
the rules of society, in particular the quality of police, with, e.g., economic and institutional variables
courts, contract enforcement and property rights. (Czernich et al. 2011; Greenstein and Spiller 1996)
and our main control variable of GDP per capita is
3.2.3 Social characteristics highly correlated (0.89) with Internet penetration in
our dataset). Instead, all our main dependent variables
We obtain social characteristic variables of general risk- measuring the penetration of Bitcoin infrastructure are
taking, general trust, and trust in banks and the financial normalized by Internet penetration, i.e., the percentage
system from the Life in Transition (LITS III) survey which of Internet users in a country.
was carried out in 34 countries from 2014 to 2016. While As a virtual currency with the potential to dis-
data used to create most of the independent variables rupt existing payment systems and possibly even
described previously was only available on the level of monetary systems (Böhme et al. 2015), bitcoin aims
countries, LITS data can be broken down to the (sub- to rival fiat currencies issued by central banks.
national) regional level, corresponding to states, provinces Barriers to adoption in each jurisdiction for bitcoins
or administrative regions in the 34 countries where this are largely enforced by regulation. Most regulation
survey was carried out. The LITS survey was, however, vis-à-vis bitcoins are issued by a country’s financial
only carried out once during the period that we investigate regulatory authorities (Law Library of Congress
in this paper,11 meaning that we have no time variation to 2018). As suggested by public choice theory
exploit for this analysis. In order to utilize as much of the (Stigler 1971), incumbents may lobby for regula-
variation in the data as possible, exploring the role of these tion to keep out competitors and create rents for
social characteristics is carried out at the regional rather themselves. To the extent that bitcoin rivals
than national level. This has the consequence that analyses government-issued currencies, regulatory barriers
investigating the role of social characteristics are restricted on bitcoins by governments can hamper their adop-
to regions in 34 countries (294 regions) predominantly in tion. We create a longitudinal indicator variable,
Europe, instead of our full set of countries, and these Restrictive regulation, equal to one in a country
variables are cross-sectional. and for all the years in which restrictive (i.e., con-
tentious or hostile) regulation is in effect in a coun-
3.3 Control variables try by its financial regulator on either transacting in
bitcoin or buying and selling goods or services in
We furthermore construct a number of control variables exchange for bitcoins. Hostile regulation towards
related to fundamental economic and legal prerequisites bitcoin consists of regulatory authorities imposing
for the growth of Bitcoin infrastructure. full prohibition of its use, or partial prohibitions
Participation in any financial system is strongly tied such as barring financial institutions from dealing
to the availability of economic resources. In our analysis with it. Contentious regulation towards bitcoin con-
of global adoption of Bitcoin infrastructure, variations in sists of some legal restrictions against use of
economic strength across countries and regions is there- bitcoin (incl. imposition of cumbersome compliance
fore a natural first factor to consider. We control for requirements) or warnings against bitcoin use by
GDP utilizing data from the World Bank. regulatory authorities going beyond discouragement
Alongside economic strength, we also expect tech- (incl. statements that bitcoin transactions may cause
nological development to be strongly related to the violation of anti-money laundering or terrorist
Global drivers of cryptocurrency infrastructure adoption 369

Table 2 Summary statistics for panels A and B

Variable name Obs. Mean Std. dev. Min Median Max

Panel A: Country-year level dataset (2014–-2018; 137 countries)

Dependent variables
Bitnode intensity 685 2.02 4.07 0.0 0.2 34.9
Unique bitnodes 685 15.12 23.45 0.0 3.6 123.4
Bitnode hours 685 431.46 334.30 0.0 608.8 750.0
Bitcoin merchants 685 1.32 5.05 0.0 0.2 114.0
Bitnode intensitypc 685 1.65 3.52 0.0 0.1 30.8
Unique bitnodespc 685 11.70 19.40 0.0 1.9 114.2
Bitcoin merchantspc 685 0.98 3.99 0.0 0.1 90.9
Control variables
GDP per capita 685 8.80 1.46 5.7 8.8 11.7
Restrictive regulation 685 0.11 0.32 0.0 0.0 1.0
Population 685 51.22 166.39 0.1 11.2 1394.1
Technological variables
Internet penetration 685 54.68 27.71 3.3 59.1 97.6
Broadband penetration 685 14.45 13.79 0.0 9.9 55.7
ICT market development 685 5.27 2.23 1.2 5.3 8.9
Latest technology 685 4.84 0.94 2.3 4.8 6.6
Internet servers 685 4.07 11.24 0.0 0.2 123.1
Mobile subscriptions 685 113.39 34.07 25.0 115.4 259.4
Financial variables
Inflation crisis 685 0.04 0.20 0.0 0.0 1.0
Inflation 685 0.05 0.07 -0.0 0.0 0.4
Unbanked 685 38.56 27.59 0.0 37.9 93.6
Internet banking 685 27.89 21.89 0.4 23.5 85.1
Five-bank asset concentration 685 78.84 16.19 27.5 80.9 100.0
Three-bank asset concentration 685 63.94 18.47 18.4 64.0 100.0
Criminality variables
Rule of law 685 0.06 0.99 −2.3 −0.1 2.1
Organized crime 665 4.77 1.02 1.5 4.8 6.9
Crime and violence costs 685 4.45 1.07 1.5 4.5 6.8
Money laundering 685 0.36 0.48 0.0 0.0 1.0
Alternative variables
Shadow economy 670 30.22 12.76 8.1 30.2 63.3
Taxation 580 25.82 12.49 −1.3 24.7 65.0
Tax burden 665 77.62 12.00 37.2 79.2 99.9
Tax haven 685 0.06 0.23 0.0 0.0 1.0
Stock market return 440 0.67 25.77 −72.1 −1.9 282.9
Crisis stock market return 460 −2.77 19.70 −51.7 −4.6 52.5
Bitcoin mining country 685 0.07 0.26 0.0 0.0 1.0
Bitcoin mining MWs 670 1.85 11.28 0.0 0.0 111.0
Electricity price 685 21.98 76.11 0.7 14.3 1038.0
Electricity cost 685 1329.8 3006.7 0.0 308.2 28,965.9
370 E. Saiedi et al.

Panel B: Region-year level dataset (2014–2018; 294 NUTS2 or equivalent regions in 34 countries)
Dependent variables
Bitnode intensity 1470 2.17 4.71 0.0 0.4 59.7
Unique bitnodes 1470 17.37 33.31 0.0 5.9 486.5
Bitcoin merchants 1470 2.03 6.90 0.0 0.0 109.7
Social variables
High risk willingness 1470 0.16 0.12 0.0 0.1 0.7
Trust to Others 1470 −0.05 0.26 −0.8 −0.1 1.0
Distrust in banks & financial system 1470 0.09 0.31 −0.9 0.1 1.0
Distrust in other institutions 1470 0.01 0.24 −0.9 0.0 0.8
Interaction variable
Bitcoin price 1470 2.76 3.21 0.3 0.6 8.5

financing rules). This is a longitudinal measure. global coverage, % of unbanked, was restricted due to
Information about regulation on bitcoin trade is the breadth of coverage of the World Bank’s Financial
hand collected from the US Library of Congress, Inclusion survey. We refer to the resulting database as
bitlegal.io, and Wikipedia. Panel A, for which summary statistics are available in
A list of all dependent and explanatory variables used Table 2. Having reduced our sample from 201 countries
in the analysis and their definitions are described in respectively for which we have Bitnode and CoinMap
Table 1, where Panel A documents our country-level data12 to 137 countries may introduce a sample selection
analysis and Panel B the regional-level analysis. Sum- bias. While this choice was driven by a need to have
mary statistics for all our variables are presented in values for all independent variables used in the analysis,
Table 2 and a correlation matrix is available in the it is justifiable as the excluded nations are smaller in size
Appendix (Table A.I). by orders of magnitude and often reliable statistics on
financial and criminal characteristics unavailable for
them.13
Our dependent variables take the value 0 for a
4 Estimation methodology
non-trivial share of our observations (i.e., for
countries and years in which no Bitcoin infrastruc-
4.1 Country-level analyses
ture is detected), and can thus be thought of as
censored variables. In our country-year analyses,
We construct a panel-data structured database of
we therefore run random-effects Tobit models
countries and years for adoption of non-hidden
lower-censored at zero with panel-data adjusted
Bitcoin nodes and Bitcoin merchants, as well as our
bootstrapped standard errors.14 We employ the fol-
independent variables. We limit our database to 2014
lowing model specification, where the subscripts i
to 2018, the period for which full Bitcoin node and
and t, refer to countries and years respectively:
merchant data are available. We regress our measures
of infrastructure adoption over two baseline variables 12
147 countries have been reported to have at least one Bitcoin
of GDP per capita and restrictive regulations— merchant, with 28 having only one. We have coded the countries
capturing fundamental economic and legal country- without reported merchants on CoinMap as having no merchants.
Given that the median GDP per capita of these countries is below
level characteristics—as well as our independent $1000, it is reasonable to assume that no merchants existed in these
variables. nations.
13
In our country-level analyses, we limit our analysis to In fact, the median population of the excluded 60 countries is less
than 0.3 million while for those included in the analyses it is 11.7
137 countries in which all our explanatory variables are
million.
available from their respective data sources for all years 14
This entails blocked bootstrapping with countries as blocks, for the
in our database. The variable with the most limited bootstrapped clustering.
Global drivers of cryptocurrency infrastructure adoption 371

Bitcoin Infrastructure Adoption Variableit and Central Asia, as well as Germany, Italy, and Greece.
We make adjustments to ensure our dependent variable
¼ β 0 þ β1  lnðGDPÞit þ β2
database regions match statistical regions from the LiTS III
 Restrictive regulationit þ β3 survey, and then calculate region-level dependent variables
of bitnode intensity, unique bitnodes and bitnode hours.
 Inflation crisisit þ β4  Unbankedit We complement this data with population statistics from
þ β5  Five bank asset concentrationit European or national statistics agencies for these 34 coun-
tries. Hence, we have effectively restricted our analysis
þ β6  Money launderingit from 137 countries in the country-year level database in
Panel A, to a subset of 34 countries in the region-year level
þ β7  Rule of lawit þ WRi þ ζ t þ WRi  ζ t
database in panel B (with 294 regions). Summary statistics
þ εit ð1Þ for Panel B are available in Table 2.
Our region-level analyses use the following random-
Our Bitcoin infrastructure adoption variables are bitnode effects Tobit regression model, where subscripts j and t,
intensity, unique bitnodes, bitnode hours and Bitcoin mer- refer respectively to sub-national administrative or sta-
chants. The world region fixed effects (WRi) and year fixed tistical regions and years:
effects (ζt), respectively control for systematic differences
across world regions,15 and for changes in the price of
bitcoin and in global macroeconomic conditions over time.
Bitcoin Infrastructure Adoption Variablejt
Our interactions of world region and year fixed effects
(WRi × ζt) absorb the effect of world region-specific shocks ¼ β 8 þ β9  lnðGDPÞit þ β10
or time trends in adoption of bitcoins. Given the country-
year level variation of our Panel A data, these controls are  Restrictive regulationit þ β11
stringent enough to tease out the effect of our country-level  High risk willingnessjt þ β12
variables. The limited variation of our country-level char-
acteristics during the five years of observation does not  Trust to othersjt þ β 13
allow for the inclusion of country-level fixed effects.16 We
 Distrust in banks and the financial Systemjt
cluster all regression standard errors at the country level. In
robustness tests, we cluster our results at the country and the þ β14  Distrust in other institutionsjt þ ci
year level, and results do not significantly change.
þ ζ t þ εit
ð2Þ
4.2 Regional level analysis
This equation involves our baseline characteristics of
Our region-year analyses are meant to investigate the logged GDP and restrictive regulation to control for the
effect of social variables such as general and institutional heterogeneity specific to the economic and legal condi-
trust, and risk willingness on the adoption of Bitcoin tions, with our social characteristic variables following.
infrastructure. Our social variables are derived from the Our region-level empirical models using eq. 2 control
Life in Transition Societies (LiTS III) Survey carried out for systematic differences across nations using country
between 2014 and 2016 of 51,000 households in 34 fixed effects (ci) and for macro-economic conditions
countries. To this end, we construct a database of averages using year fixed effects (ζt). Similar to Giannetti and
of social variables for the 294 statistical regions surveyed, Wang (2016), we control for distrust (lack of trust) in
which are primarily located in nations in Eastern Europe other institutions unrelated to banks and the financial
15 system, to separate the specific effect of distrust to banks
Defined using the World Bank’s division of nations into world
regions of North America, Europe & Central Asia, East Asia & Pacific, from any idiosyncratic relationship between general
Latin America & Caribbean, Middle East & North Africa, Sub-Saharan institutional sentiments and Bitcoin infrastructure adop-
Africa, and South Asia
16 tion. In additional tests, we interact high risk willingness
Upon their inclusion, many of our country-level variables would be
omitted, due to their limited time-series variation from 2014 to 2018 in with bitcoin price to examine how this social character-
variables such as Restrictive regulation. istic varies with price increases.
372 E. Saiedi et al.

Table 3 Financial and criminality drivers of Bitcoin nodes (Panel A; Country-year level)

(1) (2) (3) (4) (5) (6)

Dependent variables Bitnode intensity Unique bitnodes Bitnode hours

Financial characteristics
Inflation crisis 0.338 0.530+ 7.038** 50.642
(0.386) (0.306) (2.596) (65.208)
Unbanked − 0.006 0.008 − 0.189* − 2.309
(0.013) (0.009) (0.078) (1.421)
Five-bank asset concentration − 0.004 0.003 0.011 − 5.080**
(0.015) (0.013) (0.069) (1.676)
Criminality characteristics
Money laundering 1.302* 1.347* 6.916** 176.344***
(0.568) (0.604) (2.381) (53.490)
Rule of law 1.343** 1.445** 9.323*** 88.669+
(0.424) (0.474) (2.557) (45.858)
Baseline controls
GDP per capita 1.425*** 1.338*** 0.617 0.681 0.039 72.001*
(0.363) (0.378) (0.415) (0.457) (2.209) (34.734)
Restrictive regulation − 0.574+ − 0.581* − 0.553+ − 0.529 3.069* 32.565
(0.314) (0.291) (0.290) (0.324) (1.347) (39.446)
Constant − 11.455*** − 10.185** − 5.464 − 6.624 19.808 382.419
(3.123) (3.616) (3.573) (4.347) (20.597) (431.859)
Year FEs Yes Yes Yes Yes Yes Yes
World Region FEs Yes Yes Yes Yes Yes Yes
Year-World Region FEs Yes Yes Yes Yes Yes Yes
Panel Data Specifications Tobit Tobit Tobit Tobit Tobit Tobit
Observations 685 685 685 685 685 685
Number of Countries 137 137 137 137 137 137
Pseudo R-Squared 0.34 0.4 0.34 0.41 0.62 0.22
Log-likelihood − 1314.9 − 1314.6 − 1306.7 − 1306.1 − 2268.1 − 2386.8
RHO 0.74 0.74 0.72 0.72 0.54 0.62

All models are Tobit random effects models lower-censored at 0 and their associated pseudo R-squared are reported. Panel-data adjusted
bootstrapped standard errors, clustered at the country level, are reported in parentheses. The symbols ***, **, *, + mean that the reported
coefficients are statistically different from zero, respectively, at the 0.1, 1, 5, and 10% level. Model 6 is additionally upper-censored at 750
(upper limit of active bitnode hours in a month)

5 Estimation results frequency of bitnode activity in a country, relative to

the number of Internet users. These dependent variables
5.1 Financial and criminality drivers are standardized measures described in Table 1, namely
bitnode intensity, unique bitnodes, and bitnode hours. In
Our first set of analyses focus on associations between all regression models (1) to (4), our primary dependent
use and running of bitnodes or merchants and the crim- variable of bitnode intensity is regressed against a base-
inality and financial characteristics discussed previous- line of controls, GDP per capita and a dummy variable
ly. We first report results on bitnodes and subsequently for restrictive regulation being in effect. Models (2) and
on Bitcoin merchants. (3) add financial characteristics and criminality charac-
Table 3 reports results of our random effects regres- teristics, respectively, and model (4) expands the base-
sions for three dependent variables measuring the line estimation to include both sets of financial and
Global drivers of cryptocurrency infrastructure adoption 373

criminality characteristics. The magnitude and standard notion that Bitcoin infrastructure is being set up as
deviations of coefficients for either sets of variables do a substitute to the financial system in countries
not alter in a significant way and are robust to inclusion where financial inclusion and penetration is low,
of both. Models (5) and (6) respectively regress unique but rather that Bitcoin infrastructure emerge where
bitnodes and bitnode hours on the full set of country- the penetration of the financial system is greater.
level characteristics. The Internet Appendix Tables IA.I This result goes against our hypothesis 2a, in favor
to IA.III show the effect of each individual variable of our counter-hypothesis 2c.
being added separately to the baseline model, for each The coefficient on Five-bank asset concentration in
of our dependent variables. model 6 is negative and significant at 1%, indicating that
Model 1 is our baseline model. Fixed effects for greater concentration in the financial sector reduces the
years, world regions and their interactions together with intensity of use of Bitnodes. This result goes against our
our two basic controls account for a substantial fraction hypothesis 2b assertion, in favor of the opposing hy-
of the variation in Bitnode intensity, corresponding to a pothesis 2d. Together with the findings for unbanked
pseudo R2 of 0.34. As expected, more restrictive regu- populations, results point to the penetration of Bitcoin
lations vis-à-vis bitcoins are associated with below- infrastructure increasing in well-developed banking sys-
average total intensity of Bitnode activity (models 1 to tems (hypotheses 2c and 2d), where the general famil-
4), but with the occurrence of an above-average number iarity with online finance is greater and the related
of unique bitnodes. The effect of restrictive regulation is digital entrepreneurial ecosystem is more vibrant.
clearly driven by differences on the intensive margin Countries with high risk of drug-related money laun-
(i.e., how many hours a month Bitnodes are available) dering tend to exhibit greater Bitnode intensity of activity
rather than difference on the extensive margin (i.e., how (model 4, p < 0.05) and Bitnode adopters (model 5,
many unique Bitnodes are being run in a country). This p < 0.01) and provide significantly more active hours of
may indicate that while restrictive regulation reduces the supply of Bitcoin nodes (model 6, p < 0.001). This latter
actual use of bitcoins, primarily by making the exchange coefficient estimate for money laundering is significant at
of bitcoins to fiat currencies illegal and secondarily by the 1% level and suggests that countries that have a high-
forbidding the trade of cryptocurrencies, it does not risk of money laundering, all else equal, run Bitcoin
effectively suppress the general interest in Bitcoin as a nodes for ~ 7 days more in a month (= 176.3/24) than
phenomenon, and to contributing to its infrastructure.17 other countries. Our hypothesis 4 is therefore validated.
The occurrence of an inflation crisis is associated The variable rule of law pertains to society’s percep-
with an increase in the number of unique bitnodes in a tion of the likelihood of crime and violence, as well as
country (p < 0.01). The coefficient estimate indicates their confidence in the quality of the police, courts, con-
that the onset of an inflation crisis in a country is tract enforcement, and property rights in each country. It
associated with an uptake of 47% of the average and is strongly associated with increases in Bitnode use in-
30% of the standard deviation of the number of unique tensity (model 4, p < 0.05) and the number of unique
bitnodes in a country. We run robustness tests by alter- bitnodes (model 5, p < 0.001). These are in agreement
ing the inflation crisis cut-off to a minimum of 15% or with our hypothesis 5. We note, however, that our Rule of
25% of annual inflation and find similar results (see law measure could be thought of as having two distinct
Internet Appendix Table IA.IV). This speaks to bitcoins components: the dependability of law enforcement, as
being utilized for financial hedging where currency well as the likelihood of crime and violence. Our hypoth-
devaluations greatly erode the value of a national cur- esized relationship relates to the former of these aspects.
rency, consistent with hypothesis 1. We therefore undertake further analysis in order to ensure
While a greater fraction of unbanked in a country that our estimates on rule of law are not driven by the
is not found to be associated with the intensity of Bitcoin infrastructure penetration being greater in coun-
use of Bitnodes (models 2, 4), it is negatively asso- tries with more crime and violence.
ciated with the number of unique Bitnode users We use costs to businesses associated with crime and
(model 5, p < 0.05). This result goes against the violence, the sub-index crime and violence costs of the
World Economic Forum’s Global Competitiveness In-
17
Plausibly, this increase in the extensive margin may result from dex, as a proxy for the intensity of crime and violence
greater publicity of bitcoins following national restrictions. in nations. A related variable from the same source, the
374 E. Saiedi et al.

degree of Organized crime, is also investigated. Inserting financial and criminality characteristics to our baseline
these variables into our main models yields insignificant model (1). We expand the baseline estimation in model
estimates throughout. We interpret these results (reported (4) to include both sets of characteristics.
in Internet Appendix, Table IA.V) to indicate that the Model 4 seems to indicate that national financial
association between rule of law and our dependent vari- characteristics dominate criminality explanations in ac-
ables is, as argued in the corresponding hypothesis, driv- ceptance of bitcoins by merchants. Our coefficient esti-
en by country-level differences in the reliability and mates on the unbanked fraction (β = − 0.048) in model 4
effectiveness of police and the legal system. is statistically significant at the 10% level, suggesting
Table 4 presents the results of our analyses on deter- that adoption of bitcoins by merchants is more popular
minants of adoption of Bitcoin merchants globally on in locations where adults have bank accounts to a greater
the same countries and using the same Panel A database extent in agreement with hypothesis 2c rather than 2a.
on which our Table 3 results are based on. Similar to Together with the results in the same direction in Table 3
Table 3, models 2 and 3 progressively add national and contrary to a widely purported promise of bitcoins

Table 4 Financial and criminality drivers of Bitcoin merchants (Panel A; Country-year level)

(1) (2) (3) (4)

Dependent variable Bitcoin merchants

Financial characteristics
Inflation crisis − 0.418 0.639
(1.175) (1.807)
Unbanked − 0.059+ − 0.048+
(0.031) (0.029)
Five-bank asset concentration − 0.041 − 0.037
(0.032) (0.037)
Criminality characteristics
Money laundering 1.163* 0.761
(0.560) (0.867)
Rule of law 1.448* 1.303*
(0.712) (0.628)
Baseline characteristics
GDP per capita 1.401* 0.481 0.368 − 0.234
(0.552) (0.495) (0.531) (0.579)
Restrictive regulation − 0.020 − 0.667 0.111 − 0.388
(0.997) (0.922) (0.736) (0.908)
Constant − 12.041* 1.336 − 3.980 6.166
(5.301) (4.511) (4.916) (5.096)
Year FEs Yes Yes Yes Yes
World Region FEs Yes Yes Yes Yes
Year-World Region FEs Yes Yes Yes Yes
Panel Data Specification Tobit Tobit Tobit Tobit
Observations 685 685 685 685
Number of Countries 137 137 137 137
Log-likelihood − 1418.2 − 1415.8 − 1415.1 − 1413.7
RHO 0.23 0.23 0.22 0.22

All models are Tobit random effects models lower-censored at 0. Panel-data adjusted bootstrapped standard errors, clustered at the country
level, and are reported in parentheses. The symbols ***, **, *, + mean that the reported coefficients are statistically different from zero,
respectively, at the 0.1, 1, 5, and 10% level
Global drivers of cryptocurrency infrastructure adoption 375

(see, e.g., Vigna and Casey 2015), this points to a picture The coefficient estimate on rule of law is significant
of Bitcoin infrastructure serving the banked population (p < 0.05) in models (3) and (4) where we control for
more than the unbanked. It is noteworthy that our other covariates and money laundering is only statisti-
Bitcoin merchants variable is more concentrated in cer- cally significant in model (3), in agreement with our
tain countries, especially in the earlier years of the hypothesis 5.
sample. This has the effect that the broad and stringent
set of fixed effects interacting world regions and year 5.2 Social drivers
dummies, and our baseline characteristics, account for a
great deal of the variation in our data, reducing the Table 5 presents the results of our region-level
variation that can be attributed exclusively to our ex- analyses. This set of analysis uses regional-level
ploratory variables. versions of the dependent variables and baseline

Table 5 Social drivers of adoption of Bitcoin infrastructure (Panel B; Region-year level)

(1) (2) (3) (4) (5) (6)

Dependent variables Bitnode Unique Bitcoin Bitnode Unique Bitcoin
intensity bitnodes merchants intensity bitnodes merchants

Social characteristics
High risk willingness 4.848 27.595 17.267*** 6.309+ 38.307 20.057**
(3.421) − 23.031 (4.552) (3.400) (23.387) (6.603)
High risk willingness x Bitcoin − 0.581** − 4.928* − 1.824*
price (0.000) (0.002) (0.911)
Trust in others 1.804 17.246+ − 0.819 1.945 17.833+ 0.441
(1.229) − 9.43 (2.561) (1.186) (9.374) (2.713)
Distrust in banks & financial 4.383** 25.271* − 0.942 4.266** 24.855* 2.581
system (1.638) − 12.525 (3.210) (1.581) (12.449) (3.310)
Distrust in other institutions − 2.827 − 18.204 − 8.955 *
− 2.632 − 17.708 − 3.834
(2.311) − 16.9 (4.354) (2.248) (16.809) (4.351)
Bitcoin price 0.011 − 2.985*** 0.269
(0.038) (0.391) (0.169)
Baseline characteristics
GDP per capita − 1.886* − 33.543*** − 6.211 − 1.884* − 33.558*** − 6.341
(0.869) − 8.313 (4.548) (0.821) (8.239) (4.225)
Restrictive regulation − 0.139 3.468 − 0.441 0.049 5.212 0.240
(0.422) − 4.051 (2.254) (0.401) (4.036) (2.133)
Constant 8.381 239.257*** 35.253 8.424 239.675*** 32.577
(7.710) − 71.928 (38.691) (7.301) (71.287) (35.990)
Year FEs Yes Yes Yes Yes Yes Yes
Country FEs Yes Yes Yes Yes Yes Yes
Panel Data Specifications Tobit Tobit Tobit Tobit Tobit Tobit
Observations 1470 1470 1470 1470 1470 1470
Number of Statistical Regions 294 294 294 294 294 294
Number of Countries 34 34 34 34 34 34
Log-likelihood − 2800.7 − 5221.2 − 2742.9 − 2796.7 − 5218.4 − 2740.9
RHO 0.78 0.59 0.24 0.78 0.59 0.24

Notes: All models are Tobit random effects models, lower censored at 0 and their associated overall R-squared is reported. Their standard
errors are reported in parentheses. The symbols ***, **, *, + mean that the reported coefficients are statistically different from zero,
respectively, at the 0.1, 1, 5 and 10% level
376 E. Saiedi et al.

characteristics of Table 3, and in addition includes bitcoin. However, our measure of self-reported willing-
social variables and country-level fixed effects in ness to take risks by individuals is found to be signifi-
order to control for unobserved country-level het- cant at the 0.1% level consistent with our hypothesis 3c.
erogeneity. Country-level differences in financial This suggests that adoption of bitcoins by merchants
and criminality factors are absorbed by these fixed occurs more where the population have a higher tenden-
effects and are not included individually in this cy to engage in risky and speculative behavior.
analysis. Dependent variables are standardized by We conduct a further test regarding the hypothesis 3c.
the number of Internet users. Year fixed effects In developing this hypothesis, we argued that risk-
control for unobserved year-specific events. willingness can be expected to be associated with the
Models 1 and 2 present estimates of the model spec- penetration of Bitcoin infrastructure since the level risk-
ification in Eq. (2) for bitnode intensity and unique willingness could be expected to influence how many
bitnodes. In regressing the trust in others variable against people that are interested in holding bitcoin. If this is the
unique bitnodes, the coefficient estimate of model (1) is case, we should be able to observe signs of Countercycli-
insignificant. In model (2), the coefficient of β = 17.2 is cal Risk Aversion. Such behavior, which has been identi-
significant at the 10% level. In terms of magnitude, these fied as explaining behavior on the stock market (Cohn
suggest that a one standard deviation decrease in trust in et al. 2015; Guiso et al. 2018), would imply that when
others is associated with 4.5 fewer unique bitnode prices go down, generally risk-averse individuals are even
adopters per million Internet users or 31.3% of average less willing than before to take on risks. While risk-
adopters of bitnodes in statistical regions. We thus find willingness would generally be associated with greater
only weak and partial support for hypothesis 3a. participation in financial markets, this association would
The coefficient estimate for our variable distrust in hence in the presence of countercyclical risk aversion be
banks and the financial system is statistically significant strengthened in periods when assets trade at a relatively
in models (1) to (2), respectively at the 1% and 5% level. lower price. We test this prediction in models 4 to 6 which
In terms of magnitude, a one standard deviation increase use the same variables used respectively in models 1
in distrust in banks and the financial system equivalent through 3, but add an interaction term of the annually-
to 31% more distrustful people18 is associated with an averaged bitcoin price variable with high risk willingness.
increase of 8 unique bitnode adopters per million Inter- The estimates on these interaction terms are consistently
net users (equivalent to 54% of mean adopters in statis- and significantly negative. We interpret these findings as
tical regions), or an increase equivalent to 34% of the providing relatively strong support for hypothesis 3c, and
standard deviation of bitnode intensity. These effects are for the notion that there is greater interest to hold
fairly large, and corroborate the role of crisis-related bitcoins—both by merchants themselves and by their
factors and distrust in banks being a factor in the early costumers—in regions with higher levels of risk-taking.
use of other peer-to-peer fintech technologies (see, e.g.,
Saiedi et al. 2017; Thakor and Merton 2018), in agree- 5.3 Robustness tests
ment with hypothesis 3b.19
Model 3 documents region-level social determinants We consider a range of robustness tests to lend empirical
of the adoption of bitcoin by merchants. Our hypothe- support to our findings, with respect to (i) alternative
sized trust variables do not yield any statistically signif- definitions of variables utilized to test our hypotheses, (ii)
icant association to merchants’ rate of adoption of an alternative modeling approach, and (iii) alternative strat-
egies for choosing the reference sample (e.g., restricting
18
Equivalent to 115% of the time-wise increase in Americans’ distrust
analysis to geographies with non-zero infrastructure).
in banks seen from 2006 till 2010 (which was 26.7% based on the We investigate the construct validity of our financial
Gallup survey), the year this distrust peaked at the height of the variables by constructing alternative measures relating
financial crisis.
19 to the relevant hypotheses. These alternative variables
Given Bitcoin’s origins in the cypherpunk and crypto-anarchist
subcultures (Swartz 2018), it is possible that distrust in government are listed along with the original variables in Table 1.
may be have greater association with the adoption of bitcoin. We Each alternative measure is significantly correlated, at
explored this, by substituting for distrust in banks with a similarly- below 1% significance, with the original variable.
constructed distrust in government variable from the same sources, and
do not find any explanatory power to this latter variable in regressing The results of this exercise are shown in Table 6.
either of our Bitcoin infrastructure variables. Models 1 and 4 are provided for easy reference and are
Global drivers of cryptocurrency infrastructure adoption 377

Table 6 Robustness tests with alternative measures for financial variables (Panel A; Country-year level)

(1) (2) (3) (4) (5)

Dependent variables Unique bitnodes Bitnode hours

Inflation crisis 7.038** 6.204* 50.642 50.322

(2.596) (2.943) (65.208) (69.477)
Inflation − 22.106
(28.046)
Inflation2 146.589*
(69.948)
Unbanked − 0.189* − 0.177* − 2.309 − 2.373
(0.078) (0.074) (1.421) (1.588)
Internet banking 0.168*
(0.079)
Five-bank asset concentration 0.011 0.011 0.008 − 5.080**
(0.069) (0.063) (0.062) (1.676)
Three-bank asset concentration − 4.647***
(1.255)
GDP per capita 0.039 1.072 1.072 72.001* 71.732*
(2.209) (1.884) (1.884) (34.734) (33.292)
Restrictive regulation 3.069* 3.368* 3.368* 32.565 30.525
(1.347) (1.580) (1.580) (39.446) (46.289)
Money laundering 6.916** 6.939** 6.939** 176.344*** 180.962***
(2.381) (2.295) (2.295) (53.490) (38.249)
Rule of law 9.323*** 9.477*** 9.477*** 88.669+ 84.178*
(2.557) (2.534) (2.534) (45.858) (39.757)
Constant 19.808 - 0.718 − 0.718 382.419 292.711
(20.597) (14.874) (14.874) (431.859) (386.696)
Year and World Region FEs Yes Yes Yes Yes Yes
Year-World Region FEs Yes Yes Yes Yes Yes
Observations 685 685 685 685 685
Log-likelihood − 2268.1 − 2264.9 − 2269.5 − 2385.1 − 2264.9
Number of Countries 137 137 137 137 137
RHO 0.54 0.54 0.52 0.62 0.61

All models are Tobit random effects models lower-censored at 0 and their associated pseudo R-squared are reported. Panel-data adjusted
bootstrapped standard errors, clustered at the country level, are reported in parentheses. The symbols ***, **, *, + mean that the reported
coefficients are statistically different from zero, respectively, at the 0.1, 1, 5, and 10% level. Models 4 and 5 are additionally upper-censored
at 750 (upper limit of active Bitnode hours in a month)

duplicates of models 5 and 6 in Table 3. In model 2, a 5%. Moreover, inflation and inflation2 are jointly
linear and quadratic combination of the continuous significant at less than 0.1%. The coefficients imply
measure inflation are replacing the indicator variable that on average, starting at inflation values of 15.1%
inflation crisis. A non-linear specification is used increases in inflation contribute to broadening the
since our Hypothesis 1 states that it is very high population of bitcoin adopters. In further work, re-
inflation levels which are expected to be associated ported in Table IA.IV of the Internet Appendix, we
with greater uptake of Bitcoin infrastructure. In confirm that using indicator variables for > 15% or >
agreement with such hypothesizing, the estimate on 25% inflation, instead of the > 20% indicator used in
the quadratic component is positive and significant at our main analysis yields qualitatively similar results.
378 E. Saiedi et al.

In model 3, we replace unbanked for the proportion of Bitnodes and merchants in parallel, our separate econo-
those with Internet banking. We find a positive and signif- metric models for these two could be under-utilizing our
icant coefficient on this variable, which as before supports related dependent variables. Since there may be unob-
our hypothesis 2c that greater financial inclusion leads to served factors (e.g., the number of bitcoin users in a
greater adoption of Bitcoin infrastructure. In model 5 we country) which are related to both the intensity of
replace five-bank asset concentration with another measure bitnode operations and the number of Bitcoin mer-
for the concentration in the banking industry of three-bank chants, the error terms (or unaccounted part) of each
asset concentration. Also, with this alternative measure, we model could be expected to be positively correlated.
find support for our hypothesis 2d in that greater compe- Running these two sets of analysis as a system is there-
tition in the banking sector is associated with greater fore of interest. We implement a generalized simulta-
uptake of Bitcoin infrastructure. neous equation model (GSEM). Table 7 reports results
Next, we consider an alternative modeling approach. of a Tobit model fitted with GSEM for three possible
While we have presented our two sets of analyses of pairings of our three bitnode dependent variables with

Table 7 Robustness tests using generalized simultaneous equation modeling (Panel A; country-year level)

(1) (2) (3)

Dependent variables Bitnode Bitcoin Unique Bitcoin Bitnode Bitcoin

intensity merchants bitnodes merchants hours merchants

Financial characteristics
Inflation crisis 1.534+ 1.149 11.008** 1.149 87.005 1.149
(0.852) (1.120) (3.806) (1.120) (77.197) (1.120)
Unbanked 0.010 − 0.049 − 0.129* − 0.049 − 1.585 − 0.049
(0.011) (0.032) (0.061) (0.032) (1.349) (0.032)
Five-bank asset 0.011 − 0.038 0.010 − 0.038 − 5.084*** − 0.038
concentration
(0.018) (0.036) (0.068) (0.036) (1.244) (0.036)
Criminality characteristics
Money laundering 1.481* 0.594 6.942** 0.594 111.079** 0.594
(0.635) (0.653) (2.454) (0.653) (37.625) (0.653)
Rule of law 1.937*** 1.173* 9.693*** 1.173* 73.946* 1.173*
(0.457) (0.566) (1.942) (0.566) (34.603) (0.566)
Baseline characteristics
GDP per capita 0.552+ − 0.162 1.147 − 0.162 43.281 − 0.162
(0.329) (0.504) (1.514) (0.504) (31.870) (0.504)
Restrictive regulation − 0.122 − 0.692 0.024 − 0.692 19.883 − 0.692
(0.404) (0.772) (1.781) (0.772) (49.044) (0.772)
Constant − 6.473* 5.854 7.629 5.854 493.289 5.854
(3.257) (5.702) (15.480) (5.702) (318.059) (5.702)
Year FEs Yes Yes Yes Yes Yes Yes
World Region FEs Yes Yes Yes Yes Yes Yes
Year-World Region FEs Yes Yes Yes Yes Yes Yes
Observations 685 685 685
Log-likelihood − 2960.1 − 3814.1 − 5410.2
Number of Countries 137 137 137

All models are Generalized Simultaneous Equation Models fitted with Tobit models lower-censored at 0 and their associated log-likelihoods
are reported. Standard errors, clustered at the country level, are reported in parentheses. The symbols ***, **, *, + mean that the reported
coefficients are statistically different from zero, respectively, at the 0.1, 1, 5 and 10% level
Global drivers of cryptocurrency infrastructure adoption 379

our Bitcoin merchant variable. Our results are highly first order tests, we explore measures accounting for more
similar to those found using single equation models advanced types of Internet use. Beyond the previously
(models 4 to 6 in Table 3; model 4 in Table 4). In model used Internet penetration, we also extract measures of
1, rule of law is more significant in explaining bitnode broadband penetration and mobile telephone subscriptions
intensity at 0.1% with a greater coefficient magnitude, from the International Telecommunications Unit (ITU)’s
while other variables retain their significance. Mean- World Telecommunication/ICT Development reports, re-
while, rule of law is similarly significant in explaining spectively defined as percentage of households with access
Bitcoin merchants. Results for unique bitnodes and to broadband, and the number of mobile telephone trans-
bitnode hours in models 2 and 3 are very similar to our actions per 100 adults in the population. We obtain ICT
previous findings in our main analysis in Table 3 and capabilities and ICT market development from the ITU’s
confirm our previous findings. Measuring the Information Society Reports; the former
Finally, we repeat our main analyses excluding coun- measures user sophistication as a composite index of adult
tries in which no bitnode activity (7 countries) or Bitcoin literacy, and secondary and tertiary education, and the latter
merchants (24 countries) exist during our sample period. measures Information and Communication Technology
Tables IA.VII and IA.VIII of the Internet Appendix, (ICT) development as a composite index of ICT infrastruc-
present these results respectively for our financial and ture, ICT capabilities (skills and knowledge) and ICT use
criminality variables analysis (Panel A; Country-Years) of intensity. We measure the extent to which the latest
and for our social variables analysis (Panel B; Region- technology is available, using the Executive Opinion Sur-
Year). This is to investigate to what extent our results are vey of the World Economic Forum’s Global Competitive-
driven by countries that do not have any Bitcoin infra- ness Report. We obtain Internet servers data from the
structure at any time of observation. Our coefficient World Bank and netcraft.com, as the number of servers
magnitudes are very similar and statistical significances using encryption technology in Internet communication
remain largely unchanged from our main findings pre- per one million people. We utilize electricity cost and
sented earlier. electricity price data from the World Bank’s Doing
Business datasets, respectively measuring the median
5.4 Additional analysis cost in percent terms relative to income per capita of
connecting and running a warehouse with electricity, and
In this section, we perform additional analysis in order the price in US cents per kWH of monthly electricity
to set our main analysis into context. Specifically, we consumption for a warehouse based in the largest
investigate a set of factors beyond the main variables business city of a country for the month of March.
and controls used in the main model, which may drive Table 8 shows the effect of these variables on per-capita
the diffusion of Bitcoin infrastructure. In a first set of standardized dependent variables. Model 1 shows, for
analysis, we explore a wider set of measures of techno- reference, the relationship between GDP per capita and
logical development. Thereafter, we seek to rule out the bitnode intensity. In all other models, we do not include
following alternative explanations: (i) shadow economy GDP per capita due to very high correlations—as high as
and tax evasion, (ii) rewards in cryptocurrency mining 0.91—with the technology variables (see Appendix,
of bitcoins, and (iii) hedging of equity markets. Table A.I). Models 2 and 3 respectively show that the
Internet and broadband penetration are significant (at
5.4.1 Technological/infrastructural characteristics 0.1% and 5% respectively) in explaining bitnode intensity.
Models 4 through to 7 examine the effect of each explan-
In our main models, we account for the role of techno- atory variable in Eq. (1) independently on bitnode intensity.
logical development by normalizing the dependent var- ICT market development and latest technology are highly
iables by the number of Internet users. In further work, significant (at 0.1%), while secure Internet servers and
we construct a range of variables to explore further mobile subscribers are significant at 10% and 5% respec-
nuances in the relationship between technological de- tively. While not reported, we also tested ICT capabilities
velopment and the penetration of Bitcoin infrastructure. separately which was highly significant at 0.1%. Models 9
Since Information and Communication Technology through to 12 represent the results of Eq. (1), by progres-
(ICT) is an enabler and prerequisite for financial technol- sively regressing all infrastructural variables onto our three
ogies (Haddad and Hornuf 2019; Huang et al. 2019), as per-capita dependent variables. ICT market development
 380

Table 8 Technological/infrastructural drivers of Bitcoin nodes (Panel A; country-year level)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Dependent variables Bitnode intensitypc Unique bitnodespc Bitnode hours Bitcoin merchantspc

Internet penetration 0.044***

(0.013)
Broadband penetration 0.080*
(0.035)
ICT market development 0.613*** 0.43*** 6.393*** 103.41*** 0.487*
(0.173) (0.100) (0.925) (23.450) (0.224)
Latest technology 1.344*** 0.628* 2.540 63.585+ 0.744+
(0.339) (0.271) (1.740) (37.540) (0.404)
+ +
Internet servers 0.079 0.079 − 0.011 − 4.988 0.017
(0.046) (0.041) (0.137) (13.35) (0.045)
Mobile subscriptions 0.017* 0.006 0.004 − 0.374 − 0.002
(0.008) (0.007) (0.045) (0.849) (0.008)
GDP per capita 1.333***
(0.352)
Restrictive regulation − 0.652+ − 0.577 − 0.520+ − 0.653* − 0.160 − 0.558+ − 0.126 3.610*** 24.854 0.009
(0.337) (0.382) (0.314) (0.306) (0.189) (0.303) (0.255) (1.069) (38.522) (0.564)
Constant − 11.0*** − 1.131+ − 0.025 − 2.31* − 5.8*** 1.080* − 0.936 − 5.34** − 26.59** − 84.64 − 5.921*
(3.127) (0.603) (0.692) (0.926) (1.553) (0.534) (1.200) (1.325) (9.417) (244.682) (2.665)
Year FEs Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
World Region FEs Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Year-World Region FEs Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Panel Data Specification Tobit Tobit Tobit Tobit Tobit Tobit Tobit Tobit Tobit Tobit Tobit
Observations 685 685 685 685 685 685 685 685 685 685 685
Number of Countries 137 137 137 137 137 137 137 137 137 137 137
Pseudo R-Squared 0.37 0.31 0.35 0.33 0.36 0.40 0.24 0.48 <0 <0 <0
Log-likelihood − 1235.1 − 1246.3 − 1247.2 − 1243.3 − 1239.8 − 1208.4 − 1249.1 − 1190.6 − 2248.6 − 2351.5 − 1404.0
RHO 0.72 0.76 0.75 0.75 0.73 0.77 0.79 0.71 0.60 0.70 0.23

All models are Tobit random effects models lower-censored at 0 and their associated pseudo R-squared are reported. Panel-data adjusted bootstrapped standard errors, clustered at the country
level, are reported in parentheses. The symbols ***, **, *, + mean that the reported coefficients are statistically different from zero, respectively, at the 0.1, 1, 5 and 10% level. Model 10 is
additionally upper-censored at 750 (upper limit of active Bitnode hours in a month)
E. Saiedi et al.
Global drivers of cryptocurrency infrastructure adoption 381

and latest technology remain positive and significance in It is possible that use of Bitcoin infrastructures (esp.
most regressions at 0.1% and below 10% respectively. This Bitcoin nodes) are primarily being driven by
aligns with the expectation that Bitcoin infrastructure are cryptocurrency miners vying to reap the financial re-
more likely to appear in countries where ICT is more wards associated with mining bitcoins. Hence, we ex-
advanced. Still, analysis of the pseudo-R2s suggest that plore if bitcoin mining is a major driver of adoption of
economic strength (GDP per capita) is a more efficient our variables, by regressing a bitcoin mining country
predictor of Bitcoin infrastructure penetration than ICT indicator variable for countries identified by the Cam-
market development. This supports our choice to anchor bridge Centre for Alternative Finance’s two versions of
our main analysis in models controlling for this factor. the Global Cryptocurrency Benchmarking Study
(Hileman and Rauchs 2017; Rauchs et al. 2018) as
having medium-to-large scale mining operations, and
5.4.2 Alternative explanations for bitcoin adoption additionally find no significant effect in our robustness
tests (Tables 9 and 10, Model 5). We include estimates
We additionally test a range of other variables argued for of mining intensity (in MW) and do not find any signif-
in the emerging literature to drive the use of icant effect in our robustness tests (Tables 9 and 10,
cryptocurrencies.20 Since they enable global transac- Model 6). We attribute this to operation of Bitcoin nodes
tions, digital assets can be used for shadow banking being driven by other than purely financial incentives
and tax evasion (Van Alstyne 2014). Currently, many and rewards to mine bitcoins, while this may still be a
national governments and their tax administrations are contributing factor. Bitcoin mining facilities are highly
in the process of penning legislation on assets held in concentrated in certain countries with cheap and reliable
cryptocurrencies, while international task forces are de- electricity, cold climate and friendly regulatory environ-
vised to combat cryptocurrency-enabled tax crime (J5 ment (e.g., Rauchs et al. 2018 identify only 128 mining
2018). Most countries, however, have still not imposed facilities worldwide, where only 7 countries are mining
any regulation or indeed effective regulation (Akins more than 40 MW whereas 1.7GW have been identified
et al. 2015; Marian 2013). Hence, it is conceivable that worldwide), whereas nodes are less concentrated. In
bitcoins and Bitcoin infrastructure would find the fact, Miller et al. (2018) find only 2% of nodes on the
greatest use and support where incentives for tax avoid- Bitcoin network to account for three quarters of the
ance and evasion are highest. To test this, in Tables 9 and mining power. We additionally test electricity prices in
10 we separately regress our dependent variables of USD per kWh in different countries and years and
bitnode intensity and Bitcoin merchants, on measures similarly find no effect (Tables 9 and 10, Model 7).
for the size of the shadow economy relative to the GDP One of the touted benefits of cryptocurrencies in the
(shadow economy), an indicator variable for whether a financial literature has been in enabling diversification
jurisdiction is a tax haven (tax haven), tax rates on (Bouri et al. 2019a) or as a hedge against equity markets
income, and capital gains (taxation), and a measure of (Bouri et al. 2019b), although some studies have found
tax burden across countries (tax burden). We maintain them to be a weak hedge for stock markets (Kliber et al.
the baseline controls, but omit other controls to avoid 2019; Shahzad et al. 2019). Certainly bitcoins are held
confounding of variables. We do not find an indication due to their speculative nature (Rauchs et al. 2018; Baur
to support the shadow economy and taxation drivers et al. 2018b), in congruence with the reported effects of
(Tables 9 and 10; Models 1 to 4). This could be attrib- our high-risk willingness variable, in addition to serving
uted either to more hidden uses of bitcoins in such as an alternative investment (Glaser et al. 2014). We test
nations or that the new potentials that cryptocurrencies to see if Bitcoin infrastructure has seen an uptick world-
enable for tax evasion and shadow economy, are not yet wide as a hedge for low returns in national stock mar-
extensive enough to constitute a large part of bitcoin kets. Model 8 in Tables 9 and 10 presents results of that
use. and does not show evidence for hedging as a main driver
of bitcoin use. In Model 9 of these tables, we addition-
20
Internet Appendix, Table IA.VI provides the variable description ally test to see if purely low returns during the global
and data sources for these variables. Our analyses of these additional financial crisis can be attributable to an upsurge in
characteristics, reported in Tables 9 and 10, are restricted to fewer than
all 137 countries in Panel A, as certain variable sources do not report adoption, and similarly do not find statistically signifi-
values for all countries in our sample. cant effects on our crisis stock market return variable.
382 E. Saiedi et al.

Table 9 Additional analysis on the role of alternative potential drivers (Panel A; country-year level)

Dependent variable (1) (2) (3) (4) (5) (6) (7) (8) (9)
Bitnode intensity

Shadow economy − 0.037

(0.031)
Tax haven − 1.052
(0.665)
Taxation 0.011
(0.021)
Tax burden −0.011
(0.027)
Bitcoin mining 0.374
country
(0.996)
Bitcoin mining MWs − 0.013
(0.031)
Electricity price − 0.002
(0.035)
Stock market return − 0.002
(0.005)
Crisis stock market 0.005
return
(0.019)
GDP per capita 1.173** 1.452*** 1.735*** 1.640*** 1.420*** 1.408*** 1.425*** 1.680*** 1.384**
(0.373) (0.377) (0.421) (0.334) (0.333) (0.351) (0.334) (0.370) (0.477)
Restrictive regulation − 0.500 − 0.579* − 0.474 − 0.570+ − 0.582* − 0.466 − 0.582* − 0.636 − 0.744
(0.348) (0.279) (0.363) (0.331) (0.295) (0.293) (0.283) (0.439) (0.490)
Constant − 8.222* − 11.567*** − 14.639*** − 12.552** − 11.480*** − 11.260*** − 11.281*** − 13.802*** − 10.959**
(3.646) (3.065) (3.731) (4.325) (2.869) (2.936) (3.366) (3.326) (4.252)
Year FEs Yes Yes Yes Yes Yes Yes Yes Yes Yes
World Region FEs Yes Yes Yes Yes Yes Yes Yes Yes Yes
Year-World Region Yes Yes Yes Yes Yes Yes Yes Yes Yes
FEs
Observations 670 685 580 665 685 670 685 440 460
Log-likelihood − 1291.6 − 1314.4 − 1131.3 − 1267.2 − 1314.8 − 1269.2 − 314.6 − 975.1 − 1023.1
Number of Countries 134 137 116 134 137 134 137 92 92
RHO 0.74 0.74 0.77 0.75 0.74 0.75 0.74 0.76 0.75

Notes: All models are Tobit random effects models lower-censored at 0 and their associated pseudo R-squared are reported. Panel-data
adjusted bootstrapped standard errors, clustered at the country level, are reported in parentheses. The symbols ***, **, *, + mean that the
reported coefficients are statistically different from zero, respectively, at the 0.1, 1, 5 and 10% level

6 Discussion and limitations supporting Bitcoin’s peer-to-peer network (the

hosting of bitnodes, which are infrastructure meant
To date, bitcoin is clearly the most prominent and to validate and verify transactions) as well as
widespread cryptocurrency. A decade on from the infrastructure facilitating the integration of bitcoin
first bitcoin transaction, this paper studies the de- into the regular economy (merchants’ acceptance
terminants of adoption of infrastructure supporting of bitcoins as payment) has been explored. The
bitcoin’s use as a digital currency. Infrastructure results are summarized in Table 11.
Global drivers of cryptocurrency infrastructure adoption 383

Table 10 Additional analysis (continued) on the role of alternative potential drivers (Panel A; Country-year level)

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Dependent variable Bitnode merchants

Shadow economy − 0.003

(0.033)
Tax haven − 0.431
(1.276)
Taxation 0.003
(0.038)
Tax burden − 0.071
(0.047)
Bitcoin mining country 1.003
(1.283)
Bitcoin mining MWs 0.002
(0.031)
Electricity price 0.020
(0.025)
Stock market return 0.020
(0.019)
Crisis stock market return − 0.010
(0.025)
GDP per capita 1.343* 1.413** 1.550* 1.137** 1.387** 1.474** 1.372* 0.590 0.689+
(0.648) (0.492) (0.633) (0.438) (0.535) (0.557) (0.588) (0.364) (0.362)
Restrictive Bitcoin Regulation − 0.024 − 0.024 0.461 − 0.129 − 0.109 − 0.132 − 0.053 − 0.035 − 1.203
(0.821) (0.750) (0.941) (0.793) (0.725) (1.040) (0.970) (0.708) (1.315)
Constant − 11.121 +
− 12.096 *
− 13.596 *
− 3.933 − 12.092 *
− 12.620 *
− 11.805 *
− 4.499 − 5.190
(6.577) (4.786) (5.954) (4.898) (5.281) (5.213) (5.633) (3.641) (3.540)
Year FEs Yes Yes Yes Yes Yes Yes Yes Yes Yes
World Region FEs Yes Yes Yes Yes Yes Yes Yes Yes Yes
Year-World Region FEs Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 670 685 580 665 685 670 685 440 460
Log-likelihood − 1382.1 − 1418.2 − 1278.1 − 1407.8 − 1418.0 − 1377.5 − 1417.8 − 1151.0 − 1195.8
Number of Countries 134 137 116 134 137 134 137 92 92
RHO 0.23 0.23 0.21 0.20 0.23 0.23 0.23 0.18 0.19

All models are Tobit random effects models lower-censored at 0 and their associated pseudo R-squared are reported. Panel-data adjusted
bootstrapped standard errors, clustered at the country level, are reported in parentheses. The symbols ***, **, *, + mean that the reported
coefficients are statistically different from zero, respectively, at the 0.1, 1, 5, and 10% level

Our results offer some support for the view that bitcoin fintech technologies (see, e.g., Saiedi et al. 2017; Bertsch
adoption is driven by perceived failings of traditional et al. 2017; Thakor and Merton 2018). We also find greater
financial institutions and systems (see, e.g., Cohen 2017; supply of and demand for Bitcoin infrastructure in years in
Shiller 2019; Vigna and Casey 2015). In particular, we find which countries undergo inflation crises, potentially indic-
more activity in the running of Bitcoin nodes in regions ative of a loss of faith in central-bank issued currencies or
with low trust in banks and the financial system. These using bitcoin as an investment or store of value. Our
corroborate the role of distrust in banks, which spread findings suggest that the ideological motives for use of
widely in the aftermath of the financial crisis of 2007– cryptocurrencies in entrepreneurial financing (e.g., Fisch
2009, being a factor in the early use of other peer-to-peer et al. 2019) extend to sentiments of infrastructure adopters,
384 E. Saiedi et al.

Table 11 Summary of hypotheses and results of hypothesis tests

Hypothesis number Hypothesis Results Support for nodes and/or

merchants?

H1 The occurrence of inflation crises is associated Supported Nodes

with increased adoption of Bitcoin infrastructure.
H2a (vs. countering H2c) The lower (for H2c: greater) the population of H2a rejected in Nodes and partially merchants
financially included adults, the greater the favor of H2c
adoption of Bitcoin infrastructure.
H2b (vs. countering H2d) The lower (for H2d: higher) the level of competition H2b rejected in Nodes
in banking markets, the greater the adoption of favor of H2d
Bitcoin infrastructure.
H3a The greater the level of trust in others, the greater the Partially supported Nodes
adoption of Bitcoin infrastructure.
H3b The greater the level of distrust in banks and the Supported Nodes
financial system, the greater the adoption of
Bitcoin infrastructure.
H3c The higher the risk-willingness, the greater the Supported Merchants and partially nodes
adoption of Bitcoin infrastructure.
H4 The more money-laundering activities taking place, Supported Nodes and partially merchants
the greater the adoption of Bitcoin infrastructure.
H5 The stronger the rule of law, the greater the adoption Supported Nodes and merchants
of Bitcoin infrastructure.

who are active supporters of the cryptocurrency allowing illicit business activity. Greater rule of law in-
ecosystem. creases the adoption of both types of Bitcoin infrastructure
Furthermore, we find that lower industry concentration we have analyzed. Higher rule of law may plausibly spur
in the banking market—which we take to indicate higher those desiring to engage in illicit activity in countries with
banking competition—are positively associated with run- high quality of police and courts to turn to online trade as a
ning of Bitcoin nodes. This finding is in line with contrary safer means of procuring illegal drugs (Barratt et al. 2016).
to findings for certain peer-to-peer financial technologies Results suggest that countries home to a greater risk of
(e.g., crowdfunding by Rau 2017), but in line with studies money laundering are more likely to host infrastructure for
finding complementarities between banking markets and bitcoin transaction verification and validation, being in
fintech development (Hornuf et al. 2018; Klus et al. line with evidence that bitcoins are utilized greatly for
2019). Our data also indicate that contrary to expectations drug trade and other illicit activities (Athey et al. 2016;
put forth by some (e.g., Killeen 2015; Vigna and Casey Böhme et al. 2015; Foley et al. 2019).
2015), greater Bitcoin infrastructure adoption has not yet Finally, our results lend support for Bitcoin node
emerged in countries with a larger unbanked population. infrastructure being driven, to a greater extent, by bank-
Together, these results paint a picture of bitcoins being ing market development, crime, and societal distrust in
used as a complement to existing financial institutions, banks and the financial system, while they lend support
rather than a substitute. It is likely that, as a recent study by for Bitcoin merchants being driven more by risk will-
Yermack (2018) of fintech adoption in Africa suggests, a ingness and criminality measures (specifically rule of
lack of information technology infrastructure underlies law). While the emergence of these infrastructures co-
failures of fintech platforms to succeed in countries with incides across geographies (c.f. Figs. 2b and 3) and time
low financial inclusion. (c.f. Figs. 4 and 5), results suggest they may be subject
Geographical variation in self-reported willingness to to different mechanisms. A possible interpretation is that
take risks is found to be correlated to the acceptance of the pattern for nodes reflects attitudes to a greater extent
bitcoins by merchants, suggesting that the high volatility (or active support for Bitcoin), whereas the pattern for
of bitcoins and uncertain regulatory environment re- merchants reflects the user base more (or interest in
duces its appeal in more risk-averse societies. Bitcoin) (Fig. 5).
Our results also support the view that bitcoin adoption Our study has certain limitations relating to the
is driven partly by the cryptocurrency’s usefulness in availability of data. One such limitation is that we do
Global drivers of cryptocurrency infrastructure adoption 385

Fig. 3 Total Bitcoin merchants per million users (see Table 1 for variable definition). Top 15 countries are San Marino, Liechtenstein,
Slovenia, Czech Republic, Montenegro, Belize, Estonia, Lithuania, Austria, Switzerland, Slovakia, Monaco, Latvia, Malta, and Greece

not directly observe the use of bitcoin as a currency assessing drivers of cryptocurrency infrastructure. In
(e.g., in the number of people holding bitcoin in a particular, more sophisticated users and criminals
country). We are therefore not able to explore to what using bitcoins are more likely to use firewalls to
extent the associations that our study uncovers are avoid revealing their network presence. We therefore
driven by country-level differences in bitcoin adop- suggest that the size of the effect we attribute to such
tion, or by other related factors. A further limitation technological or criminality characteristics can be
of our study that is in measuring the penetration of seen as a lower-bound to their true effect. Hence, by
Bitcoin infrastructure across space and time, we are directly measuring open support for the digital cur-
only capturing nodes and merchants who do not hide rency ecosystem, this study determines the effect of
their location. We believe such open support for financial characteristics on bitcoin adoption, while
Bitcoin’s network to provide a proxy for the adoption ascribing lower-bound for the effects of technologi-
of Bitcoin infrastructure. Nevertheless, our inability cal, criminality, and social drivers.
to observe hidden infrastructure can result in In choosing appropriate variables to specify and test
underweighting the role of certain characteristics in hypotheses about illicit activities as drivers of interest in

Fig. 4 Monthly time-series

variation of addition of Bitcoin
merchants worldwide on
CoinMap
386 E. Saiedi et al.

Fig. 5 Quarterly time-series

variation of Bitnode intensity
worldwide

Bitcoin, we had to resort to measures that we expect to et al. 2018) and are expected to have significant conse-
proxy for illicit activities, since more direct measures quences for business in the relatively near future (e.g.,
were not available with sufficient global comparability. see Dierksmeier and Seele 2018). Understanding the
We did not, for example, use actual dark-net drug sales drivers behind their spread thus far is therefore impor-
or drug-trade statistics (e.g., the Global Drug Survey; tant to scholars studying digital currencies and their role
see UNODC 2019), as they are currently only available in present and future digital entrepreneurial ecosystems.
for a limited subset of countries, and with varying de- While papers on potential risks and opportunities
grees of reliability. Another limitation is the lack of posed by cryptocurrencies are flourishing (e.g., Choo
appropriate proxies for certain relevant constructs relat- 2015; Dierksmeier and Seele 2018; Kethineni and Cao
ed to attitudes towards cryptocurrencies and their poten- 2019; Tasca 2015), as well as those posed by
tial advantages over fiat currencies. For example, cul- blockchains and ICOs (e.g., Kewell et al. 2017; Martino
tural preferences for privacy is a potentially relevant et al. 2019), little is empirically known about the spread,
factor, and data is available from surveys. However, growth, and patterns of use of these innovations. To the
recent research casts doubts as to their reliability due best of our knowledge, this paper is the first academic
to the Digital Privacy Paradox, in that stated privacy empirical inquiry with global geographic coverage of
preferences are different from observed privacy prefer- the emergence of Bitcoin as an alternative financial
ences (Athey et al. 2017). Certain relevant national system. A significant contribution of this study is there-
characteristics were also omitted as they underlie uti- fore analyzing large global cryptocurrency databases
lized variables, as is for instance the case of financial and providing the annualized data online21 for further
literacy (not included as a separate variable; Grohmann empirical studies in this nascent and emerging field. The
et al. 2018) and financial inclusion (included in our study also offers important insights to scholars seeking
models). to understand the growth of cryptocurrencies through a
grounded global analysis, not least to separate a great
deal of myth from facts.
7 Conclusions and implications While descriptive in nature, our results paint a picture
of the adoption of cryptocurrencies as being related to its
The global spread of the use of cryptocurrencies is a usefulness as a complement to traditional financial ser-
notable feature of the twenty-first century economy. vices. Such usefulness derives on the one hand from
Already an important locus of entrepreneurial activity,
cryptocurrencies can alter the business models of 21
Data is included as supplementary material, accessible from this
existing actors within entrepreneurial finance (Block journal's webpage.
Global drivers of cryptocurrency infrastructure adoption 387

inefficiencies and failings (e.g., inflation crises) of tra- 7.2 Implications for actors in digital ecosystems
ditional economic or financial systems, and on the other
hand on the opportunity to conduct illegal or shady This paper depicts how a decentralized infrastructural
transactions using cryptocurrencies. system grows organically and offers indirect evidence of
We now turn to the implications of our study for motives that underlie the fast-growing and evolving
various scholars, as well as for institutions which may digital currency ecosystem. Our findings indicating that
shape, regulate, or lend oversight to digital currencies interest in Bitcoin thus far is strongest in the most
now or in the future. financially well-developed locations may spur entrepre-
neurs in the digital currency ecosystem to develop new
means to simplify the process of using bitcoins for it to
7.1 Implications for scholars of digital currency have a chance of reaching the unbanked, or integrating
their products on more widespread communication tech-
With the support for cryptocurrency infrastructure and nologies (e.g., mobile phones). Some efforts, such as the
interest in making financial transactions using integration of bitcoin on the popular M-Pesa system in
cryptocurrencies expected to keep growing over time, Kenya (Vigna and Casey 2015) have been made, while
further research should seek to explore further what of greater use towards poverty reduction could be fo-
drives participation, and what consequences such sup- cusing on, e.g., reducing persistently high international
port may have on incumbent financial structures. This remittance fees (Beck and Martínez Pería 2011).
study has demonstrated that institutional and social fac- Technology firms planning or exploring to issue their
tors had important roles to play in the growth of the own digital currencies, such as Facebook’s Libra project
Bitcoin system to date, which suggests that such factors (Libra Association 2019), can draw lessons from the
should also be carefully considered in future research on historical patterns of adoption of bitcoins depicted in this
digital currency development. paper. In particular, the findings indicating that interest in
The study shows that despite the much-touted Bitcoin is associated with distrust in banks and established
promise of bitcoins (e.g., Dierksmeier and Seele financial institutions, would seem to point to an advantage
2018; Killeen 2015; Vigna and Casey 2015) and that can be exploited in marketing new currencies. The
blockchain technologies (e.g., Larios-Hernández greater trust of consumers in technology companies than
2017) in increasing financial inclusion and poverty in banks (Bain and Company 2019; Forbes 2015) could
reduction, it is the countries where such issues are provide an edge in harnessing fintech technologies, albeit
less poignant that have the strongest penetration of if they can retain such trust (Nair 2019).
Bitcoin infrastructure. On the one hand, this finding
calls for further research on barriers to adoption of 7.3 Implications for central banks and public authorities
digital currencies in less developed settings, and
how these may be overcome. On the other hand, The debate on the risk and opportunities of central banks
our results may be seen as fuel for a broader issuing cryptocurrencies, or more broadly central bank
discussion about the role of cryptocurrencies. While digital currencies (CBDC), is intensifying (Borio et al.
clearly potentially helpful for citizens of countries 2017; Conti-Brown et al. 2018; Engert and Fung 2017;
beset by hyperinflation, evidence on the strong Niepelt 2020). Many central banks currently conduct re-
connection to illicit activity and the, as of yet, search on digital currency issuance (Cœuré 2018). The
unrealized promise of providing a means to in- real-world uptake and support for bitcoins can be seen as
crease financial inclusion would seem to weigh in a case study for the risks and demand for preserving
with arguments that the cryptocurrency’s growth is anonymity in digital transactions, as alluded to in Acquisti
to be seen as cause for concern—e.g., in the light et al.’s (2016) survey on the economics of privacy. The
of price manipulation (Griffin and Shams 2019), history of bitcoin use so far can help inform CBDC-design
heavy electricity use associated with Bitcoin mining as regulators weigh trade-offs between the need to control
(Krause and Tolaymat 2018) and the increased oc- transaction fraud and the desire to avoid invasive or costly
currence of cybercrimes and frauds involving monitoring of individuals. Some economic theorists sug-
cryptocurrencies (Corbet et al. 2019; Kamps and gest that this trade-off will exist despite any technological
Kleinberg 2018; Li et al. 2018; Su 2019). progress (Kahn et al. 2005; Kahn and Roberds 2008).
388 E. Saiedi et al.

The correlation of distrust in banks with bitcoin ensure security and to prevent circumvention of consensus
adoption should be of concern to financial regulators rules such as double-spending (i.e., when a user attempts
for whom financial stability and transactions through to spend the same bitcoin twice), each node uses a ran-
legitimate monetary channels is tantamount. Nowhere domized protocol to select eight peers with which it estab-
is this more a concern than in countries experiencing lishes long-lived outgoing connections, and exchange
very high inflation. Low faith in the currency’s stability, views of the state of the blockchain with their incoming
coupled with a distrust in the national financial system, and outgoing peers in order to verify and validate new
can drive the populace into digital currencies, rendering transactions (Heilman, Kendler, and Zohar 2015).
a monetary authorities’ monetary supply less effective. If a node fully validates every block and transaction, by
Law enforcement agencies must react to the use of independently checking them against all consensus rules of
digital assets for narcotics-related money laundering. Bitcoin, it is termed a full node. Some of the primary
Finally, there is cause for concern for tax administrations consensus rules of Bitcoin, verified by full nodes, are: (a)
and revenue agencies in use of digital currencies for tax Spent transactions require correct signatures, (b) Blocks
evasion, as evidenced by formation of international task may only release a limited number of bitcoins as a block
forces (J5 2018). Further studies could shed more light reward, (c) Blocks and transactions are to follow the
on empirically establishing the extent of use of consensus format and (d) A transaction output cannot be
cryptocurrencies for money laundering and tax evasion. double-spent. Such full nodes would require a bitcoin core
client installed on a computed connected to the Internet. A
Acknowledgments We owe thanks to Silvio Vismara (the edi- full note contains the entire record of bitcoin’s blockchain,
tor), the anonymous referees, Gustav Martinsson, Per Strömberg, a distributed public ledger of all transactions.22 Archival or
Christopher Baum, Sergio Correia, Andreas Stephan, David
Yermack, Ali Mohammadi, Mariassunta Giannetti, Matthew listening nodes (also called reachable nodes) are a subset of
Stogsdill, Zheng Xiaochen, Mike van Rossum, Carolin Bock, full nodes that accept incoming connections and upload or
Yury Kucheev, Joaquin Ordieres, and seminar participants at the “relay” complete copies of archival bitcoin blocks and
4th EntFin Conference, Swedish House of Finance and Royal transaction data to peers on the bitcoin network. They are
Institute of Technology, for helpful comments and discussions.
We owe thanks to Addy Yeow, the SatoshiLabs, and NORC at the useful for bootstrapping new full node peers or helping
University of Chicago for the databases utilized and Patrícia light-weight nodes discover their balance. Light-weight
Estavão for an infographic. All remaining errors are our own. nodes are meant for market participants who do not want
to run full nodes and use a Simplified Payment Verification
Funding information Open access funding provided by Royal method to verify transactions by downloading only a part
Institute of Technology. We would also like to thank two anony- of the blockchain (e.g., only block headers and partial
mous reviewers whose comments greatly improved this manu-
script. This paper is produced as part of the EMJD Program, blocks).
European Doctorate in Industrial Management (EDIM), funded Running a full node is the only way for a bitcoin
by the European Commission, Erasmus Mundus Action 1. market participant to independently ascertain that all
consensus rules are followed and for full removal of a
need for a trusted alternative node or third party. Run-
Appendix ning full nodes also much increases a market partici-
pants’ security as many attacks against lightweight
Bitcoin infrastructure nodes do not affect full nodes, and they offer the most
privatized and anonymized way to transact in bitcoins,
Bitcoin nodes hiding addresses belonging to the full node operator. It
is, hence, in the interest of market participants to run a
Relating to the supply-side of global bitcoin infrastructure, full node.23
we focus our study on Bitcoin nodes, i.e., bitnodes. Bitcoin
relies on a decentralized network of nodes, which use 22
The size of the entire blockchain has exponentially increased, from
computational proofs-of-work to reach consensus on the 21.7GB to 176.5GB from Sept 2014 to 2018. Through the same time
bitcoin blockchain (Nakamoto 2008). To validate and period the price of storage has decreased by ~ 45% (http://jcmit.
net/diskprice.htm).
relay transactions, the Bitcoin network must broadcast 23
https://en.bitcoin.it/wiki/Clearing_Up_Misconceptions_About_
messages using nodes (see Fig. 1 for validation and Full_Nodes#Myth:_There_is_no_incentive_to_run_nodes_so_the_
verification actions in a bitcoin transaction life cycle). To network_relies_on_altruism
Global drivers of cryptocurrency infrastructure adoption 389

Bitcoin nodes contribute to bitcoin’s aim to serve as a (a) Downloading ultra-compressed files containing
medium of exchange, and contribute to bitcoin’s goal to thousands of JSON files for first month of each
serve as a store of value. Nodes are critical to the quarter, with each JSON file being a python-
functioning of the entire bitcoin ecosystem and were a crawled database snapshot taken nearly every
core part of the functioning of Bitcoin’s blockchain from 6 min containing such information as IP addresses,
its design inception. Figure 2a shows a global map of the city, country, and coordinate information relating
revealed intensity of bitcoin node adoption, averaged for to nodes
the five years from 2014 to 2018. Figure 2b shows a (b) Retaining every 10th file corresponding to nearly
map of the number of non-hidden unique bitcoin nodes hourly records of all crawled non-hidden Bitcoin
per month, averaged for the same period. nodes
(c) Re-structuring the hourly files to form monthly
databases
Bitnodes dataset preparation (d) Calculating all variables separately for countries
and administrative regions of 34 countries included
We obtain data on reachable (aka listening) full nodes on in LITS III survey. The region-level analysis
the Bitcoin network from Bitnodes.earn.com.24 Our consisted of:
database contains all reachable nodes running Bitcoin
protocol version 70,001 (i.e., Bitcoin core 0.8.x), which i. Matching world cities to NUTS-2 regions of 34
was introduced on February 2013,25 or later versions. countries in Life in Transition Societies’ Social
These nodes may or may not be full nodes; however, Survey
they have been set up to be reachable. Reachable nodes ii. Geo-locating regions of unidentified IP ad-
have port 8333, or a port they have configured the dresses, using geo-location tools of IP ad-
software to, open for incoming connections. The dresses meant for the STATA software, specif-
default configuration in node software is to have ically the SSC package geocode_ip, written by
listening enabled. However, home connections are Correia (2017), yet modified to use the API
usually filtered by a modem or router. Bitcoin core service from ipgeolocation.io.
will request automatic configuration of your router to
allow incoming connections. Unfortunately, many (e) Creating country-year and region-year data of non-
routers do not allow automatic configuration and hidden bitcoin node use, in a panel data structure.
require manual configuration. Moreover, firewalls may There are 201 countries worldwide for which in at
also need to be configured to allow inbound connections least one year between 2014 and 2018 a Bitnode is
to port 8333. Hence, it is conceivable that there are as operated.
many as 6 to 8 times more nodes on the network than
our database could reach (Wang and Pustogarov 2017). The Bitnodes’ database uses Python’s GeoIP2
However, our database can be assumed to provide a geo-locating package to identify countries of nodes.
reasonable proxy for the total geographical distribution Also the geolocation we use to locate unidentified
of nodes in the bitcoin network. IP addresses uses two alternative and independent
From the extensive Bitnode database (~ 1 Terabyte), IP databases. These IP address geolocation services
location-specific measures for the adoption of Bitcoin state the accuracy of IP geolocation to be are
nodes was constructed through the following procedure: around 95% to 99% at the country-level, and 55%
to 80% at the region level (See https://www.
iplocation.net/geolocation-accuracy). Poese et al.
24
(2011), Komosny et al. (2017) and Shavitt and
Network snapshots are taken every few minutes in JSON format,
Zilberman (2011) find that the main geolocation
and have been compressed and stored in monthly Tarball files from
September 2014 onwards. The python crawler implementation of this databases are accurate at the country level. Hence,
website is made available by Addy Yeow on GitHub, https://github. our Bitnode analyses at the country and region
com/ayeowch/bitnodes.
25 level, while providing a best-estimate of location
To illustrate speed of adoption of a protocol, a year on from release
of the following protocol, nearly 98% of nodes had updated to the of bitcoin nodes, is limited by the accuracy of geo-
newer protocol. location.
390 E. Saiedi et al.

Bitcoin merchants a Stata-readable format). CoinMap is a website where

either business owners or interested customers or users
We also study bitcoin infrastructure driving the demand would add a business that accepts bitcoins (or bitcoin and
for bitcoins, using data on how many merchants that other cryptocurrencies) as payment in exchange for a
accept bitcoin as a currency for selling their goods or service or good. Some of the businesses on these websites
services. By adopting bitcoin payment technology, mer- only accept bitcoins on their online websites.
chants contribute to bitcoin serving as a medium of CoinMap started in early 2013 with self-reporting of
exchange. They also facilitate bitcoin’s adoption as a unit venues on an OpenStreetMap26 map, being done by
of account, by enabling the quotation of prices in bitcoins. adding a tag of bitcoin = yes (or selecting bitcoin as a
Merchants’ adoption of technology to accept bitcoins payment type in a pre-defined drop-down list) to venues
as payment should largely be driven by perceived demand on its extensive interactive map. One would have needed
for such services among its customers. A certain amount to sign up and enter in edit mode and use OpenStreetMap’s
of computer savviness would be expected for merchants to user-friendly editor interface to select a building or busi-
start accepting bitcoins; businesses would need to down- ness and tag it as accepting bitcoins. CoinMap would
load a bitcoin client and give customers a bitcoin address synchronize with any such entries made into
where payments can be sent. However, bitcoin payment OpenStreetMap up until the middle of 2017, after which
service providers such as Bitpay, Bitgo, or Coinbase.com OpenStreetMap integration stopped and adding venues on
offer web and mobile apps that simplify the process of CoinMap’s own interface by logging in via Google/
becoming a bitcoin-accepting merchant. Facebook/etc. and right-clicking, became the only way to
It has been reported that merchants reported satisfaction add a bitcoin-accepting merchant. Soon after launching in
with bitcoin payment processing due to its strikingly low- 2013, CoinMap became the primary website worldwide,
cost (Böhme et al. 2015). As an example, Coinbase (the as referred to in online Bitcoin forums, to find businesses
largest bitcoin payment processing firm) currently charges where bitcoin holders can spend their bitcoins. On online
zero fees on incoming payments, greatly reduced from the forums, bitcoin enthusiasts express a great desire to devel-
fees they bear when credit card are used for payment. op an extensive network of merchants that accept bitcoins,
Compared with other payment forms, however, bitcoin so it could serve their hope for it to serve as a formidable
payments expose the merchants to a new type of volatility alternative to fiat currencies.
risk, since the exchange rate of any bitcoins that is received The data contains coordinates and categories of busi-
in payment may vary substantially over time. Payments nesses accepting bitcoins as payment, together with the
can be accepted using hardware terminals, touch screen date and time they were added to CoinMap’s online
apps or simple wallet addresses using QR codes. Most map. The top categories are shopping, food, ATMs,
bitcoin merchants prefer to minimize the volatility risk of lodging, cafes, nightlife, and sports venues. We geo-
holding bitcoins by offloading or selling bitcoins as soon locate all addresses using the coordinates provided to
as possible after the sale of their service or product. identify which city, state and country they are located in.
While this study is focused on infrastructure As of the end of 2018, the database consists of 14,149
supporting Bitcoin, many infrastructure providers for venues. Removing duplicated entries of venues, our
Bitcoins simultaneously extend support to other final list has 13,741 merchants. Figure 4 shows the
cryptocurrencies, as it has become prevalent with monthly frequency of addition of bitcoin merchants
multi-currency platforms payment service providers over time on CoinMap since its inception. No data is
used by cryptocurrency merchants. Figure 3 shows the available on the size of transactions.
worldwide of the number of merchants per country,
averaged by the nation’s population.

Bitcoin merchants dataset preparation

26
OpenStreetMap is one of the world’s largest collaborate mapping
Our data for merchants accepting payments in bitcoin is projects. Developed as a free editable map of the world in 2004, it is
extracted from data stored using CoinMap’s API (The data used by large established corporations as well as independent website
owners to develop a navigable map with complete business/
is available at https://coinmap.org/api/v1/venues/, whereas institutional venue information. As the world’s largest database of
an R-code was written to convert the JSON format data to buildings, OpenStreetMap covers an extensive database of businesses.
Table A.I Correlation matrix

Variable no./name 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1 Bitnode intensity 1

2 Unique bitnodes 0.66 1

3 Bitnode Hours 0.45 0.52 1
4 Bitcoin merchants 0.25 0.24 0.2 1
5 GDP per capita 0.52 0.59 0.64 0.21 1
6 Restrictive regulation − 0.16 − 0.19 − 0.04 − 0.08 − 0.09 1
7 Internet penetration 0.48 0.53 0.68 0.21 0.89 − 0.05 1
8 Broadband penetration 0.6 0.66 0.64 0.25 0.83 − 0.15 0.83 1
9 ICT market development 0.53 0.61 0.7 0.22 0.91 − 0.07 0.95 0.89 1
10 Latest technology 0.52 0.56 0.55 0.19 0.8 − 0.09 0.76 0.75 0.78 1
11 Secure Internet servers 0.63 0.29 0.31 0.22 0.42 − 0.11 0.39 0.49 0.43 0.4 1
12 Mobile subscriptions 0.2 0.26 0.38 0.09 0.57 0.03 0.56 0.38 0.58 0.45 0.13 1
Global drivers of cryptocurrency infrastructure adoption

13 Electricity cost − 0.2 − 0.24 − 0.41 − 0.09 − 0.56 − 0.07 − 0.52 − 0.39 − 0.53 − 0.43 − 0.15 − 0.45 1
14 Inflation crisis − 0.09 − 0.09 − 0.09 − 0.03 − 0.07 − 0.03 − 0.14 − 0.13 − 0.13 − 0.3 − 0.07 − 0.18 0.15 1
15 Inflation − 0.22 − 0.23 − 0.21 − 0.08 − 0.27 − 0.05 − 0.31 − 0.32 − 0.31 − 0.45 − 0.15 − 0.27 0.21 0.82 1
16 Unbanked − 0.51 − 0.58 − 0.59 − 0.23 − 0.82 0.02 − 0.77 − 0.81 − 0.83 − 0.77 − 0.41 − 0.48 0.44 0.16 0.32 1
17 Internet banking 0.56 0.54 0.35 0.19 0.58 − 0.11 0.51 0.62 0.55 0.64 0.43 0.17 − 0.12 − 0.07 − 0.19 − 0.68
18 5-bank asset concentration 0.09 0.07 − 0.2 0.03 0.09 − 0.22 0.04 0.1 0.04 0.06 0.07 − 0.05 0.12 0.01 − 0.07 − 0.05
19 3-bank asset concentration 0.14 0.1 − 0.2 0.01 0.08 − 0.21 0.03 0.11 0.04 0.04 0.08 − 0.05 0.1 0.04 − 0.02 − 0.04
20 Rule of law 0.61 0.65 0.57 0.22 0.78 − 0.16 0.76 0.82 0.8 0.86 0.47 0.45 − 0.4 − 0.29 − 0.47 − 0.8
21 Organized crime 0.4 0.42 0.27 0.15 0.54 − 0.06 0.54 0.51 0.54 0.6 0.27 0.37 − 0.24 − 0.19 − 0.35 − 0.53
22 Crime and violence costs 0.35 0.41 0.28 0.15 0.52 − 0.04 0.53 0.51 0.54 0.55 0.26 0.34 − 0.23 − 0.18 − 0.32 − 0.48
23 Money laundering 0.21 0.2 0.39 0.03 0.28 0.02 0.21 0.23 0.23 0.22 0.12 0.12 − 0.19 0 0 − 0.23
24 Shadow economy − 0.42 − 0.44 − 0.44 − 0.15 − 0.68 0.06 − 0.61 − 0.62 − 0.65 − 0.64 − 0.36 − 0.31 0.32 0.01 0.17 0.64
25 Taxation 0.02 − 0.03 0.08 − 0.06 0.12 − 0.06 0.01 0.05 0.01 0.19 0.13 − 0.07 − 0.1 − 0.02 − 0.02 − 0.1
26 Tax burden − 0.27 − 0.4 − 0.27 − 0.17 − 0.29 0.15 − 0.2 − 0.47 − 0.27 − 0.35 − 0.28 0.11 − 0.04 − 0.01 0.09 0.36
27 Tax haven − 0.1 − 0.1 − 0.04 − 0.04 0.11 0.01 0.08 0.02 0.08 0.13 − 0.08 0.21 − 0.09 − 0.05 − 0.04 − 0.11
28 Stock market return − 0.02 − 0.09 0.04 0.03 − 0.01 − 0.05 0.02 0.01 0.02 − 0.03 0 − 0.02 − 0.03 0.21 0.2 − 0.02
29 Crisis stock market return − 0.14 − 0.22 − 0.11 − 0.15 − 0.02 0.04 − 0.15 − 0.28 − 0.18 − 0.01 − 0.09 0.05 0.05 0.08 0.16 0.2
30 Bitcoin mining country 0.16 0.17 0.27 0.03 0.19 0.08 0.18 0.21 0.21 0.12 0.1 0.18 − 0.04 0.08 0.05 − 0.24
31 Bitcoin mining MWs 0.02 0.01 0.16 − 0.02 0.06 0.18 0.04 0.11 0.07 0 0.01 0.01 − 0.06 − 0.03 − 0.06 − 0.1
32 Electricity price − 0.03 − 0.01 0.08 0 0.01 − 0.04 0.01 − 0.02 0 − 0.1 − 0.03 − 0.07 0.19 0.4 0.45 − 0.03

Variable no./name 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

2
3
391
Table A.I (continued)
392

Variable no./name 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

4
5
6
7
8
9
10
11
12
13
14
15
16
17 1
18 0.16 1
19 0.15 0.94 1
20 0.68 0.09 0.07 1
21 0.53 0.17 0.15 0.7 1
22 0.45 0.08 0.07 0.61 0.91 1
23 0.12 − 0.29 − 0.24 0.13 − 0.06 − 0.1 1
24 − 0.53 − 0.08 − 0.1 − 0.68 − 0.46 − 0.46 − 0.11 1
25 0.13 0.03 0 0.2 − 0.02 − 0.1 0.13 − 0.13 1
26 − 0.49 − 0.07 − 0.08 − 0.42 − 0.12 − 0.09 − 0.15 0.25 − 0.29 1
27 0.04 0.07 0.04 0.05 0.07 0 − 0.06 − 0.02 − 0.04 0.1 1
28 0.01 0 0 − 0.05 − 0.09 − 0.1 0.01 − 0.02 0.07 − 0.03 − 0.02 1
29 − 0.06 − 0.08 − 0.09 − 0.08 0.01 − 0.07 0.21 0.01 0.17 0.04 0.13 − 0.01 1
30 0.25 − 0.11 − 0.08 0.18 0.09 0.09 0.14 − 0.1 0.1 − 0.06 − 0.07 0.11 0.06 1
31 0.09 − 0.15 − 0.12 0.05 0.06 0.09 0.1 − 0.07 0.05 − 0.05 − 0.04 0.03 0.05 0.7 1
32 0.02 − 0.03 − 0.03 − 0.18 − 0.19 − 0.22 0.13 0 0.13 − 0.05 − 0.03 0.38 0.02 0.3 − 0.05 1
E. Saiedi et al.
Global drivers of cryptocurrency infrastructure adoption 393

Internet appendix

Table IA.I Robustness regressions on Bitnode intensity by adding exploratory variables in isolation to baseline regression (Panel A)

(1) (2) (3) (4) (5) (6)

Dependent variable Bitnode intensity

Financial characteristics
Inflation crisis 0.530+ 0.654*
(0.306) (0.285)
Unbanked 0.008 − 0.006
(0.009) (0.013)
5-bank asset concentration 0.003 0.000
(0.013) (0.015)
Criminality characteristics
Money laundering 1.347* 1.156+
(0.604) (0.610)
**
Rule of law 1.445 1.267*
(0.474) (0.508)
Baseline characteristics
GDP per capita 0.681 1.428*** 1.339** 1.424*** 1.335*** 0.761+
(0.457) (0.337) (0.499) (0.320) (0.356) (0.395)
Restrictive regulation − 0.529 − 0.574 − 0.584 +
− 0.573 +
− 0.586 +
− 0.541
(0.324) (0.377) (0.332) (0.329) (0.343) (0.354)
Constant − 6.624 − 11.482*** − 10.492* − 11.482*** − 11.358*** − 5.935+
(4.347) (2.827) (4.802) (3.012) (3.132) (3.434)
Year FEs Yes Yes Yes Yes Yes Yes
World region FEs Yes Yes Yes Yes Yes Yes
Year-world region FEs Yes Yes Yes Yes Yes Yes
Panel data specification Tobit Tobit Tobit Tobit Tobit Tobit
Observations 685 685 685 685 685 685
Number of countries 137 137 137 137 137 137
Pseudo R-Squared 0.41 0.36 0.36 0.36 0.37 0.40
Log-likelihood − 1306.1 − 1314.8 − 1314.8 − 1314.9 − 1312.9 − 1309.4
RHO 0.72 0.74 0.74 0.74 0.74 0.73

All models are Tobit random effects models lower-censored at 0 and their associated pseudo R-squared are reported. Panel-data adjusted
bootstrapped standard errors, clustered at the country level, are reported in parentheses. The symbols ***, **, *, + mean that the reported
coefficients are statistically different from zero, respectively, at the 0.1, 1, 5, and 10% level
394 E. Saiedi et al.

Table IA.II Robustness regressions on unique bitnodes by adding exploratory variables in isolation to baseline regression (Panel A)

(1) (2) (3) (4) (5) (6)

Dependent variable Unique bitnode

Financial characteristics
Inflation crisis 7.038** 5.875*
(2.596) (2.445)
Unbanked − 0.189* − 0.288***
(0.078) (0.070)
5-bank asset concentration 0.011 0.013
(0.069) (0.082)
Criminality characteristics
Money laundering 6.916** 5.853*
(2.381) (2.770)
Rule of law 9.323*** 9.845***
(2.557) (2.332)
Baseline characteristics
GDP per capita 0.039 8.946*** 4.292* 8.903*** 8.446*** 3.342+
(2.209) (1.536) (1.749) (1.376) (1.198) (1.784)
Restrictive regulation 3.069* 3.384* 2.708* 3.386* 3.188* 3.663**
(1.347) (1.365) (1.354) (1.341) (1.596) (1.265)
Constant 19.808 − 58.362 ***
− 6.835 − 58.966 ***
− 57.456 ***
− 11.338
(20.597) (12.931) (16.692) (14.654) (10.462) (14.818)
Year FEs Yes Yes Yes Yes Yes Yes
World region FEs Yes Yes Yes Yes Yes Yes
Year-world region FEs Yes Yes Yes Yes Yes Yes
Panel data specification Tobit Tobit Tobit Tobit Tobit Tobit
Observations 685 685 685 685 685 685
Number of countries 137 137 137 137 137 137
Pseudo R-squared 0.62 0.58 0.58 0.58 0.58 0.61
Log-likelihood − 2266.7 − 2294.4 − 2284.1 − 2294.6 − 2292.0 − 2279.5
RHO 0.54 0.58 0.58 0.58 0.57 0.56

All models are Tobit random effects models lower-censored at 0 and their associated pseudo R-squared are reported. Panel-data adjusted
bootstrapped standard errors, clustered at the country level, are reported in parentheses. The symbols ***, **, *, + mean that the reported
coefficients are statistically different from zero, respectively, at the 0.1, 1, 5 and 10% level
Global drivers of cryptocurrency infrastructure adoption 395

Table IA.VII Robustness regressions on Bitnode hours by adding exploratory variables in isolation to baseline regression (Panel A)

(1) (2) (3) (4) (5) (6)

Dependent variable Bitnode hours

Financial characteristics
Inflation crisis 50.642 21.316
(65.208) (75.872)
Unbanked − 2.309 − 2.883*
(1.421) (1.316)
5-bank asset concentration − 5.080** − 5.709***
(1.676) (1.583)
Criminality characteristics
Money laundering 176.344*** 214.910***
(53.490) (55.422)
Rule of law 88.669+ 78.177
(45.858) (53.565)
Baseline characteristics
GDP per capita 72.001* 163.076*** 117.830*** 173.152*** 145.646*** 119.839**
(34.734) (30.603) (32.456) (27.254) (23.679) (40.040)
Restrictive regulation 32.565 47.150 43.588 30.404 45.289 51.749
(39.446) (34.713) (44.952) (47.014) (47.311) (54.152)
Constant 382.419 − 762.920* − 259.390 − 415.942 − 740.771* − 400.660
(431.859) (328.849) (417.735) (355.106) (330.463) (384.618)
Year FEs Yes Yes Yes Yes Yes Yes
World region FEs Yes Yes Yes Yes Yes Yes
Year-world region FEs Yes Yes Yes Yes Yes Yes
Panel data specification Tobit Tobit Tobit Tobit Tobit Tobit
Observations 685 685 685 685 685 685
Number of countries 137 137 137 137 137 137
Log-likelihood − 2386.8 − 2402.9 − 2397.6 − 2397.4 − 2403.4 − 2405.2
RHO 0.62 0.69 0.69 0.65 0.66 0.69

All models are Tobit random effects models lower-censored at 0 and upper censored at 750 (upper limit of active Bitnode hours in a month)
and their associated pseudo R-squared are reported. Panel-data adjusted bootstrapped standard errors, clustered at the country level, are
reported in parentheses. The symbols ***, **, *, + mean that the reported coefficients are statistically different from zero, respectively, at the
0.1, 1, 5, and 10% level
396 E. Saiedi et al.

Table IA.IV Robustness regressions substituting Inflation crisis with a 15% or 25% min inflation (Panel A)

(1) (2) (3) (4) (5) (6)

Dependent Variables: Bitnode intensity Unique bitnodes Bitnode hours Bitnode intensity Unique bitnodes Bitnode hours

Financial characteristics
25% Min Inflation 0.677 8.580** 74.911
(0.490) (3.202) (97.863)
15% Min Inflation 0.775* 8.038** 121.690+
(0.334) (2.452) (64.821)
Unbanked 0.008 − 0.189** − 2.320 0.008 − 0.190* − 2.376+
(0.009) (0.073) (1.623) (0.009) (0.081) (1.353)
5-bank asset concentration 0.003 0.013 − 5.071 **
0.004 0.015 − 5.035**
(0.016) (0.067) (1.555) (0.013) (0.072) (1.679)
Criminality characteristics
Money laundering 1.348** 6.945** 176.610*** 1.331** 6.825** 175.268**
(0.491) (2.532) (53.588) (0.510) (2.263) (59.596)
** *** * ** ***
Rule of law 1.469 9.634 92.375 1.470 9.513 96.620+
(0.503) (2.649) (46.450) (0.458) (2.301) (53.552)
Baseline characteristics
GDP per capita 0.656 − 0.238 68.815+ 0.679 0.009 68.917+
(0.475) (2.095) (38.871) (0.466) (1.860) (39.943)
Restrictive regulation − 0.534 3.016 +
32.214 − 0.479 3.572 *
40.662
(0.357) (1.549) (38.786) (0.372) (1.518) (37.398)
Constant − 6.412 22.056 409.386 − 6.618 19.767 406.150
(4.460) (19.682) (420.884) (4.342) (18.050) (403.880)
Year FEs Yes Yes Yes Yes Yes Yes
World Region FEs Yes Yes Yes Yes Yes Yes
Year-World Region FEs Yes Yes Yes Yes Yes Yes
Observations 685 685 685 685 685 685
Pseudo R-Squared 0.42 0.63 0.32 0.41 0.67 0.32
Number of Countries 137 137 137 137 137 137

All models are Tobit random effects models lower censored at 0 and their associated overall R-squared is reported. Panel-data bootstrapped
standard errors, clustered at the country level, are reported in parentheses. The symbols ***, **, *, + mean that the reported coefficients are
statistically different from zero, respectively, at the 0.1, 1, 5 and 10% level. Models 3 and 6 are in addition upper-censored at 750 (upper limit
of active Bitnode hours in a month)
Global drivers of cryptocurrency infrastructure adoption 397

Table IA.V Disentangling the effect of Rule of law on Bitcoin infrastructure adoption (Panel A; Country-year level)

(1) (2) (3) (4)

Dependent variables Bitnode intensity Bitcoin merchants Bitnode intensity Bitcoin merchants

Organized crime 0.152 − 0.419

(0.188) (0.373)
Crime and violence costs − 0.016 − 0.656
(0.221) (0.486)
Money laundering 0.992 0.516 1.117+ 0.554
(0.656) (0.724) (0.596) (0.718)
Inflation crisis 0.395 − 0.241 0.285 − 0.783
(0.447) (0.993) (0.418) (5.133)
Unbanked 0.009 − 0.050 − 0.004 − 0.050+
(0.011) (0.035) (0.012) (0.030)
Five-bank asset concentration 0.008 − 0.011 0.008 − 0.023
(0.019) (0.026) (0.014) (0.028)
*** **
GDP per capita 1.696 0.786 1.278 0.742
(0.480) (0.690) (0.443) (0.515)
Restrictive bitcoin regulation − 0.603+ − 0.588 − 0.612+ − 0.558
(0.316) (0.680) (0.335) (0.859)
Constant − 16.043*** − 1.822 − 11.075* 0.523
(4.393) (6.132) (4.367) (4.174)
Year FEs Yes Yes Yes Yes
World Region FEs Yes Yes Yes Yes
Year-World Region FEs Yes Yes Yes Yes
Observations 665 665 685 685
Log-likelihood − 1231.9 − 1393.8 − 1279.6 − 1400.4
Number of countries 133 133 137 137
RHO 0.75 0.20 0.73 0.22

All models are Tobit random effects models lower-censored at 0 and their associated pseudo R-squared are reported. Panel-data adjusted
bootstrapped standard errors, clustered at the country level, are reported in parentheses. The symbols ***, **, *, + mean that the reported
coefficients are statistically different from zero, respectively, at the 0.1, 1, 5 and 10% level. Models 1 and 2 exclude the 4 countries of Belize,
Bermuda, Cayman Islands and Syria, as no value was reported for the variable organized crime from the data source
398 E. Saiedi et al.

Table IA.VI Variable description and data sources of variables for alternative explanation analysis

Variable Description Source Source variable name

Panel A
Shadow Size estimates of the shadow economy as a percentage Elgin and Öztunalı (2012)
econo- of a nation’s GDP, based on last available year
my reported
Tax haven Indicator variable equal to 1 for countries either ec.europa.eu/taxation_
black-listed or gray-listed by the EU as customs/tax-common-eu-list_en,
non-cooperative tax jurisdictions accessed on December 6, 2018
Taxation Taxes on income, profits, and capital gains are levied International Monetary Fund, Government GC.TAX.YPKG.RV.ZS
on the actual or presumptive net income of Finance Statistics Yearbook and data
individuals, on the profits of corporations and files
enterprises, and on capital gains, whether realized or
not, on land, securities, and other assets.
Intragovernmental payments are eliminated in
consolidation.
Tax Composite index of the tax burden, including direct The Heritage Foundation, 2014–2018 Tax burden (or Fiscal
burden taxes, in terms of the top marginal tax rates on Freedom from 2016)
individual and corporate incomes, and overall taxes,
consisting of all forms of direct and indirect taxation
at all governmental levels, as a percentage of GDP.
This variable is defined on a 100-tax burden scale,
such that highest values correspond to lower tax
burden and vice versa. The variable is standardized
in regression analyses.
Bitcoin Indicator variable equal to 1 for countries identified as Hileman and Rauchs (2017) and Rauchs
mining having high levels of activity of bitcoin or et al. (2018)
country crypto-asset mining facilities
Bitcoin National aggregated capacity of facilities used for Hileman and Rauchs (2017)
mining crypto-asset mining measured in megawatts (MW)
MWs
Electricity Price of electricity in US cents per kWh. A monthly World Bank Doing Business Reports
price electricity consumption is assumed, for which a bill (2014–2018); “Getting Electricity”
is then computed for a warehouse based in the largest section
business city of the economy for the month of March
Stock The percentage U.S. dollar price change in the World Bank obtained from: Standard & CM.MKT.INDX.ZG
market country-specific stock markets covered by the Poor’s, Global Stock Markets Factbook
return S&P/IFCI and S&P/Frontier BMI country indices and supplemental S&P data.
(2014–2018)
Crisis The percentage U.S. dollar price change in the World Bank obtained from: Standard & CM.MKT.INDX.ZG
stock country-specific stock markets covered by the Poor’s, Global Stock Markets Factbook
market S&P/IFCI and S&P/Frontier BMI country indices and supplemental S&P data.
return for the years of 2008–2009
Global drivers of cryptocurrency infrastructure adoption 399

Table IA.VII Robustness tests of financial and criminality drivers excluding countries without Bitcoin infrastructure (Panel A; Country-
year level)

(1) (2) (3) (4)

Dependent variables Bitnode intensity > 0 Unique bitnodes > 0 Bitnode hours > 0 Bitcoin merchants > 0

Financial characteristics
Inflation crisis 0.656* 8.508** 63.253 1.690
(0.331) (3.109) (80.025) (1.276)
Unbanked 0.011 − 0.176* − 2.069 − 0.035+
(0.011) (0.088) (1.515) (0.018)
Five-bank asset concentration 0.009 0.040 − 4.496 **
− 0.014
(0.014) (0.064) (1.659) (0.027)
Criminality characteristics
Money laundering 1.392* 7.231** 181.439** 0.475
(0.607) (2.443) (57.945) (0.595)
Rule of law 1.538** 10.075*** 99.385+ 0.860+
(0.482) (2.493) (51.276) (0.513)
Baseline Controls
GDP per capita 0.557 − 0.709 58.254 0.048
(0.512) (1.749) (43.641) (0.435)
Restrictive regulation − 0.543* 3.012* 31.062 − 1.027
(0.257) (1.472) (45.992) (0.756)
Constant − 6.048 23.508 446.380 2.214
(4.790) (18.356) (438.980) (5.210)
Year FEs Yes Yes Yes Yes
World Region FEs Yes Yes Yes Yes
Year-World Region FEs Yes Yes Yes Yes
Panel Data Specifications Tobit Tobit Tobit Tobit
Observations 650 650 650 565
Number of Countries 130 130 130 113
Pseudo R-Squared 0.44 0.65 0.23 <0
Log-likelihood − 1301.6 − 2262.9 − 2382.4 − 1386.3
RHO 0.71 0.53 0.61 0.16

All models are Tobit random effects models and their associated overall R-squared is reported. Panel-data bootstrapped standard errors,
clustered at the country level, are reported in parentheses. The symbols ***, **, *, + mean that the reported coefficients are statistically
different from zero, respectively, at the 0.1, 1, 5 and 10% level. Model (3) is upper-censored at 750 (upper limit of active Bitnode hours in a
month)
400 E. Saiedi et al.

Table IA.VIII Robustness tests of social drivers excluding regions without Bitcoin infrastructure (Panel B; Region-year level)

(1) (2) (3) (4) (5) (6)

Dependent variables Bitnode Unique Bitcoin Bitnode Unique Bitcoin
intensity > 0 bitnodes > 0 merchants > 0 intensity > 0 bitnodes > 0 merchants > 0

Social characteristics
High risk willingness 4.414 20.920 14.357* 5.974 34.322 20.175**
(3.631) (23.771) (6.367) (3.670) (24.362) (6.985)
High risk willingness x − 0.605** − 5.402* − 2.012*
Bitcoin price
(0.211) (2.168) (0.991)
0.292
(0.178)
Bitcoin price 0.013 − 2.999*** 0.292
(0.039) (0.397) (0.178)
Trust in others 1.949 19.573* − 2.093 2.062+ 20.139* − 1.981
(1.219) (9.458) (2.656) (1.216) (9.449) (2.653)
** * ** *
Distrust in banks & financial 4.309 24.170 2.876 4.302 24.117 2.836
system (1.603) (12.305) (2.864) (1.599) (12.294) (2.861)
Distrust in other institutions − 3.376 − 24.477 − 6.207 − 3.409 − 24.462 − 6.181
(2.279) (16.552) (3.900) (2.275) (16.543) (3.897)
Baseline characteristics
GDP per capita − 1.911* − 33.624*** − 6.408 − 1.919* − 33.743*** − 6.181
(0.842) (8.510) (4.470) (0.837) (8.473) (4.453)
Restrictive regulation − 0.059 4.217 − 0.480 0.058 5.220 0.190
(0.406) (4.108) (2.246) (0.407) (4.126) (2.266)
Constant 17.323* 307.839*** 48.074 17.229* 308.580*** 45.291
(8.220) (76.940) (38.258) (8.183) (76.645) (38.128)
Year FEs Yes Yes Yes Yes Yes Yes
Country FEs Yes Yes Yes Yes Yes Yes
Panel data specifications Tobit Tobit Tobit Tobit Tobit Tobit
Observations 1210 1210 1000 1210 1210 1000
Number of Regions 242 242 200 242 242 200
Number of Countries 34 34 31 34 34 31
Log-likelihood − 2771.9 − 5182.3 − 2652.4 − 2767.7 − 5179.1 − 2650.3
Rho 0.77 0.55 0.08 0.77 0.55 0.08

All models are Tobit random effects models and their associated overall R-squared is reported. Panel-data bootstrapped standard errors,
clustered at the country level, are reported in parentheses. The symbols ***, **, *, + mean that the reported coefficients are statistically
different from zero, respectively, at the 0.1, 1, 5 and 10% level
Global drivers of cryptocurrency infrastructure adoption 401

Open Access This article is licensed under a Creative Commons Baek, C., & Elbeck, M. (2015). Bitcoins as an investment or
Attribution 4.0 International License, which permits use, sharing, speculative vehicle? A first look. Applied Economics
adaptation, distribution and reproduction in any medium or format, Letters, 22(1), 30–34 https://doi.org/10.1080
as long as you give appropriate credit to the original author(s) and /13504851.2014.916379.
the source, provide a link to the Creative Commons licence, and Bain & Company (2019). Many consumers trust technology com-
indicate if changes were made. The images or other third party panies more than banks. 9 January. Available at: https://www.
material in this article are included in the article's Creative Com- bain.com/insights/many-consumers-trust-technology-
mons licence, unless indicated otherwise in a credit line to the companies-morethan-banks-snap-chart/. Accessed 30
material. If material is not included in the article's Creative Com- August 2019.
mons licence and your intended use is not permitted by statutory Bank of International Settlements. (2019). BIS triennial central
regulation or exceeds the permitted use, you will need to obtain bank survey - foreign exchange turnover in April 2019.
permission directly from the copyright holder. To view a copy of Available at: https://www.bis.org/statistics/rpfx19_fx.pdf.
this licence, visit http://creativecommons.org/licenses/by/4.0/. Accessed 27 November 2019.
Barratt, M. J., Ferris, J. A., & Winstock, A. R. (2016). Safer
scoring? Cryptomarkets, social supply and drug market vio-
lence. International Journal of Drug Policy, 35, 24–31
References https://doi.org/10.1016/j.drugpo.2016.04.019.
Bashir, M., Strickland, B., & Bohr, J. (2016). What motivates
people to use bitcoin? In E. Sprio & Y. Ahn (Eds.), Social
Acquisti, A., Taylor, C., & Wagman, L. (2016). The economics of Informatics. In SocInfo 2016. Lecture notes in computer
privacy. Journal of Economic Literature, 54(2), 442–492 science (Vol. 10047, pp. 347–367). Cham: Springer
https://doi.org/10.1257/jel.54.2.442. https://doi.org/10.1007/978-3-319-47874-6_25.
Akins, B. W., Chapman, J. L., & Gordon, J. M. (2015). A whole Baur, D. G., Dimpfl, T., & Kuck, K. (2018a). Bitcoin, gold and the
new world: income tax considerations of the bitcoin econo- US dollar – a replication and extension. Finance Research
my. Pittsburgh Tax Review. https://doi.org/10.5195 Letters, 25, 103 –110 https://doi.org/1 0.10 16/J.
/taxreview.2014.32. FRL.2017.10.012.
Albrecht, C., Duffin, K. M., Hawkins, S., & Morales Rocha, V. M. Baur, D. G., Hong, K. H., & Lee, A. D. (2018b). Bitcoin: medium
(2019). The use of cryptocurrencies in the money laundering of exchange or speculative assets? Journal of International
process. Journal of Money Laundering Control, 22(2), 210– Financial Markets, Institutions and Money, 54, 177–189
216 https://doi.org/10.1108/JMLC-12-2017-0074. https://doi.org/10.1016/j.intfin.2017.12.004.
Aldridge, J., & Décary-Hétu, D. (2016a). Hidden wholesale: the Beck, T., & Martínez pería, M. S. (2011). What explains the price
drug diffusing capacity of online drug cryptomarkets. of remittances? An examination across 119 country corridors.
International Journal of Drug Policy, 35, 7–15 https://doi. World Bank Economic Review, 25(1), 105–131 https://doi.
org/10.1016/j.drugpo.2016.04.020. org/10.1093/wber/lhr017.
Aldridge, J., & Décary-Hétu, D. (2016b). Cryptomarkets and the Bertsch, C., Hull, I., Qi, Y., & Zhang, X. (2017). Bank misconduct,
future of illicit drug markets. In The internet and drug mar- trust, and online lending. Sveriges Riksbank working paper
kets (European Monitoring Centre for Drugs and Drug s e r i e s n o . 3 4 6 . h t t p : / / w w w. c h r i s t o p h b e r t s c h .
Addiction: Insights 21) (pp. 23–30). Luxembourg: com/data/TrustOnlineLending_BertschHullQiZhang.pdf.
Publications Office of the European Union.Available at: Accessed 27 November 2019.
https://daviddhetu.openum.ca/files/sites/39/2018/12/Future. Blau, B. M. (2018). Price dynamics and speculative trading in
pdf. Accessed 27 November 2019. Bitcoin. Research in International Business and Finance, 43,
Arnold, S., & Auer, B. R. (2015). What do scientists know about 15–21 https://doi.org/10.1016/j.ribaf.2017.07.183.
inflation hedging? The North American Journal of Block, J. H., Colombo, M. G., Cumming, D. J., & Vismara, S.
Economics and Finance, 34, 187–214 https://doi. (2018). New players in entrepreneurial finance and why they
org/10.1016/J.NAJEF.2015.08.005. are there. Small Business Economics, 50(2), 239–250.
Athey, S., Parashkevovz, I., Sarukkai, V., & Jing, X. (2016). https://doi.org/10.1007/s11187-016-9826-6.
Bitcoin Pricing, Adoption, and Usage: Theory and Bohannon, J. (2016). The bitcoin busts. Science, 351(6278),
Evidence. Stanford University Graduate School of Business 1144–1146. https://doi.org/10.1126/science.351.6278.1144.
Research Paper no. 16-42. Available at: https://ssrn. Böhme, R., Christin, N., Edelman, B., & Moore, T. (2015).
com/abstract=2826674 Bitcoin: economics, technology, and governance. Journal of
Athey, S., Catalini, C., & Tucker, C. E. (2017). The digital privacy Economic Perspectives, 29(2), 213–238. https://doi.
paradox: small money, small costs, small talk. SSRN Working org/10.1257/jep.29.2.213.
Paper. https://doi.org/10.2139/ssrn.2916489. Borio, C., Claessens, S., Cohen, B., Domanski, D., Halaburda, H.,
Audretsch, D. B., Heger, D., & Veith, T. (2015). Infrastructure and Hughes, K., et al. (2017). Central bank cryptocurrencies. BIS
entrepreneurship. Small Business Economics, 44(2), 219– quarterly review. Available at: https://www.bis.
230 https://doi.org/10.1007/s11187-014-9600-6. org/publ/qtrpdf/r_qt1709f.pdf. Accessed 11 November 2019.
Ba, S., & Pavlou, P. A. (2002). Evidence of the effect of trust Bouoiyour, J., & Selmi, R. (2015). What does Bitcoin look like?
building technology in electronic markets: price premiums Annals of Economics and Finance, 16(2), 449–492 Available
and buyer behavior. MIS Quarterly, 26(3), 243–268 at: https://ideas.repec.org/a/cuf/journl/y2015v16i2bouoiyour.
https://doi.org/10.2307/4132332. html.
402 E. Saiedi et al.

Bouri, E., Gupta, R., Tiwari, A. K., & Roubaud, D. (2017a). Does of the transformative potential of cryptomarkets for orga-
Bitcoin hedge global uncertainty? Evidence from wavelet- nized illicit drug crime. International Criminal Justice
based quantile-in-quantile regressions. Finance Research Review, 28(3), 255–274 https://doi.org/10.1177
Letters, 23, 87–95. https://doi.org/10.1016/j.frl.2017.02.009. /1057567718769719.
Bouri, E., Molnár, P., Azzi, G., Roubaud, D., & Hagfors, L. I. Demirguc-Kunt, A., Klapper, L., Singer, D., Ansar, S., & Hess, J.
(2017b). On the hedge and safe haven properties of Bitcoin: (2018). The global findex database 2017: measuring financial
is it really more than a diversifier? Finance Research Letters, inclusion and the fintech revolution. The World Bank.
20, 192–198 https://doi.org/10.1016/j.frl.2016.09.025. https://doi.org/10.1596/978-1-4648-1259-0.
Bouri, E., Lucey, B., & Roubaud, D. (2019a). Cryptocurrencies Dierksmeier, C., & Seele, P. (2018). Cryptocurrencies and busi-
and the downside risk in equity investments. Finance ness ethics. Journal of Business Ethics, 152(1), 1–14.
Research Letters, in press. https://doi.org/10.1016/j. https://doi.org/10.1007/s10551-016-3298-0.
frl.2019.06.009. Dodd, N. (2018). The social life of bitcoin. Theory, Culture &
Bouri, E., Shahzad, S. J. H., & Roubaud, D. (2019b). S o c i e t y, 3 5 ( 3 ) , 3 5 – 5 6 h t t p s : / / d o i . o r g / 1 0 . 11 7 7
Cryptocurrencies as hedges and safe-havens for US equity /0263276417746464.
sectors. In Quarterly review of economics and finance, in Dodgson, M., Gann, D., Wladawsky-Berger, I., Sultan, N., &
press. https://doi.org/10.1016/j.qref.2019.05.001. George, G. (2015). Managing digital money. Academy of
Campbell, J. Y. (2006). Household finance. The Journal of Management Journal, 58(2), 325–333 https://doi.
Finance, 61(4), 1553–1604 https://doi.org/10.1111/j.1540- org/10.5465/amj.2015.4002.
6261.2006.00883.x. Duarte, J., Siegel, S., & Young, L. (2012). Trust and credit: the role
Cheah, E.-T., & Fry, J. (2015). Speculative bubbles in Bitcoin of appearance in peer-to-peer lending. Review of Financial
markets? An empirical investigation into the fundamental Studies, 25(8), 2455–2484 https://doi.org/10.1093
value of Bitcoin. Economics Letters, 130, 32–36 https://doi. /rfs/hhs071.
org/10.1016/j.econlet.2015.02.029. Dwyer, G. P. (2015). The economics of bitcoin and similar private
Choo, K. K. R. (2015). Cryptocurrency and virtual currency: digital currencies. Journal of Financial Stability, 17, 81–91
corruption and money laundering/terrorism financing risks? https://doi.org/10.1016/j.jfs.2014.11.006.
In Handbook of digital currency: bitcoin, innovation, finan-
Dyhrberg, A. H. (2016). Bitcoin, gold and the dollar - a GARCH
cial instruments, and big data. https://doi.org/10.1016/B978-
volatility analysis. Finance Research Letters, 16, 85–92
0-12-802117-0.00015-1.
https://doi.org/10.1016/j.frl.2015.10.008.
CipherTrace (2018). Cryptocurrency anti-money laundering re-
Elgin, C., & Öztunalı, O. (2012). Shadow economies around the
p or t . Av a i l a b l e at : h t tp s :/ /c i ph e rt ra ce .c om /w p -
world: model based estimates. Working papers 2012/05,
content/uploads/2018/10/crypto_aml_report_2018q3.pdf.
Bogazici University, 1–48. Available at: http://www.econ.
Accessed 27 November 2019.
boun.edu.tr/public_html/RePEc/pdf/201205.pdf. Accessed
Cœuré, B. (2018). The new frontier of payments and market
27 November 2019.
infrastructure: on cryptos, cyber and CCPs. BIS. Available
at: https://www.bis.org/review/r181115a.htm. Accessed 27 Elliott, F., & Duncan, G. (2009). Chancellor Alistair Darling on
November 2019. brink of second bailout for banks. The Times, p. 1,2 and 6.
Cohen, B. (2017). The rise of alternative currencies in post-capi- London, UK. Available at: https://www.thetimes.co.
talism. Journal of Management Studies, 54(5), 739–746 uk/article/chancellor-alistair-darling-on-brink-of-second-
https://doi.org/10.1111/joms.12245. bailout-for-banks-n9l382mn62h. Accessed 27 November
Cohn, A., Engelmann, J., Fehr, E., & Maréchal, M. A. (2015). 2019.
Evidence for countercyclical risk aversion: an experiment Engert, W., & Fung, B. S. C. (2017). Central bank digital currency:
with financial professionals. American Economic Review, motivations and implications. Bank of Canada. Available at:
105(2), 860–885. https://doi.org/10.1257/aer.20131314. https://www.bankofcanada.ca/wp-content/uploads/2017/11
Conti-Brown, P., Lastra, R., Raskin, M., & Yermack, D. (2018). /sdp2017-16.pdf. Accessed 27 November 2019.
Digital currencies, decentralized ledgers and the future of Ermakova, T., Fabian, B., Baumann, A., Izmailov, M., &
central banking. In Research handbook on central banking. Krasnova, H. (2017). Bitcoin: drivers and impediments.
https://doi.org/10.4337/9781784719227.00028. SSRN Working Paper. https://doi.org/10.2139/ssrn.3017190.
Corbet, S., Cumming, D. J., Lucey, B. M., Peat, M., & Vigne, S. A. Estrin, S., Gozman, D., & Khavul, S. (2018). The evolution and
(2019). The destabilising effects of cryptocurrency adoption of equity crowdfunding: entrepreneur and investor
cybercriminality. Economics letters, in press. https://doi. entry into a new market. Small Business Economics, 51(2),
org/10.1016/j.econlet.2019.108741. 425–439. https://doi.org/10.1007/s11187-018-0009-5.
Correia, S. (2017). GEOCODE_IP: Stata module to geocode IP European Central Bank. (2019). Banknotes and coins circulation.
addresses. Statistical Software Components, (S458333). https://www.ecb.europa.eu/stats/policy_and_exchange_
Available at: https://ideas.repec.org/c/boc/bocode/s458333. rates/banknotes+coins/circulation/html/index.en.html.
html. Accessed 27 November 2019. Accessed 11 November 2019.
Czernich, N., Falck, O., Kretschmer, T., & Woessmann, L. (2011). Evans-Pughe, C., Novikov, A., & Vitaliev, V. (2014). To bit or not
Broadband infrastructure and economic growth. Economic to bit? [Bitcoin cryptocurrency]. Engineering & Technology,
Journal, 121(552), 505–532 https://doi.org/10.1111/j.1468- 9(4), 82–85. https://doi.org/10.1049/et.2014.0411.
0297.2011.02420.x. Eyal, I., & Sirer, E. G. (2014). Majority is not enough: Bitcoin
Demant, J., Munksgaard, R., Décary-Hétu, D., & Aldridge, J. mining is vulnerable. In Lecture notes in computer science
(2018). Going local on a global platform: a critical analysis (including subseries lecture notes in artificial intelligence
Global drivers of cryptocurrency infrastructure adoption 403

and lecture notes in bioinformatics). https://doi.org/10.1007 Griffin, J. M., & Shams, A. (2019). Is bitcoin really un-tethered?
/978-3-662-45472-5_28. Journal of Finance, forthcoming. https://doi.org/10.2139
Fanusie, Y. J., & Robinson, T. (2018). Bitcoin laundering: an /ssrn.3195066.
analysis of illicit flows into digital currency services. Grohmann, A., Klühs, T., & Menkhoff, L. (2018). Does financial
Washington, DC. Available at: https://cdn2.hubspot. literacy improve financial inclusion? Cross country evidence.
net/hubfs/3883533/downloads/Bitcoinlaundering.pdf. World Development, 111, 84–96. https://doi.org/10.1016/j.
Accessed 27 November 2019. worlddev.2018.06.020.
Fatás, A., & Weder, B. (2019). Initial coin offerings: fundamen- Gross, M., Hogarth, J., & Schmeiser, M. (2012). Use of financial
tally different but highly correlated. Available at: services by the unbanked and underbanked and the potential
h t t p s : / / v o x e u . o rg / a r t i c l e / i n i t i a l - c o i n - o ff e r i n g s - for mobile financial services adoption. Federal Reserve
fundamentally-different-highly-correlated. Accessed 27 Bulletin, 98(4), 20. Available at: https://www.federalreserve.
November 2019. gov/pubs/bulletin/2012/pdf/mobile_financial_
FATF (2014). Virtual currencies – key definitions and potential services_201209.pdf. Accessed 24 December 2018.
AML/CFT risks. FATF report. Available at: https://www.fatf- Guiso, L., Sapienza, P., & Zingales, L. (2018). Time varying risk
gafi.org/media/fatf/documents/reports/Virtual-currency-key- aversion. Journal of Financial Economics, 128(3), 403–421.
definitions-and-potential-aml-cft-risks.pdf. Accessed 27 https://doi.org/10.1016/j.jfineco.2018.02.007.
November 2019. Haddad, C., & Hornuf, L. (2019). The emergence of the global
Fisch, C. (2019). Initial coin offerings (ICOs) to finance new fintech market: economic and technological determinants.
ventures. Journal of Business Venturing, 34(1), 1–22. Small Business Economics, 53(1), 81–105. https://doi.
https://doi.org/10.1016/J.JBUSVENT.2018.09.007. org/10.1007/s11187-018-9991-x.
Fisch, C., Masiak, C., Vismara, S., & Block, J. (2019). Motives Heilman, E., Kendler, A., & Zohar, A. (2015). Eclipse attacks on
and profiles of ICO investors. Journal of Business Research, bitcoin’s peer-to-peer network. In Proceedings of the 24th
in press. https://doi.org/10.1016/j.jbusres.2019.07.036. USENIX Security Symposium (pp. 129–144). Available at:
Foley, S., Karlsen, J. R., & Putniņš, T. J. (2019). Sex, drugs, and https://www.usenix.org/system/files/conference/usenixsecurity15
bitcoin: how much illegal activity is financed through /sec15-paper-heilman.pdf. Accessed 9 December 2018.
cryptocurrencies? The Review of Financial Studies, 32(5), Henfridsson, O., & Bygstad, B. (2013). The generative mecha-
1798–1853. https://doi.org/10.1093/rfs/hhz015. nisms of digital infrastructure evolution. MIS Quarterly,
Folkinshteyn, D., Lennon, M., & Reilly, T. (2015). A tale of twin 37(3; special issue: Critical realism in IS research), 907–
tech: bitcoin and the WWW. Journal of Strategic and 931. Available at: https://www.jstor.org/stable/43826006.
International Studies, X(2), 82–89 Available at: https://ssrn. Accessed 20 August 2019.
com/abstract=2601617. Henry, C. S., Huynh, K. P., & Nicholls, G. (2018). Bitcoin aware-
Forbes (2015). Americans trust tech firms more than banks for ness and usage in Canada. Journal of Digital Banking, 2(4),
finance. 24 June. Available at https://www.forbes. 311–337 Available at: https://www.ingentaconnect.
com/sites/niallmccarthy/2015/06/24/americans-trust-tech- com/content/hsp/jdb001/2018/00000002/00000004
firms-more-than-banks-for-finance-infographic/#583802874 /art00004.
e94. Accessed 30 August 2019. Higgins, S. (2018). Financial technology adoption. Available at:
Fry, J. (2018). Booms, busts and heavy-tails: the story of bitcoin https://seankhiggins.com/assets/pdf/higgins_jmp.pdf.
and cryptocurrency markets? Economics Letters, 171, 225– Accessed 27 November 2019.
229. https://doi.org/10.1016/j.econlet.2018.08.008. Hileman, G., & Rauchs, M. (2017). Global cryptocurrency
Gandal, N., & Halaburda, H. (2014). Competition in the benchmarking study. https://doi.org/10.2139/ssrn.2965436.
cryptocurrency market. CEPR discussion paper no. Hoffman, P. T., Postel-Vinay, G., & Rosenthal, J.-L. (2009). The
DP10157. Available at: https://ssrn.com/abstract=2501640 role of trust in the long-run development of french financial
Gazel, M., & Schwienbacher, A. (2018). Entrepreneurial fintech markets. In K. S. Cook, M. Levi, & R. Hardin (Eds.), Who
clusters. SSRN Working Paper. https://doi.org/10.2139 can we trust? How groups, networks and institutions make
/ssrn.3309067. trust possible (pp. 249–285). New York: Russell Sage
Giannetti, M., & Wang, T. Y. (2016). Corporate scandals and Foundation.
household stock market participation. Journal of Finance, Holt, T. J., Smirnova, O., Chua, Y. T., & Copes, H. (2015).
71(6), 2591–2636. https://doi.org/10.1111/jofi.12399. Examining the risk reduction strategies of actors in online
Glaser, F., Zimmermann, K., Haferkorn, M., Weber, M. C., & criminal markets. Global Crime, 16(2), 81–103. https://doi.
Siering, M. (2014). Bitcoin - asset or currency? Revealing org/10.1080/17440572.2015.1013211.
users’ hidden intentions. In Twenty second european confer- Hornuf, L., Klus, M. F., Lohwasser, T. S., & Schwienbacher, A.
ence on information systems (ECIS). https://doi.org/10.1007 (2018). How do banks interact with fintechs? Forms of
/978-3-319-42448-4_6. strategic alliance and their economic impact. SSRN working
Goldfarb, A., & Tucker, C. (2019). Digital economics. Journal of paper. Available at: https://ssrn.com/abstract=3252318
Economic Literature, 57(1), 3–43. https://doi.org/10.1257 Hu, A., Parlour, C. A., & Rajan, U. (2018). Cryptocurrencies:
/jel.20171452. stylized facts on a new investible instrument. SSRN
Greenstein, S. M., & Spiller, P. T. (1996). Estimating the welfare Working Paper. https://doi.org/10.2139/ssrn.3182113.
effects of digital infrastructure. In National Bureau of Huang, W., Meoli, M., & Vismara, S. (2019). The geography of
Economic Research Working Paper No. 5770. https://doi. initial coin offerings. Small Business Economics, 1–26.
org/10.3386/w5770. https://doi.org/10.1007/s11187-019-00135-y.
404 E. Saiedi et al.

Iansiti, M., & Lakhani, K. R. (2017). The truth about blockchain. Lagarde, C. (2018). Winds of change: the case for new digital
Harvard Business Review. Available at: https://hbr.org/2017 currency. Singapore: International Monetary Fund. Available
/01/the-truth-about-blockchain at: https://www.imf.org/en/News/Articles/2018/11/13
Ingram Bogusz, C., & Morisse, M. (2018). How infrastructures /sp111418-winds-of-change-the-case-for-new-digital-
anchor open entrepreneurship: the case of Bitcoin and stigma. currency. .
Information Systems Journal, 28(6), 1176–1212. https://doi. Larios-Hernández, G. J. (2017). Blockchain entrepreneurship op-
org/10.1111/isj.12204. portunity in the practices of the unbanked. Business
Iyidogan, E. (2018). An equilibrium model of blockchain-based Horizons, 60(6), 865–874 https://doi.org/10.1016/j.
cryptocurrencies. SSRN Working Paper. https://doi. bushor.2017.07.012.
org/10.2139/ssrn.3152803. Lavorgna, A. (2014). Internet-mediated drug trafficking: towards a
J5. (2018). Tax enforcement authorities unite to combat interna- better understanding of new criminal dynamics. Trends in
tional tax crime and money laundering. In Montreal Organized Crime, 17(4), 250–270. https://doi.org/10.1007
https://www.irs.gov/pub/irs-utl/j5-media-release-7-2-18.pdf. /s12117-014-9226-8.
Kahn, C. M., & Roberds, W. (2008). Credit and identity theft. Law Library of Congress. (2018). Regulation of cryptocurrency
Journal of Monetary Economics, 55(2), 251–264. https://doi. around the world. Available at: https://www.loc.
org/10.1016/j.jmoneco.2007.08.001. gov/law/help/cryptocurrency/cryptocurrency-world-survey.
Kahn, C. M., McAndrews, J., & Roberds, W. (2005). Money is pdf. Accessed 27 November 2019.
privacy. International Economic Review, 46(2), 377–399. Li, T., Shin, D., & Wang, B. (2018). Cryptocurrency pump-and-
https://doi.org/10.1111/j.1468-2354.2005.00323.x. dump schemes. SSRN Working Paper. https://doi.
Kamps, J., & Kleinberg, B. (2018). To the moon: defining and org/10.2139/ssrn.3267041.
detecting cryptocurrency pump-and-dumps. Crime Science, Liang, T.-P., Wu, S. P.-J., & Huang, C. (2019). Why funders invest
7(1), 18. https://doi.org/10.1186/s40163-018-0093-5. in crowdfunding projects: role of trust from the dual-process
Kaskaloglu, K. (2014). Near zero bitcoin transaction fees cannot perspective. Information & Management, 56(1), 70–84.
last forever. In The International Conference on Digital https://doi.org/10.1016/J.IM.2018.07.002.
Security and Forensics (DigitalSec2014). Available at: Libra Association. (2019). An introduction to libra. Available at:
http://sdiwc.net/digital-library/near-zero-bitcoin-transaction-
https://libra.org/en-us/whitepaper. Accessed 30 August,
fees-cannot-last-forever.html. Accessed 27 November 2019. 2019.
Kethineni, S., & Cao, Y. (2019). The rise in popularity of
Luther, W. J. (2016). Cryptocurrencies, network effects, and
cryptocurrency and associated criminal activity.
switching costs. Contemporary Economic Policy, 34(3),
International Criminal Justice Review, 1–20. https://doi.
553–571. https://doi.org/10.1111/coep.12151.
org/10.1177/1057567719827051.
Marian, O. (2013). Are cryptocurrencies super tax havens? Mich.
Kewell, B., Adams, R., & Parry, G. (2017). Blockchain for good?
L. Rev. First Impressions, 112(1), 38–48. https://ssrn.
Strategic Change, 26(5), 429–437. https://doi.org/10.1002
com/abstract=2305863.
/jsc.2143.
Killeen, A. (2015). The confluence of bitcoin and the global Martino, P., Bellavitis, C., & DaSilva, C. M. (2019). Blockchain
sharing economy. In Handbook of digital currency (pp. and initial coin offerings (ICOs): a new way of
485–503). D. L. K: Elsevier. Edited by Chuen. https://doi. crowdfunding. SSRN Working Paper. https://doi.
org/10.1016/B978-0-12-802117-0.00024-2. org/10.2139/ssrn.3414238.
Kim, T. (2017). On the transaction cost of bitcoin. Finance Masiak, C., Block, J. H., Masiak, T., Neuenkirch, M., & Pielen, K.
Research Letters, 23, 300–305. https://doi.org/10.1016/j. N. (2019). Initial coin offerings (ICOs): market cycles and
frl.2017.07.014. relationship with bitcoin and ether. Small Business
Kliber, A., Marszałek, P., Musiałkowska, I., & Świerczyńska, K. Economics, 1–18. https://doi.org/10.1007/s11187-019-
(2019). Bitcoin: safe haven, hedge or diversifier? Perception 00176-3.
of bitcoin in the context of a country’s economic situation — Maurer, B., Nelms, T. C., & Swartz, L. (2013). “When perhaps the
A stochastic volatility approach. Physica A: Statistical real problem is money itself!”: the practical materiality of
Mechanics and its Applications. https://doi.org/10.1016/j. bitcoin. Social Semiotics, 23(2), 261–277. https://doi.
physa.2019.04.145. org/10.1080/10350330.2013.777594.
Klus, M., Lohwasser, T., Holotiuk, F., & Moormann, J. (2019). McGinn, D., Birch, D., Akroyd, D., Molina-Solana, M., Guo, Y.,
Strategic alliances between banks and Fintechs for digital & Knottenbelt, W. J. (2016). Visualizing dynamic bitcoin
innovation: Motives to collaborate and types of interaction. transaction patterns. Big Data, 4(2), 109–119. https://doi.
The Journal of Entrepreneurial Finance, 21(1), 1–23. org/10.1089/big.2015.0056.
Available at: https://digitalcommons.pepperdine. Miller, A., Litton, J., Pachulski, A., Gupta, N., Levin, D., Spring,
edu/jef/vol21/iss1/1/. Accessed 2 September 2019. N., & Bhattacharjee, B. (2018). Discovering bitcoin’s public
Komosny, D., Voznak, M., & Rehman, U. S. (2017). Location topology and influential nodes. Available at: https://cs.umd.
accuracy of commercial IP address geolocation databases. edu/projects/coinscope/coinscope.pdf. Accessed 27
Journal of Information Technology and Control, 46(3), 333– November 2019.
344. https://doi.org/10.5755/j01.itc.46.3.14451. Moser, M., Bohme, R., & Breuker, D. (2013). An inquiry into
Krause, M. J., & Tolaymat, T. (2018). Quantification of energy and money laundering tools in the bitcoin ecosystem. In 2013
carbon costs for mining cryptocurrencies. Nature APWG eCrime researchers summit (pp. 1–14). San
Sustainability, 1(11), 711–718. https://doi.org/10.1038 Fransisco: IEEE. https://doi.org/10.1109
/s41893-018-0152-7. /eCRS.2013.6805780.
Global drivers of cryptocurrency infrastructure adoption 405

Nair, S. (2019). Trust in tech is wavering and companies must act. and commodities? International Review of Financial
In Edelman. https://www.edelman.com/research/2019-trust- Analysis, 63, 322–330. https://doi.org/10.1016/j.
tech-wavering-companies-must-act. irfa.2019.01.002.
Nakamoto, S. (2008). Bitcoin: a peer-to-peer electronic cash sys- Shavitt, Y., & Zilberman, N. (2011). A geolocation databases
tem. https://bitcoin.org/bitcoin.pdf. Accessed 9 December study. IEEE Journal on Selected Areas in Communications,
2018. 2 9 (1 0), 2044 – 2 0 5 6 . h t t p s : / / d o i . o rg / 1 0 . 11 0 9
NDIC. (2008). Money laundering in digital currencies. /JSAC.2011.111214.
Washington, DC. Available at: https://www.justice. Shiller, R. (2019). Bitcoin narratives. In Narrative economics: how
gov/archive/ndic/pubs28/28675/28675p.pdf. Accessed 27 stories go viral and drive major economic events (1st ed., pp.
November 2019. 3–11). Princeton, NJ, USA: Princeton University Press.
Niepelt, D. (2020). Reserves for all? Central bank digital currency, Shin, L. (2017). Here’s the man who created ICOs and this is the
deposits, and their (non)-equivalence. International Journal new token he’s backing. Forbes. https://www.forbes.
of Central Banking, forthcoming. Available at: https://ssrn. com/sites/laurashin/2017/09/21/heres-the-man-who-created-
com/abstract=3254206 icos-and-this-is-the-new-token-hes-backing.
Ormsby, E. (2016). Silk road: insights from interviews with users Smith, K. (2019). Darknet transactions up 70% — with BTC still
and vendors. In J. Mounteney, A. Oteo, & P. Griffiths (Eds.), the currency of choice. Available at: https://bravenewcoin.
The Internet and drug markets (European monitoring Centre com/insights/darknet-transactions-up-70-with-btc-still-the-
for Drugs and Drug Addiction: Insights 21) (pp. 61–67). currency-of-choice. Accessed 27 November 2019.
Luxembourg: Publications Office of the European Union. Soska, K., & Christin, N. (2015). Measuring the longitudinal evolution
Philippon, T. (2016). The fintech opportunity. NBER working of the online anonymous marketplace ecosystem. In Proceedings
paper no. 22476. Available at: http://pages.stern.nyu. of the 24th USENIX security symposium Available at:
edu/~tphilipp/papers/FinTech.pdf. Accessed 6 March 2019. https://www.usenix.org/system/files/conference/usenixsecurity15
Poese, I., Uhlig, S., Kaafar, M. A., Donnet, B., & Gueye, B. /sec15-paper-soska-updated.pdf.
(2011). IP geolocation databases. ACM SIGCOMM Spigel, B. (2017). The relational organization of entrepreneurial
Computer Communication Review, 41(2), 53. https://doi. ecosystems. Entrepreneurship Theory and Practice, 41(1),
org/10.1145/1971162.1971171. 49–72. https://doi.org/10.1111/etap.12167.
Rau, P. R. (2017). Law, trust, and the development of crowdfunding.
Stigler, G. J. (1971). The theory of economic regulation. The Bell
Available at: https://ssrn.com/abstract=2989056
Journal of Economics and Management Science, 2(1), 3.
Rauchs, M., Blandin, A., Klein, K., Pieters, G., Recanatini, M., &
https://doi.org/10.2307/3003160.
Zhang, B. (2018). 2nd global cryptoasset benchmarking
Su, J. (2019). Hackers stole over $4 billion from crypto crimes in
s t u d y. Av a i l a b l e a t : h t t p s : / / w w w. j b s . c a m . a c .
2019 so far, up from $1.7 billion in all of 2018. Forbes.
uk/fileadmin/user_upload/research/centres/alternative-
https://www.forbes.com/sites/jeanbaptiste/2019/08/15
finance/downloads/2018-12-ccaf-2nd-global-cryptoasset-
/hackers-stole-over-4-billion-from-crypto-crimes-in-2019-
benchmarking.pdf. Accessed 2 May 2019.
so-far-up-from-1-7-billion-in-all-of-2018/
Raymaekers, W. (2015). Cryptocurrency bitcoin: disruption, chal-
lenges and opportunities. Journal of Payments Strategy & Subramanian, R., & Chino, T. (2015). The state of
Systems, 9(1), 30–46 Available at: https://www. cryptocurrencies, their issues and policy interactions.
ingentaconnect.com/content/hsp/jpss/2015/00000009 Journal of International Technology and Information
/00000001/art00005. M a n a g e m e n t , 2 4 ( 3 ) , 2 5 – 4 0 Av a i l a b l e a t :
Reinhart, C. M., & Rogoff, K. S. (2011). From financial crash to https://scholarworks.lib.csusb.edu/jitim/vol24/iss3/2.
debt crisis. American Economic Review, 101(5), 1676–1706. Sussan, F., & Acs, Z. J. (2017). The digital entrepreneurial eco-
https://doi.org/10.1257/aer.101.5.1676. system. Small Business Economics, 49(1), 55–73. https://doi.
Sadhya, V., & Sadhya, H. (2018). Barriers to adoption of org/10.1007/s11187-017-9867-5.
blockchain technology. Available at: https://aisel.aisnet. Svetlana, S., Kokkinaki, A., & Ifigenia, G. (2017). In which
org/amcis2018/AdoptionDiff/Presentations/20/. Accessed distributed ledger do we trust? A comparative analysis of
27 February 2019. cryptocurrencies. In MCIS 2017 proceedings, 21. (p. 13).
Saiedi, E., Mohammadi, A., & Broström, A. (2017). Distrust in Available at: http://aisel.aisnet.org/mcis2017/21
banks and fintech adoption: the case of P2P loans. Swedish Swartz, L. (2018). What was bitcoin, what will it be? The techno-
House of Finance research paper no. 18-6. https://doi. economic imaginaries of a new money technology. Cultural
org/10.2139/ssrn.3124301 Studies, 32(4), 623–650. https://doi.org/10.1080
Sameeh, T. (2018). ICO basics – The difference between security /09502386.2017.1416420.
tokens and utility tokens. Coinintelligence. Available at: Tasca, P. (2015). Digital currencies: principles, trends, opportuni-
https://www.cointelligence.com/content/ico-basics-security- ties, and risks. SSRN Working Paper. https://doi.org/10.2139
tokens-vs-utility-tokens/. . /ssrn.2657598.
Schuh, S., & Shy, O. (2016). US consumers’ adoption and use of Teo, E. G. S. (2015). Emergence, growth, and sustainability of
bitcoin and other virtual currencies. Federal Reserve Bank of bitcoin. In D. L. K. Chuen (Ed.), Handbook of digital cur-
Boston Working Paper. Available at: https://www. rency (pp. 191–200). Elsevier. https://doi.org/10.1016/B978-
banqueducanada.ca/wp-content/uploads/2015/12/us- 0-12-802117-0.00009-6.
consumers-adoption.pdf. Accessed 8 January 2019. Thakor, A. V. (2019). Fintech and banking. Journal of Financial
Shahzad, S. J. H., Bouri, E., Roubaud, D., Kristoufek, L., & Lucey, Intermediation, in press. https://doi.org/10.2139
B. (2019). Is bitcoin a better safe-haven investment than gold /ssrn.3332550.
406 E. Saiedi et al.

Thakor, R., & Merton, R. (2018). Trust in lending. In NBER van Wegberg, R., Oerlemans, J.-J., & van Deventer, O. (2018).
working paper no. 24778. https://doi.org/10.3386/w24778. Bitcoin money laundering: mixed results? An explorative
The Economist (2018). Chips off the block: from one study on money laundering of cybercrime proceeds using
cryptocurrency to thousands. The Economist’s technology bitcoin. Journal of Financial Crime, 25(2), 419–435.
quarterly. Available at: https://www.economist. https://doi.org/10.1108/JFC-11-2016-0067.
com/technology-quarterly/2018/09/01/from- Vigna, P., & Casey, M. J. (2015). The age of cryptocurrency: how
onecryptocurrency-to-thousands. Accessed 27 November bitcoin and digital money are challenging the global eco-
2019. nomic order. New York, N.Y: St. Martin’s Press.
Tilson, D., Lyytinen, K., & Sørensen, C. (2010). Digital infrastruc- Wang, L., & Pustogarov, I. (2017). Towards better understanding
tures: the missing IS research agenda. Information Systems of bitcoin unreachable peers. Available at: https://arxiv.
Research, 21(4), 748–759. https://doi.org/10.1287 org/pdf/1709.06837.pdf. Accessed 27 November 2019.
/isre.1100.0318. Wang, K., Bellavitis, C., & Da Silva, C. M. (2019). An introduc-
U.S. Federal Reserve Board. (2019). How much U.S. currency is tion to blockchain, cryptocurrency and initial coin offerings.
in circulation? Available at: https://www.federalreserve. In A. Quas, Y. Alperovych, C. Bellavitis, I. Paeleman, & D.
gov/faqs/currency_12773.htm. Accessed 11 November S. Kamuriwo (Eds.), New frontiers in entrepreneurial finance
2019. research (pp. 1–21). Singapore: World Scientific. https://doi.
UNODC. (2019). World drug report 2019: global overview of org/10.1142/11344.
drug demand and supply. https://doi.org/10.18356/bdc264 Wei, W. C. (2018). Liquidity and market efficiency in
f4-en. cryptocurrencies. Economics Letters, 168, 21–24.
Van Alstyne, M. (2014). Why bitcoin has value. Communications https://doi.org/10.1016/J.ECONLET.2018.04.003.
of the ACM, 57(5), 30–32. https://doi.org/10.1145/2594288. Winstock, A., Barratt, M., Maier, L., & Ferris, J. (2018). Global
Van Hout, M. C., & Bingham, T. (2014). Responsible vendors, drug survey (GDS) 2018. Key findings report. Available at:
intelligent consumers: silk road, the online revolution in drug https://www.globaldrugsurvey.com/gds-2018/. Accessed 21
trading. International Journal of Drug Policy, 25(2), 183– February 2019.
189. https://doi.org/10.1016/j.drugpo.2013.10.009. Xue, M., Hitt, L. M., & Chen, P. (2011). Determinants and out-
van Rooij, M., Lusardi, A., & Alessie, R. (2011). Financial literacy comes of Internet banking adoption. Management Science,
and stock market participation. Journal of Financial 57(2), 291–307. https://doi.org/10.1287/mnsc.1100.1187.
Economics, 101(2), 449–472. https://doi.org/10.1016/j. Yermack, D. (2015). Is bitcoin a real currency? An economic
jfineco.2011.03.006. appraisal. In D. L. K. Chuen (Ed.), Handbook of digital
van Rooij, M. C. J., Lusardi, A., & Alessie, R. J. M. (2012). currency (pp. 31–43). Amsterdam: Elsevier. https://doi.
Financial literacy, retirement planning and household wealth. org/10.1016/B978-0-12-802117-0.00002-3.
The Economic Journal, 122(560), 449–478. https://doi. Yermack, D. (2018). FinTech in sub-saharan Africa: what has
org/10.1111/j.1468-0297.2012.02501.x. worked well, and what hasn’t. In NBER working paper no.
van Slobbe, J. (2016). The drug trade on the deep web: a law 25007. https://doi.org/10.3386/w25007.
enforcement perspective. In J. Mounteney, A. Oteo, & P.
Griffiths (Eds.), The Internet and drug markets (European Publisher’s note Springer Nature remains neutral with regard to
Monitoring Centre for Drugs and Drug Addiction: Insights jurisdictional claims in published maps and institutional
21) (pp. 77–83). Publications Office of the European Union. affiliations.
Small Business Economics is a copyright of Springer, 2021. All Rights Reserved.