What drives bitcoin adoption by retailers?

Abstract

Decentralised issued crypto “currencies”, like bitcoin, have the potential to drastically change the existing retail payment system and even the monetary system. Insights into the factors that influence their adoption are therefore crucial. Using a large representative sample of retailers that sell their products online, we find that acceptance of crypto payments is currently modest (2%), but there is substantial interest among retailers to adopt crypto payments in the near future. Consumer demand, net transactional benefits and perceived adoption effort influence adoption intention and actual acceptance by retailers. Regarding non-financial factors, our findings suggest that service providers who act as intermediaries between retailers, their customers, and providers of payment instruments play a crucial role as facilitators of competition and innovation in the online retail payments market by lowering such barriers. The most serious barrier for crypto acceptance seems to be a lack of consumer demand. Information from consumers ndicate that those who possess cryptos, don’t use it for online payments.It seems therefore unlikely that the adoption of cryptos by retailers will increase substantially, making it highly unlikely that cryptos like bitcoin will drastically change the existing retail payment system.

Full text

No. 585 / February 2018
What drives bitcoin adoption by
retailers?
Nicole Jonker

De Nederlandsche Bank NV
P.O. Box 98
1000 AB AMSTERDAM
The Netherlands
Working Paper No. 585
February 2018
What drives bitcoin adoption by retailers?
Nicole Jonker *
* Views expressed are those of the author and do not necessarily reflect official positions of
De Nederlandsche Bank.

What drives bitcoin adoption by retailers?
*
Nicole Jonkera
a De Nederlandsche Bank, Payments and Market Infrastructures Division
February 2018
Abstract
Decentralised issued crypto “currencies”, like bitcoin, have the potential to drastically change the
existing retail payment system and even the monetary system. Insights into the factors that influence
their adoption are therefore crucial. Using a large representative sample of retailers that sell their
products online, we find that acceptance of crypto payments is currently modest (2%), but there is
substantial interest among retailers to adopt crypto payments in the near future. Consumer demand, net
transactional benefits and perceived adoption effort influence adoption intention and actual acceptance
by retailers. Regarding non-financial factors, our findings suggest that service providers who act as
intermediaries between retailers, their customers, and providers of payment instruments play a crucial
role as facilitators of competition and innovation in the online retail payments market by lowering such
barriers. The most serious barrier for crypto acceptance seems to be a lack of consumer demand.
Information from consumers indicate that those who possess cryptos, don’t use it for online payments.
It seems therefore unlikely that the adoption of cryptos by retailers will increase substantially, making
it highly unlikely that cryptos like bitcoin will drastically change the existing retail payment system.
Keywords: bitcoin, cryptocurrency, technology adoption, two-sided markets, retailers, network
externalities, cost, facilitating conditions.
JEL classifications: D22, E42, G20, O33.
*
Corresponding author: Nicole Jonker, phone: +31-20-5242759, email: n.jonker@dnb.nl, address: De Nederlandsche Bank,
Payments and Market Infrastructures Division, P.O. Box 98, 1000AB Amsterdam, The Netherlands. The author thanks Bas
Koolstra and Monique Timmermans for their help in the early stage of this research, Hans Brits, Carin van der Cruijsen, Daniel
Heller, Mirjam Plooij and participants of the DNB research seminar and the ECB-Banca d’Italia joint conference “Digital
transformation of the retail payments ecosystem” for their valuable comments on earlier versions of this paper, Gareth Budden
and Robert Weaver for linguistic services and Marianne van Marwijk and Jaap Wils of research company Panteia for their
help in collecting the data. All remaining errors are my own. The views expressed are mine and do not necessarily reflect those
of the Nederlandsche Bank or the European System of Central Banks.

2
1. INTRODUCTION
This paper examines the adoption intention and actual acceptance of cryptocurrency payments like
bitcoin by online retailers. Nakamoto (2008) introduced the world’s first decentralised crypto currency,
called bitcoin. Since cryptocurrencies do not fulfil all the functions of money, we use the term “crypto”
in the rest of this paper instead of the term cryptocurrency.
1
Cryptos like bitcoin represent a new payment
technology, which enable payers and payees to directly send value to each other electronically and
anonymously without the need to use the services of trusted third parties, like financial institutions
(Nakamoto, 2008). This allows them to move outside the scope of the traditional retail payment market
with its regulated payment service providers. Instead a peer-to-peer network is used, consisting of nodes
of computer systems, which provide the computer power needed to run the software for the network.
The main novelty of these networks is that they have implemented the distributed ledger technology,
which uses cryptographic techniques for the identification and validation of payments by network nodes;
that are subsequently recorded decentrally in a public distributed ledger, called the blockchain. Since
2009 also others launched (decentralised) cryptos inspired by bitcoin, and its payment technology, of
which Ethereum, Litecoin, and Ripple are well-known examples.
2
There are about 900 cryptos with a
value of USD 342 billion, which corresponds with 0.3 percent of global GDP (WorldCoinIndex, 5
February 2018).
Since the introduction of bitcoin, cryptos have received considerable media attention worldwide, fuelled
by the sharp appreciation of major cryptocurrencies like bitcoin compared to regular currencies, and the
fluctuations therein, the close links they have with the shadow economy, but also because of the question
of whether they pose a serious threat to regular currencies. It was thought that cryptos had the potential
to drastically change the existing retail payment ecosystem by making traditional financial institutions
like banks, which act as intermediaries between consumers and retailers, superfluous. Moreover, it was
thought that, if they were to be used on a large scale, they could even affect the functioning of the
monetary system (Halperin, 2013; Stevens, 2017). They are therefore of interest to economists and
central bankers. Furthermore, using cryptos also entails risks for payers and payees. The network's
decentralised nature obscures its members' responsibilities, meaning that none of them can be held
accountable in the event of irregularities. In addition, payments and holdings in cryptos of consumers
are not covered by a deposit guarantee scheme, nor can consumers rely on a compensation policy in
case of fraud.
1
In this paper we do not consider cryptocurrencies as money. According to the economic literature a cryptocurrency should
not be considered as money, as it does not fulfil the three functions of money, i.e. 1) medium of exchange, 2) store of value
and 3) unit of account. Thus far, cryptocurrencies fulfil the role of medium of exchange to a limited extent as the adoption
and usage rate among consumers and retailers is very low. Cryptocurrencies are hardly suited to fulfilling the other two roles
due to the high volatility of their exchange rates relative to regular currencies, which causes huge fluctuations in the
purchasing power of savings and in consumer prices of goods and services.
2
For more information on the technology behind bitcoin, see Nakamoto (2008) and about decentralised and centralised
cryptos in general, see e.g. ECB (2015).

3
Insight into the factors which influence the adoption of such potentially disruptive payment technologies
are therefore highly relevant. However, research on the adoption of cryptos as a means of payment by
users is still in its infancy. Schuh and Shy (2015) and Silinskyte (2014) study the adoption and usage of
cryptos among consumers, while Polasik, Piotrowska, Wisniewski, Kotkowski and Lightfoot (2015)
shed light on the features of crypto accepting vendors.
However, as far as we know, there are no studies available on the adoption of cryptos among a large
diverse group of online retailers. This paper fills that gap. Another novelty is that we enrich the economic
literature with insights from other disciplines to analyse adoption decisions by retailers. Such an
approach is supported by an increasing number of economists (see e.g. Hoff and Stiglitz, 2016) and is
shown to be successful in the payment literature (see e.g. Van der Cruijsen and Van der Horst, 2016).
Given the technical complexity and the highly innovative features of cryptos, the technology adoption
literature seems to be a natural source from which to borrow insights. We address the following research
question: Which factors influence the retailer’s adoption of cryptos like bitcoin? In our analyses we pay
attention to the influence of consumer demand for crypto payments, transactional benefits of receiving
crypto payments relative to other means of payment, and non-financial barriers to retailers’ adoption
intention and actual acceptance of crypto payments.
In November and December 2016 we conducted a survey among 768 retailers who sell their products
online to consumers inside (and outside) the Netherlands. We polled these retailers about their business,
the acceptance of payment methods, their perceptions regarding crypto payments as well as mainstream
online payment methods, their attitudes towards cryptos and their intention to adopt them as a means of
payment. We use the resulting rich dataset to answer our research question. The Netherlands provide a
good setting for this research, as it has a well-developed online retail market. The total value of online
payments was EUR 20 billion in 2016 (Thuiswinkel.org, 2017) which corresponds with 13% market
share of total retail trade.
The structure of this paper is as follows: Section 2 provides an overview on the literature on cryptos and
the factors influencing adoption decisions of novel payment instruments by retailers. We pay attention
to both the two-sided markets literature and the technology adoption literature. Section 3 formulates and
discusses the main research question, and three related sub questions on adoption intention and actual
acceptance of cryptos by retailers. Section 4 discusses the set-up of the survey and provides some
descriptive statistics. Section 5 briefly describes the econometric models used for the in-depth analyses.
Section 6 presents and discusses the estimation results and Section 7 summarises and concludes.

4
2. LITERATURE
2.1 Theoretical literature
2.1.1 Two-sided markets
Although price setting in crypto payment markets is different from price setting in the ‘traditional’ two-
sided markets for retail payment instruments, the ‘two-sided’ markets literature may still provide
insights into which factors drive retailer acceptance of cryptos. Next to pricing, we pay attention to the
influence of the net private transactional benefits and of network externalities in case of two types of
end-users.
In general, a two-sided market is a market characterised by having two demand sides instead of one, and
a platform which offers its product to both demand sides. This means that a ‘product’ will only be sold
if both sides jointly decide to ‘purchase’ the product. The platform determines the total price paid for
the jointly-bought product and the individual prices paid by these two end-users. The crypto market may
be considered as a special case of a two sided market, just like the card payments market, which has
been the focus of considerable academic attention, see Verdier (2011) or Jonker (2016) for overviews.
In the crypto payment market there is a decentralised platform consisting of multiple nodes which offers
people the opportunity to transfer funds from one person’s account to another using a particular crypto
X, such as bitcoin. This transaction will only take place if both the payer and the payee have adopted
crypto X and have agreed to use it for this specific transaction. If one of them prefers another payment
method the transfer will not take place with X. This may happen if the net transactional benefits of
another payment method or another crypto Y exceeds that of using X for either the payer or the payee.
By net transactional benefits we mean the difference between the benefits of a payment with a particular
payment method, minus the transactional costs associated with the payment.
In a two-sided market, network externalities on one side of the market positively influence demand on
the other side. For consumers adopting crypto X becomes more attractive the higher the share of retailers
who accept it, while for retailers adoption of X becomes more attractive the higher the adoption rate of
X among consumers. Generally, centralised platforms which offer payment solutions try to maximise
the platform’s output by setting the transaction fees of the payee and the payer in such a way that total
output is maximised. In practice, payment platforms often charge consumers a zero transaction fee or
even a negative fee (reward) and a positive transaction fee to retailers. The transaction fee for retailers
may be higher than the cost associated with delivering the payment service to retailers, as platforms try
to pass on part of the cost associated with consumers to retailers, as retailer demand is assumed to be
less price elastic than consumer demand. A rationale for platforms to price their payment service this
way is that they want to encourage consumers to adopt their payment method, and as the consumers’
adoption rate rises, so will the retailers’ adoption rate due to network externalities. Note that unlike

5
payment card networks, in decentralised crypto platforms, such as bitcoin, do not set transaction fees
for payees and payers. .However, payers may voluntarily pay a fee, as an incentive to the miners in the
network to process their transaction quickly.
3
Intermediaries such as non-bank payment service
providers (PSPs) which offer payment services to retailers charge transaction fees for accepting VC
payments.
In the early two-sided market models, retailers were assumed to be homogeneous and to operate in a
non-competitive market, in which either all retailers adopted a payment method or did not (Baxter,
1983). However, in reality retailers in different sectors may perceive different benefits from adopting a
payment instrument, leading to different adoption rates across sectors (Wright, 2004). In addition,
retailers may face different cost structures and consequently have different adoption rates depending on
the average transaction size or sales volume (McAndrews and Wang, 2008). Furthermore, adoption
depends on market competition. Retailers who face competition may accept a payment method even
though the net transactional benefits are negative. They do so in order to attract consumers from
competing retailers, or to prevent losing customers to competitors (Rochet and Tirole, 2002; Vickers,
2005). In highly competitive markets, platforms can therefore charge excessive fees to retailers. This
has occurred in the debit and credit cards market in several jurisdictions worldwide, and has led to
various antitrust lawsuits and even price regulation by competition authorities (for an overview see
Jonker, 2016).
A distinguishing feature of using a crypto compared to using a payment instrument based on a regular
currency concerns the exchange rate between the cryptol and the regular currency. Bolt and Van Oort
(2016) present an economic framework for analysing the functioning of the crypto market, and in
particular the development of the exchange rate of cryptos. Both the speculative demand by investors
and the transaction demand by consumers and retailers influence the development of this exchange rate.
Since their introduction, cryptos have exhibited a high volatility in exchange rate with regular currencies.
This can be considered as a symptom of the cryptos’ early development, as in the long run, if adoption
by consumers and retailers increases, there will be an equilibrium exchange rate between the crypto
”currency” and regular currency, where investors’ demand will put a “floor” under the exchange rate.
2.1.2 Technology adoption literature
In this paper we also take into account findings from the technology adoption literature, see also
Aydogan (2016) or Silinskyte (2014) for overviews. The technology adoption literature initially focused
3
When this survey was launched the median fees for bitcoin transactions were well below 1 USD, but between December
2017 -- January 2018 it was 5 USD or higher, with a peak of about USD 34 just before Christmas (see:
https://bitinfocharts.com/comparison/bitcoin-median_transaction_fee.html). Miners usually process the transactions with the
highest transaction fees first. Recently, the transaction times and transaction fees paid by payers have risen considerably, as
the demand for bitcoin and other cryptos (as a speculative investment) has increased, which made it more difficult for payers
to get their transaction into a block. The higher the voluntary transaction fee, the less blocks it takes before the transaction is
processed.

6
on the adoption of new technologies by organisations. Later on, the models used to analyse adoption by
organisations were also used for consumers. The Technology Acceptance Model (TAM) developed by
Davis (1989) is one of the most widespread technology adoption theories. In the TAM model the factors
perceived usefulness (PU) and perceived ease of use (PEOU) jointly determine the adoption intention
of a new technology by potential users. Davis defines perceived usefulness as “the degree to which a
person believes that using a particular system would enhance his or her job performance” and perceived
ease of use as “the degree to which a person believes that using a particular system would be free of
effort”. According to TAM the greater the perceived usefulness and perceived ease of use of a new
technology, the more positive people feel about it (attitude), which increases their intention to adopt it
and to actually use it. Although TAM provides a solid basic framework, researchers also felt a need to
extend TAM and to improve its explanatory power by including additional determinants. Venkatesh,
Morris, Davis and Davis (2003) introduce the Unified Theory of Acceptance and Use of Technology
(UTAUT), in which they combine insights from TAM and seven other adoption models. UTAUT
consists of four main factors determining adoption intention, i.e. performance expectancy (PE), effort
expectancy (EE), social influence (SI) and facilitating conditions (FC). SI is defined as “the degree to
which an individual perceives that important others believe he or she should use the new system” and
FC as “the degree to which an individual believes that an organizational and technical infrastructure
exists to support use of the system”. PE and EE are respectively fairly similar to PU and PEOU from
TAM.
2.2 Empirical literature
There are few empirical studies on payment technology adoption by retailers who sell their products
online. Li, Ward and Zhang (2003) and Van Hove and Karimov (2006) examine the role of risk on
retailers’ adoption of payment methods. Li et al. (2003) use information from 260 online eBay sellers
and conclude that adoption choices reflect a balanced evaluation of the cost and convenience associated
with the payment methods and the protection they provide to buyers against any risks associated with
the product sold. Van Hove and Karimov (2016) surveyed 192 retailers active in five Central Asian
countries and find that retailers who sell high-risk products (high value physical products) online are
more likely to accept low-risk, immediate payment instruments from buyers, so that they are certain that
they will receive their money. However, if buyers also run risks due to the way products are being
delivered, retailers become more prone to accept higher-risk payment instruments (pay later, no payment
guarantee) as well. This finding is in line with earlier findings by DNB (2007) on the Dutch online
payment market.
Studies on the crypto uptake of cryptos by retailers are also scarce. Polasik et al. (2016) analyse the
share of bitcoin payments in total retail sales using information of 108 bitcoin accepting retailers from
different countries. The importance of bitcoin payments is relatively large among start-ups, small

7
retailers, in developing countries or in countries with a large shadow economy. Interestingly, the share
of bitcoin in total sales increases with the bitcoin awareness of potential customers, suggesting the
existence of network externalities. Silinskyte (2014) examines bitcoin adoption among a small sample
of 111 bitcoin users and non-users worldwide using the UTAUT model. She finds that adoption intention
is significantly influenced by the respondents’ expectations regarding the performance of bitcoins and
the amount of effort required to adopt them. Furthermore, actual bitcoin usage depends on facilitating
conditions.
Schuh and Shy (2016) examine crypto adoption among a representative sample of US
consumers using the 2014-15 Survey of Consumer Payment Choice. Actual adoption turns out to be
low; about one percent or less of the consumers have ever owned cryptos. People who expect an
appreciation of a crypto relative to regular currencies are more likely to hold them, suggesting that
investment motives drive consumers’ adoption. However, people also use them to pay for goods and
services and for remittance payments to other consumers, indicating that cryptos also act as a means of
payment.
3. RESEARCH QUESTIONS
Summarising, the academic literature provides several insights into which factors influence retailers’
decision to accept payments with a particular payment instrument from their customers. The literature
also suggests that due to the heterogeneity of retailers, they may think differently about the added value
for their business to accept crypto payments. Given this background, the aim of this study is to answer
the following key research question:
Q: Which factors influence the retailer’s adoption intention / acceptance decision of crypto payments?
There is some overlap in the economics and the technology adoption literature with respect to the factors
influencing adoption decisions, such as net transactional benefits with performance expectancy and
network externalities with social influence. There are, however, also insights from the technology
adoption literature which do not have a direct counterpart in the economics literature, such as effort
expectancy and facilitating conditions which reflect non-financial barriers. Therefore, we enrich our
empirical analysis by taking non-financial barriers into account as well. Furthermore, we distinguish
between the influence of these factors on adoption intention among retailers who do not accept crypto
payments as well as on current acceptance among all retailers. To be more specific, we address the
following sub-questions:
Qa: Does the retailer’s assessment of consumer adoption of crypto payments influence his/her adoption
intention / acceptance of crypto payments?

8
In order to answer this question we use three measures for the retailer’s assessment of consumer demand.
First of all, we use the retailer’s overall assessment of the adoption rate of crypto payments by online
shopping consumers. According to the two-sided market literature, the utility of adopting a payment
instrument by retailers increases with the adoption rate by consumers. Consequently, we expect a
positive relationship between the retailers’ assessment of the consumers’ adoption rate and their
adoption intention/acceptance of crypto payments. Secondly, studies on consumer adoption of new
payment technologies show that age and gender are important (see e.g. Stavins, 2001, or Jonker, 2007).
Early adoption declines with age and is relatively high among men. We therefore use the measures
‘Gender composition customers’ which indicates the retailer’s self-reported gender composition of
his/her customers and ‘Age composition customers’ which reflects the retailer’s self-reported age
composition of his/her customers.
4
Qb: Does the retailer’s assessment of the private net transactional benefits associated with accepting
VC payments influence the adoption intention / acceptance of crypto payments?
Whether a retailer accepts a specific payment instrument depends on the net transactional benefits it
provides. Net transactional benefits reflect the difference between the transactional benefits of payment
transactions done with a particular payment instrument to the retailer (e.g. in terms of convenience or
safety/security) and the retailer’s transaction fee. Net transactional benefits influence the retailer’s
adoption intention positively. We use five indicators, see section 5.2 for further details: ‘Relatively
favourable safety crypto’ which reflects fraud and cybercrime risk to the retailer related to crypto
payments relative to other payment instruments, ‘Relative favourable labour time cost crypto’ which
reflects time needed to handle crypto transactions by the retailer’s staff compared to other means of
payment, ‘Relative favourable transaction cost crypto’ which reflects the relative level of the retailer’s
transaction fees of crypto payments compared to other instruments, ‘Exchange rate risk’ which reflects
the perceived risk associated with fluctuations in the value of crypto payments relative to other means
of payment in regular currencies and ‘Customers within euro area’ which indicates that all the retailer’s
customers live in the euro area. We expect a positive impact of the three indicators relatively favourable
safe, relatively favourable labour time cost and relatively favourable transaction cost of cryptos on
adoption intention/acceptance and we expect that perceived exchange rate risk exercises a downward
pressure on retailers’ adoption intention/acceptance. With respect to retailers mainly having customers
living in the euro area, we expect a negative impact, as they don’t experience the advantages of crypto
payments as clearly as the ones with customers from outside the euro area, such as no exchange rate
fees and shorter transfer times.
4
Gender is often known to the retailer, because customers are asked to indicate their gender when making an online purchase
for addressing and billing purposes. Retailers may also have a fairly good view on their customers’ age profile, even though
customers often do not have to provide information about their age. The products they sell may target a specific age cohort and
the first name provided for addressing/billing purposes may give an indication about a customer’s age due to trends in first
names (Gerhards and Hackenbroch, 2000; Twenge, Abeke and Campell, 2010).

9
Qc: Does the retailer’s perceived level of effort associated with accepting crypto payments influence
the adoption intention/acceptance of crypto payments?
According to the technology adoption literature, the lower retailers perceive the effort required to start
working with a new technology within a firm, the higher the adoption intention. We use two indicators
for this non-financial barrier, i.e. ’perceived ease of use’ and ‘perceived compatibility’. Both factors are
expected to have a positive impact on retailers’ adoption intention/acceptance.
4. SURVEY
4.1 Data collection
The survey was held in the period 11 November - 7 December 2016 among 768 retailers in the
Netherlands. We focussed on retailers who sell their products online, as crypto payments are typically
suitable for online payments and less suitable for point-of-sale payments. Research agency Panteia
which is specialised in retailer research was responsible for the data collection. Panteia conducted
telephone interviews in order to raise response levels and to ensure completion of the questionnaire by
the responding retailers. Panteia’s interviewers contacted the person of the establishment who was
responsible for retail payments (usually the owner), as we are interested in the drivers of the adoption
decision.
We used two sources to draw our sample. Most retailers were drawn randomly from the registers of the
Reach database of research company Van Dijk. Reach includes information on 3.6 million companies
in the Netherlands. The sample drawn from Reach was stratified into ten retail sectors and five company
sizes in order to ensure sufficient variation in the sample of retailers.
5
Table A.1 in the annex provides
an overview. In addition, Panteia contacted 102 retailers who sell products online, who were on a list of
bitcoin-accepting retailers in July 2016 and whose contact details (phone number) seemed to be
available.
6
As our main purpose is to identify drivers of crypto adoption, it is key to have sufficient
heterogeneity in the sample and to have a sufficient number of crypto accepting firms in it. However,
we have to bear in mind that our sample may not be representative for the population of retailers who
sell their products online, when interpreting the outcomes with respect to the share of retailers who
5
According to Panteia/Statistics Netherlands more than 95% of the online retailers have 10 or fewer employees. In our sample
retailers with more than 10 employees are overrepresented in order to have a sufficient number of medium sized and large
retailers to assess the influence of firm size on adoption decisions. As information is unavailable about the characteristics of
the population of retailers who sell their products online, we are not able to check to what extent our sample represents the
population of online selling retailers. Panteia drew a random stratified sample of 8,445 firms from REACH. 4,112 firms were
not usable (firm was closed down, wrong address, no phone number available, firms did not sell products online). Of the
remaining 4,333 firms, 1,695 firms refused to participate, 297 did not pick up the phone within 5 attempts, 189 were not open
during the interview period and for 49 firms the interview could not be completed due to language problems. Based on a sample
size of 4,333 firms and 741 completed interviews the response rate was 17%. Of the 102 firms on the bitcoin acceptance list,
33 firms were not usable due to several reasons. Of the remaining 69 firms, 27 completed the interview, resulting in a response
rate of 39%. Response rate between 17 – 39% are not uncommon among random samples from a population.
6
http://www.watisbitcoin.nl/

10
accept crypto payments and the share of retailers who intend to adopt. These outcomes are merely
indicative.
Of the 768 retailers in the sample, 43 accept crypto payments. 27 of them are from a bitcoin accepting
list and 16 are from the registers of Reach. The latter figure indicates that crypto acceptance of retailers
in the Netherlands is fairly low, i.e. 2% of the retailers who are active in e-commerce. In our sample the
share of crypto accepting retailers is higher and amounts to 6%. Most of the retailers accept iDEAL
7
payments (79%), online credit transfers (61%), followed by PayPal (46%), credit card (43%), the
Belgian payment solution Bancontact (22%), cash on delivery (21%), debit card on delivery (10%),
Klarna/Afterpay and the German online payment solution Sofort (both 9%).
Most of the crypto accepting retailers immediately exchange their turnover in cryptos for euros (63%),
16% exchange them for euros when the exchange rate is favourable, 2% use them for payments and
another 2% exchange them for a non-euro currency when the exchange rate is favourable. 16% do not
know what happens with their crypto receipts.
The questionnaire includes questions on the retailer’s view on the safety, transaction cost and labour
time cost associated with crypto transaction, and five commonly-used payment instruments for online
purchases (iDEAL, credit transfer, credit card, direct debit, and PayPal) using a 7 point Likert scale. It
also includes questions on crypto adoption by online shopping consumers in general, characteristics of
respondents’ customers, their payment behaviour, firm characteristics and demographic information on
the respondents themselves. Furthermore, it contains questions related to the reasons for accepting
crypto payments or not, and the intention to accept crypto payments. Lastly, the survey has questions
related to the non-financial barriers related to crypto acceptance.
Regarding the reasons given for crypto acceptance, 42% of the retailers accept them to attract extra
customers or because their customers ask for it (23%). Many retailers accept them because they are
interested in new technology (21%) or because of the low transaction fees (7%). None of the retailers
indicate that the privacy provided by crypto payments to their customers plays a role. Neither do they
indicate that the mitigation of exchange rate risk or shorter transfer time to their account influence their
adoption decision.
Unfamiliarity with cryptos is the most cited reason for non-acceptance (58%), followed by lack of
consumer demand (36%), not feeling the need for acceptance (17%), lack of trust in crypto (16%),
acceptance not being common in their industry (12%), safety concerns (9%) and perceived complexity
(5%). Overall, both the answers given by accepting and non-accepting retailers indicate that customers’
(expected) demand for crypto influences the acceptance decision.
7
iDEAL is a payment solution used in the Netherlands, offered by banks and based on online banking. In 2015 it had a market
share of 56% in the number of online payments (Betaalvereniging, 2016).

11
5. THE MODELS
5.1 Dependent variables
We construct two dependent variables: Acceptance and Adoption intention. The dependent variable
Acceptance equals 1 for retailers who accept crypto payments and zero for those who do not. Of the
respondents, 6% accept crypto payments and 94% do not. We use probit regressions to examine which
factors influence retailers’ decisions with respect to crypto acceptance. The dependent variable Adoption
intention takes a somewhat broader perspective than Acceptance. Retailers who do not accept crypto
payments were asked whether they would consider accepting crypto payments in the near future.
Adoption intention takes on three values, i.e. 1 denoting the answer ‘no’, 2 denoting the answer
‘maybe/perhaps eventually’ and 3 referring to the answer ‘yes’. We exclude respondents who could not
answer this question and respondents who already accept crypto payments from this analysis. 7% of the
non-accepting retailers intend to accept crypto payments soon, 19% reply that perhaps eventually they
will accept them and 64% know for sure that they will not accept them. We estimate ordered probit
regressions to examine which factors influence retailers’ intention to adopt virtual currencies. An
ordered probit model is an extension of the binomial probit model. The main difference is that the
dependent variable can take on more than two values, which have a natural ordering. Differences in the
levels of the dependent variable have a qualitative meaning instead of a purely metric one, which makes
this model appropriate for the analysis of adoption intention (see e.g. Cameron and Trivedi, 2010, for
more information).
5.2 Explanatory variables
Below we describe the set of explanatory variables we use to answer research questions Qa-Qc as well
as the set of other control variables.
5.2.1 Consumer adoption of crypto payments
According to the two-sided market literature, retailers’ adoption decisions depend on the adoption on
the other demand side, i.e. consumer demand. We use several variables reflecting consumer demand.
Table 1 provides the average scores for these variables for crypto accepting respondents and for those
who do not. For the latter group averages are given depending on the level of adoption intention. In
addition, we provide the results of 2 sample t-tests which test whether the average responses in two
groups differ significantly or not.

12
Consumer demand crypto reflects the retailer’s answer to the question “What share of all consumers
which made at least one online purchase in 2016 used virtual currency payments at least once?”.
8
On
average, retailers state that 8% of the consumers used crypto payments in 2016. VC-accepting retailers
think that 6% of the online shopping consumers used crypto, which is significantly lower than the 9%
according to retailers who do not accept cryptos. Retailers who do not yet accept cryptos, but who intend
to do so, assess consumer adoption slightly higher than retailers who are certain that they are not going
to accept crypto payments (9% versus 8%). However, the difference is not statistically significant.
9
The second and third measure for consumer demand consider the characteristics of the retailers’ own
customers, i.e. their gender and age. Retailers who accept cryptos indicate relatively more often than
those who do not that their customers are mainly people below the age of 30 (16% versus 13%). Of the
latter group, the likelihood that retailers who intend to adopt crypto payments have a relatively young
clientele is at 19% almost twice as high than the 11% of the retailers who know for sure they are not
going to accept cryptos, but these differences are not statistically significant. Regarding gender
10
, we
find that among the crypto accepting retailers, there are relatively many with mainly male customers
(26%) and relatively few with mainly female customers (7%) whereas the opposite holds for retailers
who do not accept crypto payments (12% mainly male customers and 25% mainly female customers).
These differences are statically significant. We see a similar picture emerging when comparing the
gender composition of retailers who intend to accept crypto payments, who may accept crypto payments
and who know for sure they are not going to accept them, but these differences are not significant.
Table 1: Comparing retailers’ perceptions with respect to consumer demand for crypto
Acceptance
Results 2-sample t-tests
Variable
Yes
No
p-value
Consumer demand crypto (in %)
6%
9%
p=0.037
Age profile own customers: mainly young (<=30 yrs)
16%
13%
p=0.53
Gender profile own customers: mainly male
26%
12%
p=0.01
Gender profile own customers: mainly female
7%
25%
p=0.01
Adoption intention
Results 2-sample t-tests
Yes
Maybe
No
Yes vs Maybe
Maybe vs No
Consumer demand crypto ( in %)
9%
9%
8%
p=0.95
p=0.17
Age profile own customers: mainly young (<=30 yrs)
19%
16%
11%
p=0.66
p=0.13
Gender profile own customers: mainly male
18%
13%
11%
p=0.44
p=0.44
Gender profile own customers: mainly female
14%
23%
26%
p=0.137
p=0.47
8
Here we provide the exact wording of the question. In 2016 the term ‘cryptocurrency’ was not mentioned (often) yet in the
media ,whereas the term ‘virtual currencies’ was. Therefore, we used the latter term in our questionnaire.
9
In Tables 1 – 3, we used two-sample mean comparison t-tests, assuming unequal variances to tests whether groups averages
are equal to each other or not.
10
We distinguish five classes: a retailer has mainly male customers, has more male than female customers, has as many male
as female customers, has more female than male customers and has mainly female customers.

13
5.2.2 Private net transactional benefits of crypto acceptance
Perceived risks and performance of crypto payments compared to other instruments for online payments
may also influence the adoption decision (see Table 2). The variable Exchange rate risk reflects the
respondents’ perceived uncertainty in the cost associated with fluctuations in the exchange rate. They
were asked the following question: ‘How large do you perceive the exchange rate risks between virtual
currencies and regular currencies?’ using a 1 (very low) to 7 (very high) scale.
11
Crypto accepting
retailers perceive the exchange rate risk as lower (average score 4.0) than the retailers who do not accept
crypto payments (average score 4.7). The difference in average scores is statistically significant at the
10% level. A similar pattern is visible within the group of non-accepting retailers distinguished by
adoption intention, although these differences are not significant. The finding that crypto-accepting
retailers perceive relatively low exchange rate risk may be explained by the role of payment service
providers (PSPs) which facilitate crypto acceptance. Most retailers in our sample who accept crypto
payments also make use of the services of a PSP (93% against 68% of the retailers who do not accept
crypto payments). These PSPs act as intermediaries between retailers, their customers and providers of
transfers using specific payment instruments. They often offer retailers services to mitigate exchange
rate risk, which is something non-accepting retailers may not be aware of.
The results for the second measure Customers within euro area do not point at a relationship between
the residence of the retailers’ customers and Adoption intention and Acceptance. This finding is
counterintuitive, as especially retailers with customers outside the euro area may benefit from crypto
payments. In contrast to crypto payments, cross-currency transfers using means of payment in regular
currencies have relatively high transaction fees and/or long transfer times.
Table 2: Retailers’ perceptions towards cryptos relative to other payment instruments
Acceptance
Results 2-sample t-tests
Variable
Yes
No
p-value
1. Exchange rate risks ( 1 =very low, 7=very high))
4.00
4.67
P=0.07
2. Customers within euro area
0.74
0.78
P=0.52
3. Relatively favourable cost crypto
1.65
1.15
P=0.00
4. Relatively favourable safety crypto
0.98
0.74
P=0.00
5. Relatively favourable labour time cost crypto
1.11
0.88
P=0.00
Adoption intention
Result 2-sample t-tests
Yes
Maybe
No
Yes vs Maybe
Maybe vs No
1. Exchange rate risks (1=very low, 7=very high high)
4.33
4.60
4.81
P=0.41
P=0.33
2. Customers within euro area
0.87
0.80
0.78
P=0.27
P=0.60
3. Relatively favourable cost crypto
1.30
1.21
1.11
P=0.36
P=0.04
4. Relatively favourable safety crypto
0.85
0.76
0.72
P=0.09
P=0.14
5. Relatively favourable labour time cost crypto
0.98
0.93
0.86
P=0.34
P=0.01
11
The question is asked to the 552 retailers who were familiar with crypto payments.

14
The third measure: Relatively favourable cost crypto equals the ratio of the perceived attractiveness of
the cost for accepting crypto payments to the average perceived attractiveness of the cost of accepting
payments with five other commonly used online payment instruments. Perceived attractiveness of the
cost is based on the answer to the question: ‘How high do you perceive the cost for companies of payment
instrument x? By cost we mean fees paid to banks and payment service providers”. Respondents could
provide an answer on a 1 (very high) to 7 (very low) scale. A ratio higher (lower) than 1 implies that the
retailer perceives the cost for accepting crypto payments as more favourable, i.e. lower (less favourable,
i.e. higher) than the average cost for the five other mainstream payment instruments. Also, for perceived
safety and labour time cost for the retailer’s staff a ratio higher (lower) than 1 implies that crypto
payments are perceived as more (less) favourable than the average of the other five payment methods.
12
The survey results show that retailers who accept crypto payments perceive them as more favourable
than non- accepting retailers for all three perception factors. The differences are significant at the 1%
level. In general, retailers who accept crypto payments consider them as equally safe as the other five
payment methods. Furthermore, they perceive them as less costly in terms of fees and with respect to
labour time cost than the other means of payment. Interestingly, also retailers who do not accept crypto
payments perceive crypto payments as relatively cheap. This holds even for retailers who will not accept
them. Regarding the other two perception factors, retailers who do not accept crypto payments clearly
perceive them as less favourable than the other payment methods. Crypto payments score particularly
low on safety. Retailers who intend to accept crypto payments in the future have a significantly more
positive attitude regarding the relative safety of crypto payments than retailers who may accept crypto
payments, but they do not differ from them with respect to their judgement of the relative transaction
cost and relative labour time cost. Retailers who may accept crypto payments do differ significantly
from retailers who will not accept crypto payments with respect to these latter two perceptions.
Adoption effort
We use two constructs from the technology adoption literature that reflect adoption effort, i.e. perceived
ease of use/learning cost and perceived compatibility of crypto payments with existing working
procedures. For both constructs respondents could provide their opinion on two statements, all using a
7 point-Likert scale, ranging from strongly disagree (1) to strongly agree (7). The questions are listed
below:
12
Perceived safety is based on the retailer’s answer to the question “How do you perceive the safety for companies of payments
with payment instrument x? Safety concerns fraud and cybercrime”. The respondents could provide an answer on a (very
unsafe) to 7 (very safe) scale. Perceived labour time cost is based on the retailer’s answer to the question “How do you perceive
the time needed for a company to handle payments with payment instrument x?”. The respondents could provide an answer on
a 1 (hardly labour intensive) to 7 (very labour intensive) scale. In order to ensure an equal interpretation of the scores for all
three perceptions (low score=bad, high score=good), the scores given to perceived cost and perceived labour time cost have
been reversed for the calculation of the relative perceived cost and labour time cost.

15
Perceived ease of use:
1. It’s easy for me and my staff to learn to accept payments in virtual currencies.
2. It’s clear and easy for me and my staff to understand how we receive payments in virtual
currencies.
Perceived compatibility:
3. The acceptance of virtual currency payments fits well with all other aspects of our firm.
4. The acceptance of virtual currency payments fits well with the way I and/or my staff want to
receive payments for our products.
Table 3 provides average group scores per construct. Crypto accepting retailers feel significantly more
positive with both perceived ease of use and perceived compatibility than the other retailers. Retailers
who do not accept crypto payments, but state they will do so, score significantly higher than those who
state they may accept them in the future. The latter group scores significantly higher than the retailers
who know for sure they are not going to accept crypto payments. The results suggest that retailers who
are quite certain about crypto acceptance, foresee a smooth transition towards crypto acceptance within
their firm, compared to retailers who are still hesitant. Their expectations are supported by the
experiences of crypto accepting retailers, as they are even more positive than the ones who intend to
adopt them.
Table 3: Retailers’ attitude towards crypto payments
Acceptance
2-sample t-tests
Construct
Yes
No
p-value
1. Perceived ease of use
5.67
2.80
P=0.00
2. Perceived compatibility
5.29
2.58
P=0.00
Adoption intention
2-sample t-tests
Yes
Maybe
No
Yes vs Maybe
Maybe vs No
1. Perceived ease of use
4.43
2.98
2.57
P=0.00
P=0.03
2. Perceived compatibility
4.49
3.45
1.99
P=0.00
P=0.00
Other variables
We also include variables which reflect demographic characteristics of the retailers (age and educational
level) as well as firm characteristics (founding date, firm size measured by the number of employees,
whether the retailer makes use of the services of a payment service provider or not) in the set of control
variables as well as sector variables. In addition, we control for the competitiveness of the market.

16
6. ESTIMATION RESULTS
This section presents and discusses the estimation results of the regression analyses. Table 4 shows the
estimation results for the dependent variable Adoption intention measuring the relative intention to
accept crypto payments by retailers who do not accept crypto payments yet using the ordered probit
regression model. Table 5 presents the results for the dependent variable Acceptance based on
information of all respondents using the probit regression model. In order to check for the robustness of
the estimated effects and to assess the added value of the three sets of key variables, we estimate models
only containing the basic variables and the set of variables related to a specific research question (Model
1 for Qa, Model 2 for Qb and Model 3 for Qc), and we estimate a model including all variables (Full
Model), for which we present the estimated parameter coefficients (β) and average marginal effects
(AMEs).
13
6.1 Effect of consumer adoption crypto payments
We find that two of the three indicators of the retailer’s assessment of consumer adoption of crypto
payments significantly influence the intention to adopt crypto payments (Model 1 and Full Model, Table
4) and that one indicator influences the acceptance decision (Model 1 and Full Model, Table 5). In line
with the two-sided market literature, adoption intention is positively influenced by the retailer’s overall
assessment of crypto adoption by online shopping consumers. The average marginal effects indicate that
a one percentage point higher assessment of crypto adoption by consumers, increases the probability
that a retailer wants to adopt crypto payments by 0.2 percentage points and decreases the probability
that (s)he does not intent to adopt them by 0.5 percentage points. The results also show a significant
effect of gender composition of the retailer’s customers. Retailers whose clientele mainly consists of
women are 4.1 percentage points less likely to be quite certain to adopt crypto payments and 9.6
percentage points more likely not to be willing to adopt crypto payments than retailers who have a mixed
clientele with respect to gender (reference group). The age structure of the retailer’s customers does not
affect adoption intention.
13
Average marginal effects (AMEs) are marginal effects which are averaged across the respondents in the sample, and
evaluated relative to the corresponding reference category, see e.g. Cameron and Trivedi (2010). For adoption intention, the
AMEs show the impact of the explanatory variables on the probabilities that the retailer does not intend to adoption VC
payments (AME on adoption intention = ‘no’) and that the adoption intention is very high (AME on adoption intention = ‘yes‘),
relative to the reference group (perhaps eventually/maybe). So, for the binomial explanatory variable ‘PSP’, the AMEs show
how the probabilities for answer categories ‘yes’ and ‘no’ would change if a retailer made use of the services of a PSP to accept
online payments from customers, compared to one who does not make use of a PSP. For a continuous variable such as ‘age’
the AMEs show the change in probabilities if the retailer’s age increases by one year.

17
TABLE 4: Adoption intention crypto payments by retailers
Model 1
Model 2
Model 3
Full model
Dependent variable: Adoption intention
β

β
β
AME Acceptance=no
AME Acceptance=yes
Retailer characteristics
Age (yrs)
-0.022**
-0.022***
-0.014***
-0.016***
0.004***
-0.002***
(0.005)
(0.005)
(0.006)
(0.006)
(0.002)
(0.001)
Education: Bachelor degree
0.087
0.097
0.030
0.084
-0.023
0.010
(0.116)
(0.115)
(0.141)
(0.015)
(0.039)
(0.017)
Education: Master degree
0.038
0.075
-0.058
0.028
-0.007
0.003
(0.157)
(0.160)
(0.176)
(0.183)
(0.049)
(0.021)
Firm age: less than 2 years
0.224
0.244
0.452**
0.467**
-0.125**
0.055**
(0.155)
(0.155)
(0.202)
(0.212)
(0.055)
(0.024)
Firm age: 2 – 5 years
0.264*
0.277**
0.343*
0.382**
-0.102**
0.044**
(0.139)
(0.138)
(0.177)
(0.187)
(0.049)
(0.022)
Firm size: 1 person
0.056
-0.034
-0.184
-0.134
0.036
-0.015
(0.193)
(0.182)
(0.219)
(0.234)
(0.062)
(0.027)
Firm size: 2 – 4 people
0.010
0.0082
-0.227
-0.165
0.044
-0.019
(0.198)
(0.192)
(0.237)
(0.234)
(0.065)
(0.028)
Firm size: 5 – 19 people
0.004
-0.007
-0.224
-0.205
0.055
-0.023
(0.187)
(0.183)
(0.228)
(0.231)
(0.061)
(0.026)
Uses services PSP
0.328***
0.296**
0.185
0.272*
-0.073*
0.031*
(0.122)
(0.122)
(0.156)
(0.160)
(0.042)
(0.018)
Sector: media
0.338*
0.446**
0.317
0.411*
-0.110*
0.047*
(0.186)
(0.182)
(0.230)
(0.238)
(0.063)
(0.028)
Sector: electronics
0.510***
0.533***
0.556**
0.481**
-0.128**
0.055**
(0.191)
(0.184)
(0.233)
(0.239)
(0.063)
(0.027)
Competition: no to weak
0.071
0.038
-0.049
-0.022
0.006
-0.003
(0.191)
(0.188)
(0.244)
(0.261)
(0.070)
(0.030)
Competition: strong to perfect
-0.046
-0.104
-0.102
-0.102
0.027
-0.012
(0.121)
(0.124)
(0.165)
(0.173)
(0.046)
(0.020)
Consumer adoption crypto
Customers: mainly male
-0.024
-0.108
0.029
-0.012
(0.157)
(0.185)
(0.049)
(0.021)
Customers: mainly female
-0.283**
-0.360**
0.096**
-0.041*
(0.134)
(0.180)
(0.047)
(0.021)
Customers: mainly 30 years or younger
0.037
-0.142
0.039
-0.016
(0.153)
(0.190)
(0.051)
(0.022)
Perceived degree of consumer adoption
crypto
0.016**
0.018**
-0.005**
0.002**
(0.006)
(0.009)
(0.002)
(0.001)
Missing value Perceived degree of
-0.399***
-0.091
0.051
-0.022
consumer adoption crypto (dummy 0/1)
(0.145)
(0.212)
(0.056)
(0.024)
Retailer’s net transactional benefits
Relatively favourably cost crypto
0.233**
0.214*
-0.057*
0.025*
(0.115)
(0.124)
(0.033)
(0.014)
Relatively favourable labour time cost
crypto
0.458**
0.059
-0.016
0.007
(0.204)
(0.226)
(0.060)
(0.026)
Relatively favourable safety crypto
0.201
0.092
-0.025
0.011
(0.183)
(0.224)
(0.060)
(0.026)
Exchange rate risk crypto
-0.066*
-0.029
0.008
-0.003
(0.039)
(0.045)
(0.012)
(0.005)
Customers: within euro area
-0.099
-0.272
0.073*
-0.031
(0.126)
(0.167)
(0.044)
(0.019)

18
Table 4 continued
Model 1
Model 2
Model 3
Full model
Dependent variable: Adoption intention
β

β
β
AME
Acceptance=no
AME
Acceptance=yes
Adoption efforts
Perceived ease of use
-0.004
-0.005
0.001
-0.001
(0.041)
(0.042)
(0.011)
(0.005)
Perceived compatibility
0.381***
0.382***
-0.102***
0.044***
(0.041)
(0.043)
(0.009)
(0.006)
µ1
0.170
0.690*
1.063***
1.631***
(0.316)
(0.412)
(0.369)
(0.570)
µ2
1.166***
1.693***
2.332***
2.951***
(0.319)
(0.416)
(0.387)
(0.594)
No. of observations
650
650
444
444
Log likelihood
-458.38
-456.06
-295.41
-286.01
Pseudo R-squared
0.083
0.088
0.208
0.233
Notes. The table shows coefficients (β) and average marginal effects (AMEs) based on ordered probit regressions with Adoption intention as
dependent variable. Robust standard errors are between parentheses. The sample excludes retailers who accept crypto payments or did not
know their adoption intention. Reference characteristics of the firm are: firm’s age higher than 5 years, firm size: 20 people and more, does not
make use of the services of a PSP, sector: other than media or electronics, the firm experiences moderate competition, has a mixed clientele
with respect to gender ( more male than female, as many male as female, more female than male), the age of the firm’s clientele is mixed or
mainly consists of people aged 31 years and older, the firm accepts payments within and outside the euro area. . *p<.1, **p<.05, *** p<.01
(two-sided t-tests).
We have mixed results regarding the influence of perceived consumer adoption on retailers’ current
crypto acceptance (see Model 1 and Full model, Table 5). As expected, we find a negative impact of
having mainly female customers on crypto acceptance; these retailers are 0.8 percentage points less
likely to accept crypto payments than retailers with a mixed clientele. However, the result of general
consumer adoption seems at first sight counterintuitive; it has a negative impact on retailer’s crypto
acceptance (Model 1) or no effect at all (Full model). A possible explanation may be that retailers who
already accept crypto payments have learned about actual consumer usage of cryptos, and have
developed a more realistic view on actual consumer adoption than non-accepting retailers. 44% of the
crypto accepting retailers in our survey did not receive any crypto payments in 2016 and 42% reported
an up to 5% share of crypto payments on total payments, which suggest a much lower consumer adoption
rate than the average estimated consumer adoption of 9% by non-accepting retailers (Table 1). As with
consumer adoption, the age structure of the retailer’s clientele does not influence retailers’ current crypto
acceptance.
6.2 Effect of net transactional benefits
The estimation results show that three of the five factors reflecting the retailer’s net transactional benefits
associated with crypto acceptance significantly influence adoption intention (Model 2 and Full Model,
Table 4), and that four of them relate significantly with crypto acceptance (Model 2, Table 5).

19
Retailers who anticipate relatively favourable cost for crypto transactions compared to other payment
instruments have a relatively favourable attitude towards crypto adoption. The estimated average
marginal effects indicate that a 1 point increase in the relatively favourable cost ratio (indicating a more
favourable relative cost position of crypto payments) decreases the probability that retailers do not intend
to adopt crypto payments by 5.7 percentage points and increases the probability that they want to adopt
crypto payments by 2.5 percentage points (see Full model, Table 4).
We also find that retailers who expect relatively less labour time cost for handling crypto payments
compared to other payment instruments have a relatively high tendency to adopt crypto payments. In
addition, the perceived exchange rate risk between crypto and regular currencies by retailers has a
negative impact on adoption intention. However, the estimated effects for ‘Exchange rate risk crypto’
and ‘Relatively favourable labour time cost crypto’ are statistically significant in model 2, but not in the
full model, where also the indicators reflecting required effort to adopt crypto payments are included
as control variables. As the magnitude of the estimated effects is also smaller in the full model than in
models 1 – 3 it may be the case that the estimates suffer from multicollinearity bias. We examine this in
section 6.5. Furthermore, the indicator ‘Customers: within euro area’ is significant and has the expected
sign in the full model. The average marginal effect indicates that retailers who only trade with customers
inside the euro area are 7.3 percentage points more likely not to intend to accept crypto payments than
retailers who sell both inside and outside the euro area.
The results of the variables reflecting net transactional benefits on crypto acceptance are to a large extent
in line with those for adoption intention. Model 2 shows significant results with the expected sign for
the explanatory variables ‘Relatively favourable cost crypto’, ‘Relatively favourable labour time cost
crypto’ and ‘Exchange rate risk crypto’. In addition, the estimation results point at a significant positive
correlation between ‘Relatively favourable safety crypto’ and crypto acceptance. However, as in the
adoption intention model, in the full model none of these four variables turn out to be significant,
although ‘Relatively favourable cost crypto’ and ‘Relatively favourable safety crypto’ come with p-
values of 0.103 respectively 0.105 very close to significance at the 10% level. Interestingly, relative
safety was not significant in the adoption intention equation (Table 4, model 2 and full model), but
proves to be significant in the acceptance model ( Table 5, model 2). There may be two explanations for
this difference. It may be the case that retailers with most confidence in the safety of crypto payments
were the first to accept them. An alternative explanation may be that the causality is the other way round;
once retailers accept crypto payments they learn that these payments have relatively few safety issues.
Regarding the residence of the customers, we do not find a significant effect for having customers from
within the euro area on crypto acceptance, unlike crypto adoption intention.

20
TABLE 5: Acceptance crypto payments by retailers
Model 1
Model 2
Model 3
Full model
Dependent variable: Acceptance
β

β
β
AME Acceptance=yes
Retailer characteristics
Age (yrs)
-0.030***
-0.025***
-0.018
-0.021**
-0.0003***
(0.008)
(0.008)
(0.011)
(0.010)
(0.0003)
Education: Bachelor degree
0.098
0.023
0.030
0.079
0.001
(0.192)
(0.199)
(0.241)
(0.264)
(0.005)
Education: Master degree
0.328
0.233
-0.024
-0.114
-0.002
(0.224)
(0.235)
(0.293)
(0.356)
(0.005)
Firm age: less than 2 years
-0.206
-0.289
-0.188
-0.137
-0.002
(0.275)
(0.300)
(0.363)
(0.368)
(0.005)
Firm age: 2 – 5 years
0.261
0.265
0.124
0.118
0.002
(0.218)
(0.244)
(0.275)
(0.284)
(0.005)
Firm size: 1 person
0.176
-0.106
-0.367
-0.473
-0.0075
(0.315)
(0.204)
(0.401)
(0.415)
(0.007)
Firm size: 2 – 4 people
0.245
0.127
-0.232
-0.208
-0.003
(0.314)
(0.315)
(0.405)
(0.404)
(0.005)
Firm size: 5 – 19 people
-0.663*
-0.732*
-1.369**
-1.285**
-0.010**
(0.392)
(0.418)
(0.579)
(0.598)
(0.007)
Uses services PSP
0.753***
0.715***
0.523
0.449
0.006
(0.230)
(0.256)
(0.343)
(0.328)
(0.007)
Sector: media
0.014
0.139
0.052
0.165
0.003
(0.296)
(0330)
(0.354)
(0348)
(0.008)
Sector: electronics
-0.151
0.024
-0.345
-0.562*
-0.005*
(0.254)
(0.247)
(0.306)
(0.331)
(0.004)
Competition: no to weak
0.059
0.186
0.021
0.071
0.001
(0.332)
(0.332)
(0.511)
(0.510)
(0.010)
Competition: strong to perfect
0.276
0.435*
-0.040
0.023
0.0042
(0.217)
(0.226)
(0.271)
(0.291)
(0.005)
Consumer adoption crypto
Customers: mainly male
0.345*
0.012
0.0002
(0.208)
(0.276)
(0.005)
Customers: mainly female
-0.628**
-0.829**
-0.008**
(0.283)
(0.388)
(0.007)
Customers: mainly 30 years or younger
-0.074
-0.003
-0.0001
(0.251)
(0.305)
(0.005)
Perceived degree of consumer adoption crypto
-0.029**
-0.008
-0.0001
(0.014)
(0.174)
(0.0003)
Retailer’s net transactional benefits
Relatively favourable cost crypto
0.435***
0.283
0.005
(0.128)
(0.174)
(0.005)
Relatively favourable labour time cost crypto
0.546*
0.191
0.003
(0.309)
(0.416)
(0.006)
Relatively favourable safety crypto
0.606**
0.591
0.010
(0.268)
(0.365)
(0.008)
Exchange rate risk crypto
-0.111*
-0.086
-0.001
(0.063)
(0.063)
(0.002)
Customers: within euro area
-0.326
0.132
0.002
(0.202)
(0.249)
(0.005)

21
Table 5 continued
Model 1
Model 2
Model 3
Full model
Dependent variable: Acceptance
β

β
β
AME Acceptance=yes
Adoption effort
Perceived ease of use
0.362***
0.372***
0.006***
(0.071)
(0.080)
(0.004)
Perceived compatibility
0.300***
0.275***
0.004***
(0.069)
(0.070)
(0.003)
Constant
-1.221***
-2.407***
-3.451***
-3.697***
(0.447)
(0.632)
(0.604)
(0.842)
No. of observations
761
761
521
521
Log likelihood
-133.99
-121.19
-84.46
-75.07
Pseudo R-squared
0.190
0.267
0.431
0.494
Notes. The table shows coefficients (β) and average marginal effects (AME) based on probit regressions with crypto acceptance as dependent
variable. Robust standard errors are between parentheses. Reference characteristics of the firm are: firm’s age higher than 5 years, firm size:
20 people and more, does not make use of the services of a PSP, sector: other than media or electronics, the firm experiences moderate
competition, has a mixed clientele with respect to gender, the age of the firm’s clientele is mixed or mainly people aged 31 years and older,
the firm accepts payments from inside and outside the euro area. . *p<.1, **p<.05, *** p<.01 (two-sided t-tests).
6.3 Effect of adoption effort
Regarding the drivers reflecting the effort required to adopt a new technology, we find a positive effect
for ‘Perceived compatibility’ on adoption intention. This holds for both model 3 and the full model. The
average marginal effects indicate that a 1 point higher score for perceived compatibility (1-7 scale)
decreases the probability that a retailer does not intend to accept crypto payments by 10.2 percentage
points and increases the probability that (s)he intends to adopt them by 4.4 percentage points. However,
we do not find a significant impact of ‘Perceived ease of use’ on adoption intention. This indicates that
either the extent to which retailers think it will be easy for their staff to learn to use a new technology
does not influence adoption intention or that due to multicollinearity between ‘Perceived ease of use’
and ’Perceived compatibility’ (correlation between the two indicators is 0.52, see Table B.2) the estimate
for ‘Perceived ease of use’ is biased downwards.
Both drivers correlate positively and significantly with crypto acceptance (Model 3 and full model,
Table 5). Regarding ‘Perceived ease of use’, we feel this may imply that retailers who have already
adopted crypto payments anticipated relatively low learning cost compared to non-accepting retailers
anticipate, but it may also be the case that they found it easier to learn to handle crypto transactions ex
ante than they expected a priori. A similar interpretation may be given to ‘Perceived compatibility’. The
average marginal effects indicate that a 1 point higher score given for ‘Perceived ease of use’ and
‘Perceived compatibility’ go together with a 0.6 respectively 0.4 percentage point higher crypto
acceptance rate.

22
6.4 Effect of other control variables
Next to variables reflecting consumer demand, net transactional benefits and adoption effort, we include
control variables reflecting characteristics of the respondents, firms and sector. We find that adoption
intention and crypto acceptance are both negatively related with the respondents’ age, although the
estimated average marginal effect for adoption intention is larger than for acceptance. A 1 year increase
in age corresponds with a 0.4 percentage point higher probability that the retailer does not intend to
accept crypto payments and a 0.2 percentage point lower probability that (s)he intends to accept crypto
payments. The age effect on actual acceptance is smaller: a 1 year increase in age results in a 0.06
percentage point lower probability that a retailer actually accepts crypto payments. The respondent’s
educational level does not influence adoption intention and actual acceptance.
Regarding firm characteristics, we find a negative effect of the firm’s age, with firms existing less than
5 years having a significantly higher adoption intention than firms which have existed 5 years or longer.
However, the firm’s age does not affect actual acceptance. Firm size as measured by staff size does not
influence adoption intention, but turns out to relate significantly to current acceptance. Firms with 5 -
19 employees are 1.0 percentage point less likely to accept crypto payments than firms with 20 or more
employees (reference group). Furthermore, adoption intention is positively related with whether the
retailer uses the services of a PSP to handle customer payments. The average marginal effect of the
intention not to adopt crypto payments drops by 7.3 percentage points if a retailers uses a PSP, while
the intention to accept crypto payments increases by 3.1 percentage points. Actual acceptance increases
by 0.6 percentage points, though this effect is not significant in the Full model. Note however, that PSP
usage is statistically significant in models 1 and 2, and the estimated coefficients are also higher than in
model 3 or the full model. Maybe, the variables ‘Perceived ease of use’ and/or ‘Perceived compatibility’
pick up some of the effect of using a PSP. If a retailer uses the services of a PSP, which acts as an
intermediary between the retailer and its customers, crypto acceptance may not lead to changes in the
working processes of the firm itself as it has outsourced customer payment handling to the PSP.
Likewise, the retailer’s own staff does not have to learn new skills to handle payments with the new
payment method, as this only holds for the PSP’s staff. However, note that there are no strong indications
of multicollinearity between using a PSP and the two indicators of adoption effort (see section 6.5).
Regarding sector, we find that retailers who are active in the sectors ‘Media’ or ‘Electronics’ have a
significantly more positive attitude towards crypto adoption than retailers active in other sectors
(reference group). However, with respect to current acceptance, we do not find a significant sector effect.
We only find a negative effect for retailers active in the electronics sector, but this only holds in the full
model, not for models 1 – 3. Regarding competition, we do not find any effect of it on adoption intention,
but according to models 1 and 2 in Table 5, retailers who face strong to perfect competition are more

23
likely to accept crypto payments than retailers who face moderate competition (reference group).
However, this effect is not present in model 3 and in the Full model.
6.5 Robustness check
The explanatory power of the estimated models for Adoption intention and Acceptance increase
considerably when including the two adoption effort indicators ‘Perceived ease of use’ and ‘Perceived
compatibility’ as explanatory variables. This indicates that enriching economic models with insights
from the technology adoption literature when analysing the uptake of new payment technologies may
be promising. The results also show that some of the explanatory variables which are significant in
models 1 and/or 2 are not significant anymore when including these two indicators as explanatory
variables.
There may be two possible explanations for this. First, the different composition of the retailers in the
sample in Models 1 and 2 compared to Models 3 and the Full model may affect the estimation results.
Many respondents find it difficult to express their ‘Perceived ease of use’ or ‘Perceived compatibility’
of working with crypto payments. These people are included in the regressions of Models 1 and 2, but
not of Model 3 and the Full model. We have re-estimated Models 1 and 2 for both Adoption intention
and Acceptance using retailers with responses on ‘Perceived ease of use’ and ‘Perceived compatibility’
(Table B.1, Annex B). It turns out that the estimation results for Models 1 and 2 are robust to the adjusted
sample; the estimated effects of the variables reflecting usage of a PSP, consumer adoption of crypto
and retailer’s net transactional benefits remain fairly the same, as well as the estimated explanatory
power of the models. There are only a few variables which are not significant anymore at the 10% level,
though the magnitude of the estimated effect remains roughly the same.
A second explanation may be that explanatory variables suffer from multicollinearity with the
explanatory variables ‘Perceived ease of use’ and ‘Perceived compatibility’. According to the
correlation matrix in Table B.2 there are no signs of strong correlation between these and the other
explanatory variables. Apart from the strong correlation of 0.52 between ‘Perceived ease of use’ and
‘Perceived compatibility’, there is moderate correlation ranging between 0.15 and 0.19 between the
variables ‘Perceived ease of use’ and ‘Using services PSP’, ‘Relatively favourable cost crypto’ and
‘Relatively favourable labour time cost crypto’ and between the variables ‘Perceived compatibility’ and
‘Sector Electronic’, ‘Relatively favourable cost crypto’, ‘Relatively favourable labour time cost crypto’
and ‘Relatively favourable safety crypto’. Also the Variance Inflation Factors (VIFs) of the explanatory
variables do not point at multicollinearity (See Table B.3 in Appendix B). The average VIF is 1.45, the
minimum VIF found is 1.08 and the maximum is 3.17. As a rule of thumb a VIF smaller than 10 is fine.

24
Given that especially the estimated effects of ‘Using services PSP’, ‘Relatively favourable cost crypto’
and ‘Relatively favourable labour time cost crypto’ become smaller and insignificant after including
indicators of adoption effort as explanatory variables suggests that these variables are to some extent
alike. Therefore, when discussing the results for Qb (net transactional benefits) in the concluding
remarks we will focus on the results for Model 2. Furthermore, the moderate, positive correlation
between ‘Using services PSP’ and ‘Perceived ease of use’ and ‘Perceived compatibility’ indicates that
PSP usage actually acts as a facilitating condition for online retailers to accept crypto payments by
removing or lowering the required effort for retailers to adopt the new payment technology.
7 CONCLUDING REMARKS
Currently, the acceptance of crypto payments by retailers who sell their products online is modest.
However, there is interest among retailers to adopt crypto payments in the near future, indicating that
acceptance may rise once certain (perceived) barriers are lowered. In this paper we examine which
factors drive retailer adoption intention and actual acceptance of crypto payments. We pay special
attention to the impact of consumer adoption of crypto payments (Qa), the retailer’s perceived net
transactional benefits associated with crypto payments (Qb) and the retailer’s perceived level of
adoption effort (Qc). We find that that all these three factors influence the adoption intention of online
retailers in the expected way. Furthermore, we find that net transactional benefits and perceived adoption
effort correlate positively with current crypto acceptance.
The reason why acceptance has remained limited, is because most retailers feel no to limited added value
of crypto payments compared to other payment methods. In this respect, the survey results suggest an
important role for PSPs. PSPs facilitate crypto acceptance by mitigating risk (e.g. volatility in exchange
rate) and by handling the crypto payments on behalf of retailers. In that respect, PSPs fulfil an important
role in the retail payments industry. They may enhance innovation and competition in the provision of
payment services by acting as intermediaries between (new) players and retailers.
A crucial factor limiting crypto adoption by retailers turns out to be low consumer demand. Further
research is needed to gain more insight into the factors influencing consumer adoption of cryptos and
the barriers consumers encounter. The upward trend in the transfer times and transaction fees for
crypto payments paid by payers may act as hurdles for consumers who want to use cryptos for peer-to-
peer payments or for paying online purchases. In that respect, it seems unlikely that crypto acceptance
by online retailers will rise substantially in the near future.

25
References
Aydogan, S. (2016), Insights on Consumer Payment Behavior: What Factors Determine and Allow to
Influence the Use of Payment Instruments?, PhD thesis, Vrije Universiteit Brussel.
Betaalvereniging (2016), Annual Report 2015, available via https://www.betaalvereniging.nl/wp-
uploads/2016/07/Annual-Report-Dutch-Payments-Association-2015.pdf.
Bolt, W. and M. van Oordt (2016), On the value of virtual currencies, DNB Working Paper 521.
Cameron, A. C. and P. K. Trivedi (2010) Microeconometrics using Stata. Texas: Stata Press.
Davis, F.D. (1989), Perceived usefulness perceived ease of use, and the user acceptance of information
technology. MIS Quarterly 13(3), 319-340.
DNB (2007), Customer satisfaction with Internet payments, DNB Quarterly Bulletin March 2007, 33-
36.
ECB (2015), Virtual currency schemes – a further analysis, February 2015, Frankfurt.
Eurostat (2016), E-commerce statistics for individuals, 7 February 2017, available via
http://ec.europa.eu/eurostat/statistics-explained/index.php/E-
commerce_statistics_for_individuals
Gerhards, J. and R. Hackenbroch (2000), Trends and causes of cultural modernization. An empirical
study of first names. International Sociology 15(3), 501 – 531.
Halperin, D. (2013), Bitcoin, Information paper for the Payments System Board, May 2013 meeting,
RBA, available via https://www.rba.gov.au/information/foi/disclosure-log/pdf/131419.pdf
Hoff, K. and J.E. Stiglitz (2016), Striving for balance in economics: Towards a theory of the social
determination of behaviour, Journal of Economic Behaviour and Organization 126(B), 25-57.
Jonker, N (2007), Payment instruments as perceived by consumers – a public survey, De Economist
161(1),19 – 44.
Jonker, N. (2016), Regulating Interchange Fees for Card Payments in J. Górka: “Transforming
Payment Systems in Europe”, Palgrave Macmillan, Hampshire, 149 – 181.
Li, H., R. Ward and H. Zhang (2003), Risk, convenience, cost and online payment choice: a study of
eBay transactions, Working Paper, Georgia Institute of Technology.
Nakamoto, S. (2008), Bitcoin: a peer-to-peer electronic cash system, unpublished manuscript,
retrieved from https://bitcoin.org/bitcoin.pdf.
McAndrews, J. and Z. Wang (2008), The economics of two-sided payment card markets: pricing,
adoption and usage, Research Working Paper 08-12 (December 2008), The Federal Reserve
Bank of Kansas City.
Polasik, M., A.I. Piotrowska, T.P. Wisnieuwski, R. Kotkowski and G. Lightfoot (2015), Price
fluctuations and the use of bitcoin: an empirical enquiry, International Journal of Electronic
Commerce 20(1), 9-49.
Rogers, E.M. (1995), Diffusion of Innovations, 4th edition, The Free Press, New York.
Schuh, S. and O. Shy (2016), U.S. consumers’ adoption and usage of bitcoin and other virtual
currencies, working paper.

26
Silinskyte, J. (2014), Understanding Bitcoin adoption: Unified theory of acceptance and use of
technology (UTAUT) application, Master’s thesis, Leiden University.
Stavins, J. (2001), Effect of Consumer Characteristics on the Use of Payment Instruments, New
England Economic Review 3, 21-31.
Stevens, A. (2017), Digital currencies: Threats and opportunities for monetary policy, NBB Economic
Review, Chapter 5, June 2017.
Thuiswinkel.org (2017), Online bestedingen passeren de grens van € 20 miljard in 2016 (Online
expenditures exceeded EUR 20 billion in 2016), press release 3 March 2017, available via
https://www.thuiswinkel.org/nieuws/3348/online-bestedingen-passeren-de-grens-van-20-
miljard-in-2016.
Twenge, J., E.M. Abeke and W. K. Campell (2010), Trends in American Parents’ choices for
children’s names 1880 – 2007, Social Psychological and Personality Science 1(1), 19 – 25.
Van der Cruijsen, C, and F. van der Horst (2016), Payment behaviour: the role of socio-psychological
factors, DNB Working Paper 532.
Van Hove, L and F. Karimov (2016), The love of risk in retailers adoption of payment methods:
evidence for transition countries, Electronic Commerce Research 16(1), 27-72.
Venkatesh, V., M.G. Morris, F.D. Davis and G.B. Davis (2003), User acceptance of Information
Technology: Toward a unified view, MIS Quarterly 27(4), 425-478.
Venkatesh V., M.G. Morris, F.D. Davis and G.B. Davis (2003), User acceptance of Information
Technology: Toward a unified view, MIS Quarterly 27(4), 425-478.
Verdier, M. (2011), Interchange fees in card payment systems. A survey of the literature, Journal of
Economic Surveys 35(10), 273 – 297,
Vickers (2005), Public policy and the invisible price: competition law, regulation and the interchange
fee, Proceedings of a Conference on Interchange Fees in Credit and Debit Card Industries,
Federal Reserve bank of Kansas City, Santa Fe, New Mexico, 231 – 247.
WorldCoinindex (2017), 10 August 2017, available via https://www.worlVCoinindex.com.
Wright, J. (2004), The determinants of optimal interchange fees in payment systems, Journal of
Industrial Economics 2(1), 1 – 26.
Rochet, J.C. and J. Tirole (2002), Cooperation among competitors: some economics of payment card
associations, RAND journal of economics 33(4), 549 – 570.

27
Annex A. Sample characteristics
Table A.1: Decomposition sample by industry and firm size (number of workers)
Number of workers
Industry
1
2-4
5-19
20-99
>=100
n.a.
Total
Retail trade: Consumer
electronics, telecom & a white
goods
31
32
12
8
1
0
84
Retail trade: home, garden &
kitchen
57
38
14
6
5
0
120
Media & entertainment
43
12
15
7
1
0
78
Fashion
77
31
18
9
4
0
139
Travel (flights, hotels, etc.)
2
1
9
10
3
0
25
Sports & recreation
23
12
20
3
2
0
60
Tickets (parks, events, etc.)
1
1
8
13
12
0
35
Food & drinks
25
18
12
9
3
1
68
Health & personal care
20
15
5
2
0
0
42
Other products /services
57
26
22
9
3
0
117
Total
336
186
135
76
34
1
768

28
Annex B. Robustness check
TABLE B.1: Adoption intention and acceptance crypto payments using restricted and unrestricted
samples
Dependent variable: Adoption intention
Dependent variable: Acceptance
Model 1
(Table 4)
Model 1
restricted
sample
Model 2
(Table 4)
Model 2
restricted
sample
Model 1
(Table 5)
Model 1
restricted
sample
Model 2
(Table 5)
Model 2
restricted
sample
β
β

β
β
β

β
Age (yrs)
-0.022**
-0.019***
-0.022***
-0.018***
-0.030***
-0.030***
-0.025***
-0.026***
(0.005)
(0.006)
(0.005)
(0.006)
(0.008)
(0.008)
(0.008)
(0.009)
Education: Bachelor degree
0.087
0.047
0.097
0.037
0.098
0.037
0.023
-0.044
(0.116)
(0.132)
(0.115)
(0.134)
(0.192)
(0.202)
(0.199)
(0.211)
Education: Master degree
0.038
0.062
0.074
0.076
0.328
0.261
0.233
0.180
(0.157)
(0.181)
(0.160)
(0.186)
(0.224)
(0.237)
(0.235)
(0.247)
Firm age: less than 2 years
0.224
0.349*
0.244
0.390**
-0.206
-0.222
-0.289
-0.296
(0.155)
(0.187)
(0.155)
(0.189)
(0.275)
(0.293)
(0.300)
(0.316)
Firm age: 2 – 5 years
0.264*
0.392**
0.277**
0.418**
0.261
0.252
0.265
0.266
(0.139)
(0.165)
(0.138)
(0.167)
(0.218)
(0.231)
(0.244)
(0.235)
Firm size: 1 person
0.056
0.023
-0.003
-0.036
0.176
0.136
-0.106
-0.174
(0.193)
(0.218)
(0.182)
(0.210)
(0.315)
(0.334)
(0.204)
(0.308)
Firm size: 2 – 4 people
0.010
-0.095
0.008
-0.087
0.245
0.189
0.127
0.047
(0.198)
(0.222)
(0.192)
(0.225)
(0.314)
(0.338)
(0.315)
(0.337)
Firm size: 5 – 19 people
0.004
-0.108
-0.007
-0113
-0.663*
-0.648
-0.732*
-0.765*
(0.187)
(0.210)
(0.183)
(0.211)
(0.392)
(0.413)
(0.418)
(0.430)
Uses services PSP
0.328***
0.407***
0.296**
0.333**
0.753***
0.766***
0.715***
0.712***
(0.122)
(0.150)
(0.122)
(0.150)
(0.230)
(0.246)
(0.256)
(0.270)
Sector: media
0.338*
0.330
0.446**
0.431**
0.014
0.136
0.139
0.327
(0.186)
(0.224)
(0.182)
(0.215)
(0.296)
(0.324)
(0330)
(0345)
Sector: electronics
0.510***
0.712**
0.533***
0.725***
-0.151
-0.159
0.024
-0.016
(0.191)
(0.227)
(0.184)
(0.220)
(0.254)
(0.271)
(0.247)
(0.264)
Competition: no to weak
0.071
0.122
0.038
0.024
0.059
0.101
0.186
0.177
(0.191)
(0.241)
(0.188)
(0.236)
(0.332)
(0.357)
(0.332)
(0.363)
Competition: strong to perfect
-0.046
0.011
-0.104
-0.080
0.276
0.273
0.435*
0.384
(0.121)
(0.147)
(0.124)
(0.150)
(0.217)
(0.230)
(0.226)
(0.236)
Consumer adoption crypto
Customers: mainly male
-0.024
-0.090
0.345*
0.279
(0.157)
(0.174)
(0.208)
(0.214)
Customers: mainly female
-0.283**
-0.293*
-0.628**
-0.618**
(0.134)
(0.164)
(0.283)
(0.289)
Customers: mainly 30 years or younger
0.037
-0.072
-0.074
-0.096
(0.153)
(0.182)
(0.251)
(0.267)
Assessment consumer adoption crypto
0.016**
0.018**
-0.029**
-0.022*
(0.006)
(0.008)
(0.014)
(0.013)
MV assessment consumer adoption crypto
-0.399***
0.143
(0.145)
(0.189)
Retailer’s net transactional benefits
Relatively favourable cost crypto
0.233**
0.217*
0.435***
0.406***
(0.115)
(0.119)
(0.128)
(0.136)
Relatively favourable labour time cost
crypto
0.458**
0.428*
0.546*
0.504
(0.204)
(0.220)
(0.309)
(0.323)
Relatively favourable safety crypto
0.201
0.128
0.606**
0.616**
(0.183)
(0.211)
(0.268)
(0.275)
Exchange rate risk crypto
-0.066*
-0.063
-0.111*
-0.090
(0.039)
(0.039)
(0.063)
(0.058)

29
Table B.1 Continued
Dependent variable: Adoption intention
Dependent variable: Acceptance
Model 1
Model 1
restricted
sample
Model 2
Model 2
restricted
sample
Model 1
Model 1
restricted
sample
Model 2
Model 2
restricted
sample
β
β

β
β
β

β
Customers: within euro area
-0.099
0.180
-0.326
-0.309
(0.126)
(0.149)
(0.202)
(0.209)
µ1 (adoption intention)/
Constant (acceptance)
0.171
0.279
0.690*
0.766*
-1.221***
-1.058**
-2.407***
2.161***
(0.316)
(0.363)
(0.412)
(0.464)
(0.447)
(0.476)
(0.632)
(0.633)
µ2
1.166***
1.339***
1.693***
1.844***
(0.319)
(0.366)
(0.416)
(0.471)
No. of observations
650
444
650
444
761
521
761
521
Log likelihood
-458.38
-340.86
-456.06
-336.206
-133.99
-124.37
-121.19
-122.62
Pseudo R-squared
0.083
0.086
0.088
0.098
0.190
0.162
0.267
0.2413
Notes. The table shows coefficients (β) and average marginal effects (Mfx) based on ordered probit regressions with adoption intention as
dependent variable. Robust standard errors are between parentheses. The sample excludes retailers who accept crypto payments. Reference
characteristics of the firm are: firm’s age higher than 5 years, firm size: 20 people and more, does not make use of the services of a PSP, sector:
other than media or electronics, the firm experiences moderate competition, has a mixed clientele with respect to gender, the age of the firm’s
clientele is mixed or mainly people aged 31 years and older, the firm accepts payments within and outside the euro area. *p<.1, **p<.05, ***
p<.01 (two-sided t-tests).

30
TABLE B2: Correlation matrix key explanatory variables
Based on 521 respondents
Uses
services
PSP
Sector:
media
Sector:
electronics
No-weak
competition
Strong -
perfect
Competition
Customers
mainly
male
Customers
mainly
female
Customers
mainly
<=30 yrs
Consumer
adoption
crypto
Relatively
favourable
cost crypto
Relatively
favourable
labour time cost
crypto
Relatively
favourable
safety
crypto
Exchange
rate risk
crypto
Customers
within
euro area
Perceived
ease of
use
Perceived.
Compatibi
lity
Uses services
PSP
1.00
Sector: media
-0.17
1.00
Sector:
electronics
0.17
-0.10
1.00
Competition: no
to weak
-0.04
0.04
-0.04
1.00
Competition:
strong to perfect
0.01
0.00
0.06
-0.51
1.00
Customers:
mainly male
0.02
0.03-
0.09
0.08
-0.03
1.00
Customers:
mainly female
0.04
-0.10
-0.16
-0.07
0.06
-0.22
1.00
Customers:
mainly 30 years
or younger
0.11
-0.07
0.13
0.01
0.01
0.02
0.13
1.00
Assessment
consumer
adoption crypto
-0.15
-0.06
-0.08
-0.00
-0.03
0.01
0.15
-0.01
1.00
Relatively
favourable cost
crypto
0.05
-0.10
0.04
0.01
0.01
0.02
-0.01
-0.06
0.02
1.00
Relatively
favourable
labour time cost
crypto
0.07
-0.08
0.07
-0.07
0.04
0.06
-0.00
0.01
-0.00
0.31
1.00
Relatively
favourable
safety crypto
0.03
-0.10
0.04
0.02
-0.00
0.10
0.01
-0.07
0.03
0.23
0.23
1.00
Exchange rate
risk crypto
-0.02
-0.03
-0.00
-0.13
0.10
-0.01
-0.03
0.01
-0.10
-0.07
-0.11
-0.14
1.00
Customers:
within euro area
0.04
-0.13
0.07
-0.08
0.10
-0.12
0.12
-0.04
0.09
0.10
0.11
-0.04
-0.07
1.00
Perceived ease
of use
0.15
-0.02
0.14
-0.03
0.08
0.03
-0.06
0.07
-0.09
0.17
0.17
0.14
-0.05
-0.05
1.00
Perceived
compatibility
0.19
-0.03
0.18
-0.03
0.09
0.09
-0.07
0.13
-0.07
0.19
0.19
0.18
-0.10
-0.04
0.52
1.00

31
TABLE B.3: Variance inflation matrix explanatory variables
Based on 521 respondents
Variable
VIF
SQRT (VIF)
Age (yrs)
1.34
1.16
Education: Bachelor degree
1.24
1.11
Education: Master degree
1.26
1.12
Firm age: less than 2 years
1.69
1.30
Firm age: 2 – 5 years
1.62
1.27
Firm size: 1 worker
3.17
1.78
Firm size: 2 – 5 workers
2.70
1.64
Firm size: 6 – 19 workers
1.91
1.38
Uses services PSP
1.14
1.07
Sector: media
1.13
1.06
Sector: electronics
1.16
1.08
Competition: no to weak
1.43
1.19
Competition: strong to perfect
1.41
1.19
Customers: mainly male
1.12
1.06
Customers: mainly female
1.27
1.13
Customers: mainly 30 years or younger
1.13
1.06
Assessment consumer adoption crypto
1.14
1.07
Relatively favourable cost crypto
1.18
1.09
Relatively favourable labour time cost crypto
1.26
1.12
Relatively favourable safety crypto
1.19
1.09
Exchange rate risk crypto
1.08
1.04
Customers: within euro area
1.13
1.07
Perceived ease of use
1.47
1.21
Perceived compatibility
1.54
1.24
Mean VIF
1.45

Previous DNB Working Papers in 2018
No. 583
Dorinth van Dijk, David Geltner and Alex van de Minne, Revisiting supply and demand
indexes in real estate
No. 584
Jasper de Jong, The effect of fiscal announcements on interest spreads: Evidence from the
Netherlands

De Nederlandsche Bank N.V.
Postbus 98, 1000 AB Amsterdam
020 524 91 11
dnb.nl