Behavior or culture? Investigating the use of cryptocurrencies for electronic commerce across the USA and China

2.1 Cryptocurrencies: a theoretical premise and a comparison with stablecoins

Cryptocurrencies are based on an open-source, public and decentralized network; owing to this, the entire transaction history is stored in a series of chains, which refer to the “blockchain” (Ammous, 2018) – allowing to bypass the established banking system. Cryptocurrencies first appeared publicly in January 2009 with the advent of Bitcoin, a digital representation of value ideated by a computer programmer named Satoshi Nakamoto (diffusely considered as a nickname). Yet, these new electronic commerce means are not issued or controlled by any specific institute and can be used to securely store and transfer any number of values anywhere in the world without the permission of a third party (e.g. banks and their systems). This point, however, is both an advantage and a disadvantage. Indeed, the lack of control of a qualified third-party (i.e. bank) over cryptocurrencies’ transactions, does not allow to check whether users are operating legal or illegal operations (e.g. money laundering) (Dierksmeier and Seele, 2018).

However, scholars interested in the investigation of cryptocurrencies – still not a huge number – have pointed out some beneficial trends that are favoring the adoption of this new mean for electronic commerce. DeVries (2016), for example, highlighted that “the ability for cryptocurrency to perform microtransactions may allow it to bridge an economic gap that traditional state sponsored currencies would not be able to solve” (p. 8); and the same would apply to smart contracts, thus programmed payments happen when a set condition occurs.

Nevertheless, despite having a market capitalization close to $2.6tn (CoinMarketCap, 2021b) and exponential growth of the possibility of using cryptocurrencies in everyday life (Treiblmaier et al., 2020; Bezovski et al., 2021), the cryptocurrency market is characterized by extreme volatility. For instance, scholars like Corbet et al. (2020) have tracked the 1-h fluctuations during a year time span (2019) recording extreme values of −23.91% (Bitcoin Cash) and +40% (Stellar) of loss and gain, respectively.

Stemming from the above, a new bundle of digital assets, always based on blockchain technology, have appeared in the past years: stablecoins. These are defined as a particular kind of cryptocurrencies that – by combining collaterals (such as precious metals and fiat currencies) or algorithmic mechanisms of stabilization with the management of the supply – hold a stable value against a target price, which is usually $1 (Dell'Erba, 2019). Precisely, the first stablecoin, i.e. Tether, was created in 2014 as to offer a solid alternative to cryptocurrencies, such as Bitcoin, Ethereum and Ripple (Bullmann et al., 2019; Baur and Hoang, 2021). Nowadays, according to CoinMarketCap (2021a), Tether represents more than 50% ($73bn) of the worldwide stablecoins market capitalization ($141bn). However, different types of stablecoins exist according to the adopted stabilization mechanisms (Berentsen and Schär, 2019):

• commodities-backed stablecoins: whose value is anchored to real estate or gold or other precious metals’ value;

• fiat currencies-backed stablecoins: whose value is anchored to standard currencies such as US Dollar, Euro or Chinese Yuan;

• cryptocurrencies-backed stablecoins: whose value is anchored to cryptocurrencies, such as Bitcoin or Ethereum; or

• seigniorage-backed stablecoins: meaning that they do not rely on any kind of collateral value except on the algorithms that control the supply volume to keep constant their market value.

From the above, cryptocurrencies and stablecoins share and differ on some features, see Table 1. About the commonalities, cryptocurrencies and stablecoins are both:

• founded upon the blockchain; and

• suffering of a paucity of international, and often even national, legal regulations (Iredale, 2021; DeNicola, 2021).

Instead, for what concerns the differences, stablecoins diverge from cryptocurrencies on different aspects (Geroni, 2021; Lorentz, 2021), such as:

• Volatility for conventional cryptocurrencies is high and proved, while stablecoins have a partially confirmed stability.

• Collaterals for conventional cryptocurrencies are featured by the absence of collaterals, while stablecoins are baked by different kinds of collaterals as aforementioned.

• Control for conventional cryptocurrencies is actively played by investors/users who can leverage volatility, while stablecoins being pegged with third assets do not allow investors/users to leverage their value (Carstens, 2019).

However, despite stablecoins’ features can push toward thinking they have a greater potential than cryptocurrencies in electronic commerce, a recent study did not support their superior performance. In particular, Baumöhl and Vyrost (2020) analyzed 1-min price variations, for the entire year 2019, of the five largest cryptocurrencies (Bitcoin, Ethereum, Ripple, Bitcoin Cash and Litecoin) and the six largest stablecoins (Tether, USD Coin, True USD, Paxos Standard Token, Dai and Gemini Dollar) on 18 different exchange markets. Results shown that only three stablecoins (True USD, Paxos Standard Token and Gemini Dollar) may serve as “safe havens” – assets that are uncorrelated with other assets during times of market turmoil (Wang et al., 2020; Baur and Hoang, 2021). Thus, despite stablecoins’ apparently “safer” characteristics, they are not always consistently stable over time; in other words, stablecoins are not generally superior to cryptocurrencies.

2.2 Cryptocurrencies in electronic commerce

Electronic commerce became a worldwide business model (Treiblmaier and Sillaber, 2021), with retail electronic commerce sales amounting to US$2.89tn in 2021 and an expected increase to US$6.39tn by 2024 (Statista, 2021a). In comparison, the total market capitalization of cryptocurrencies amounted to US$2.89bn in 2021 (Statista, 2021b), with a big jump from the US$900mn in 2020 that surpassed any reasonable projection. Taking into account already discussed pros and cons of cryptocurrencies, such as being based on the blockchain technology, they have the potential to facilitate electronic sales that are trustless and allow performing them without dedicated intermediaries (Treiblmaier and Sillaber, 2021). In this regard, however, only a tiny amount of platforms – around 2% according to Jonker (2019) – are accepting cryptocurrencies, but, the volume of transactions of goods and services paid by cryptocurrencies is already impressing: over $4bn in 2019 (Popper, 2020) and a forecasted exponential growth all over the world is expected (Treiblmaier and Sillaber, 2021).

Related to the above, an increasing number of organizations – such as AXA Insurance, AT&T, Expedia, Tesla, Visa, KFC, Burger King (in Germany), Norwegian Air and AirBaltic – allow their clients to pay by using cryptocurrencies (Walsh, 2021). In this vein, some of the most famous precursors have been Virgin Galactic and Microsoft which, since 2013 and 2014, respectively, started allowing customers to book trips into space and to buy digital contents by Bitcoins (Cryptolocally, 2020). In addition, rumors claim that soon also Amazon will accept payments by the mean of cryptocurrencies (Day and Chiglinsky, 2021).

As a result of the reported cryptocurrencies’ exciting dynamics, many scholars have devoted their research to highlight the main e-commerce platforms’ features that might favor or, conversely, interfere with the users’ willingness to adopt cryptocurrencies as a mean of payment during online transactions. In particular, according to Mendoza-Tello et al. (2018), the main issue that restrains consumers to fully embrace cryptocurrencies for online shopping activities lays on the reliability of transactions – delegated to a cryptographic test and not to the supervision of a control authority. To develop possible solutions to this lack of trust in the usage of cryptocurrencies on e-commerce platforms, Mendoza-Tello et al. (2018) have analyzed the questionnaires administered to 125 e-commerce users. Results of their research point out that “social commerce” – defined as an internet-based commercial application supporting social interaction and user-generated content, which influence decision purchases – may represent a solution to overcome such lack of trust in cryptocurrencies use. In fact, these authors claimed that the features provided by social commerce platforms, such as the users’ ability to exchange advices foster the creation of relationships based on mutual trustworthiness, which – by leveraging the overall participation of users – may positively enhance the propagation of higher volumes of information concerning the feasibility of adopting cryptocurrencies as mean of payment.

In the same vein, Treiblmaier et al. (2020), by surveying a sample population of 161 Asian travelers, highlighted the most impacting factors that foster, or not, the buyers’ willingness to pay by the mean of cryptocurrencies on electronic travel-related platforms. On one hand, they have found evidence that electronic commerce’s ability to furnish – in terms of proper information on information technology standards – adequate levels of security, easiness of use and customer care foster users’ enthusiasm toward payment made by the use of cryptocurrencies. On the other hand, the presence of insufficient levels of the aforementioned factors pushes shoppers to use the traditional payment methods (e.g. debit card and bank transfer). Treiblmaier and Sillaber (2021) furtherly advanced such findings by adding that electronic commerce platforms that held their transactions via blockchain can increase a greater sense of trust in buyers when the platform is committed in storing purchase-related data in an unchangeable way; this, indeed, helps improve data provenance and traceability, fostering the buyers’ feeling of confidence in using cryptocurrencies – in line with Lo et al. (2017).

Therefore, from the aforementioned contributions, it is evident that the main concerns about using, or not, cryptocurrencies in electronic commerce lay on the perceived trust and security that the online platforms are able to communicate to their users (Jonker, 2019; Bezovski et al., 2021).

3.1 General factors influencing the intention to use cryptocurrencies for electronic commerce

As introduced, some scholars have already investigated the factors that influence using cryptocurrencies for electronic commerce. Among these scholars, it is noteworthy to recall the contribution brought by Li and Wang (2017) who have, through the use of a theory-driven empirical investigation concerning Bitcoin vs USD exchange rate determination, firstly indicated that the use of cryptocurrencies is going to be more affected by the economic factors (i.e. cryptocurrencies’ interest rate, trade volume and price volatility) and less affected by the technological ones (e.g. troubles concerning the mining technologies and processes). Nevertheless, the investigation completed by Li and Wang (2017) was carried out without checking the major determinants that push respondents to use, or not, cryptocurrencies. In contrast to this latter, Sohaib et al. (2019), using a survey administered to over 100 students and staff from the University of Technology Sydney, have found that technology readiness (i.e. discomfort and insecurity) of cryptocurrencies has a significant relationship with technology acceptance (i.e. perceived ease of use and perceived usefulness). From that, early adopters of technology are the more willing to use cryptocurrencies and this attitude increases as the complexity in using cryptocurrencies decrease. This last result was previously intuited by Shahzad et al. (2018) who, moreover, found that awareness and perceived trustworthiness toward cryptocurrencies are comparatively significant factors to determine the intention to use them in electronic commerce. However, none of the reported studies adopted a clear and recognized model, such as the TPB that clearly links behavioral factors with the intention to use cryptocurrencies as well as having determined the influence of culture.

3.2 Theory of planned behavior factors influencing the intention to use cryptocurrencies for electronic commerce

The TPB was conceptualized by Ajzen (1991, 2002) as a direct evolution of the theory of reasoned action (TRA). According to TPB’s principles, economic agents’ willingness to conduct a specific action can be forecasted by paying attention to the:

• attitude toward the consequences, resulting from that specific action as well as the expectation about the effects originated from it;

• subjective norms (also known as normative belief), this bundle of rules reflects the consciousness that specific behavior is or is not envisioned by individuals significant for the investor (e.g. family); and the

• perceived behavioral control concerns how well people can handle actions needed for dealing with forthcoming situations.

Specifically, the TPB unquestionably diverges from the TRA since the involvement of the perceived behavioral control variable, which has been confirmed, guiding to make improved forecasting in terms of the tendency of translating an intention into a behavior than the former (Madden et al., 1992).

Ordinarily, TPB has been repeatedly adopted in a broad array of behavioral investigations, such as envisioning customers’ intention to pick a specific product, people’s willingness to start a business or even the individuals’ preferences for what concerns household financing (Ajzen, 1991).

About the attitude variable, studies that implement TPB for the investigation of consumers’ intention toward electronic means (Liu et al., 2019; Sardanelli et al., 2019) generally found a positive influence of attitude toward their adoption. About the use of cryptocurrencies, in particular, Gazali et al. (2019) have studied the correlation between attitudes, subjective norms, financial risk tolerance and perceived benefits from (the last two variables have been conveniently added to the theoretical model) the intention to use cryptocurrencies – in particular, Bitcoin – highlighting the positive influence, played by the aforementioned factors, on the willingness to use cryptocurrencies for electronic commerce. Nevertheless, as previously claimed, the findings advanced by Gazali et al. (2019) can be considered as not fully satisfactory because of the narrow sample size (45 respondents). Yet, in their recent contribution, Mendoza-Tello et al. (2018) have empirically explained, by the use of a survey administered to over 100 respondents, that users see some psychological benefits in cryptocurrencies, among the ones the most claimed was “feeling safe,” that elicit their intention to use them. From that, we hypothesize the following: (Table 7)

Behavioral consumers’ researchers have studied the influential role played by subjective norms, through the implementation of the TPB lens, in shaping how usage decisions are made. In this vein, scholars such as Arias-Oliva et al. (2019) have not found any important influence of this variable for what concerns the willingness to use cryptocurrencies for electronic commerce purposes. Other researchers, instead, have found hints suggesting that cryptocurrencies users’ choices are frequently made in correspondence to the suggestions proposed by close people such as colleagues, friends and relatives (Liu et al., 2019) – that, sometimes, are intentionally looked for (Sondari and Sudarsono, 2015). Concluding, subjective norms can progressively pressurize cryptocurrency users that have to make decisions – also in circumstances where they have not received any kind of suggestions; indeed, individuals appear to act resolutely also just observing the behaviors of others, and from them they are socially influenced. Results on the positive influence, in terms of raising the intention to use cryptocurrencies for electronic commerce, some subjective norms have been verified also by other recent studies about cryptocurrency’s consumer behavior (Gazali et al., 2019). Therefore, H2 could be stated as follows:

Liu et al. (2019) in their meta-analysis about the factors influencing consumers’ mobile payment behavior found that there is a significantly positive influence of the perceived ease of use on consumers’ attitude toward mobile payment. In practice, the user that does not experience troubles in the adoption of a new mean is more prone to its permanent adoption because generating a sense of control and usefulness of the new mean. This can be seen in the study proposed by Ali (2011), according to which perceived behavioral control represents a factor for carrying out a specific behavior. Specifically, these scholars have found a positive level of influence played by behavioral control on the willingness to use cryptocurrencies for electronic commerce. The same positive influence has been tested by Arias-Oliva (2019) on the intention to adopt cryptocurrencies in the Spanish market. Thus, H3 could be advanced as follows:

3.3 Other behavioral factors influencing the intention to use cryptocurrencies for electronic commerce

Narayanan et al. (2016) conceptually advanced that users of cryptocurrencies may be attracted by their ability to trade illegal products/services without being traced, because of the anonymity provided; because of the latter, tax evasion is another emerging big concern for regulators. This has been recently advanced also by Dyntu and Dykyi (2018) through an analysis of historical stages of cryptocurrency creation and cases of money laundering, where criminals who used cryptocurrency have been identified and charges have been pressed; it resulted in anonymity and decentralization, i.e. the two main innovative features of cryptocurrencies, which are the characteristics that push people to use them for illegal activities. Above results are in line to what emerged in prior studies looking at deviant behavior of consumers in digital channels (Higgins et al., 2008; Williams et al., 2010). From that, we hypothesize the following:

Building on the results of Kengatharan and Kengatharan (2014), da Gama Silva et al. (2019) recently tried demonstrating the presence of herding behavior among cryptocurrency users. In particular, through an indirect analysis (based on cross-sectional absolute deviation and cross-sectional standard deviation tests) of the 50 most liquid and capitalized cryptocurrencies, they found that cryptocurrency users tend to follow and copy what other users are doing within the cryptocurrency market – leading to the excess of volatility and short-term trends that feature in this market (Liu and Tsyvinski, 2018). Results of this study have been later supported by Gurdgiev and O’Loughlin (2020) – despite the fact they did not directly test this behavioral variable in the same way as da Gama Silva et al. (2019). Therefore, we propose the following hypothesis:

About the perception of risk on the intention to use cryptocurrencies for electronic commerce, Arias-Oliva et al. (2019) did not find any significant influence in their empirical research comparing characteristics and behavior of cryptocurrency and noncryptocurrency users (in line with Sardanelli et al., 2019). However, Pelster et al. (2019) found that the overall behavior of cryptocurrency users is driven by excitement-seeking; in particular, when engaging in cryptocurrency, users simultaneously increase their risk-seeking behavior in exchanging them as they increase their trading intensity. This is in contrast to the meta-analysis of Liu et al. (2019) on the factors influencing consumers’ adoption of mobile payment (i.e. perceived risk toward a new exchange mean negatively influences the intention to use it); but, the context of adoption is substantially different from the prior reported words and the one at the center of this proposed work. Accordingly, we propose the following:

To explain the gender gap in the knowledge of cryptocurrency characteristics, Bannier et al. (2019) found that measures for actual financial literacy accounts for approximately 40% of the gender gap in Bitcoin literacy. This proposes financial literacy as an explanatory variable of the behavior of users toward cryptocurrencies – in line with other works assigning value to the financial literacy variable to explain the willingness to use cryptocurrencies for electronic commerce (Stolper and Walter, 2017). However, the same was not found by Arias-Oliva et al. (2019), whose empirical analysis of behavioral variables influencing users showed that financial literacy did not have a significant influence on the intention to use cryptocurrencies for electronic commerce. This last result is in contrast with the important discovery by Lusardi and Mitchell (2014), who found a positive result for this relationship through their review of empirical papers and their resultant findings on the influence of financial literacy on economic decision making. Accordingly, we propose the following:

3.4 Hofstede’s cultural dimensions and their interrelationships with theory of planned behavior and the other behavioral factors

From time to time, studies that implemented TPB also included cultural influences as to be comprehensive in the explanation of what “really” drives the intention to act, mainly using the Hofstede model of cultural dimensions (Hassan et al., 2016). Hofstede (2001) identifies five dimensions of national culture, namely:

• individualism, i.e. the degree to which individuals are integrated into groups, such as that individualistic societies show loose ties between individuals and seek forms of society that are more independent;

• power distance, i.e. the extent to which the less powerful members of institutions and organizations within a country expect and accept that power is distributed unequally;

• uncertainty avoidance, i.e. the extent to which people feel threatened by ambiguous situations and create beliefs and institutions in an attempt to avoid them, such as that people with high uncertainty avoidance seek stability, predictability and low risk rather than change and new experiences

• masculinity, i.e. the degree to which a society is characterized by either assertiveness or nurturance, such as that more masculine countries place greater emphasis on wealth, success, ambition, whereas more feminine countries place greater value on people and helping others; and

• long-term orientation, i.e. the extent to which a society exhibits a pragmatic, future-oriented perspective rather than a conventional historic or short-term perspective.

In particular, stemming from the fact that favorable attitude is expected to ease online transactions (Pavlou, 2002) and reduce barriers to the adoption of cryptocurrencies for electronic commerce, Pavlou and Chai (2002) – through the adoption of TPB and Hofstede’s cultural dimensions – found that attitude positively influences consumer behavioral intentions to engage in product purchases across the USA – an individualistic society – and China – a collectivist society – such that relationship between attitude and transaction intention is stronger in collectivist than in individualist societies. This happens because members of a collectivist culture would like to maintain harmonious consumer–vendor relationships. Moreover, members of individualistic societies tend to be independent, while those in collectivistic cultures endorse the interdependent natures of their societies and obligations to the group (Pavlou and Chai, 2002), accepting to be influenced by the societal norm (one operationalization of subjective norm; Hofstede, 2001) while building their intention to act. From that, we propose the following:

Collectivist societies are also featured by a high power distance: people accept a hierarchical system with an unequal power distribution (Hofstede, 2001). This leads, of course, to a greater social influence – which reflects family, friends’ and influencers’ opinions – and to a positive relationship with electronic commerce intentions (Pavlou and Chai, 2002). The same result has been recently supported by other cross-cultural studies on the intention to trade within e-commerce web-sited and social media commerce platforms (Han and Kim, 2018). From that, we propose the following:

Chinese culture, in particular, and collectivist cultures in general, influence the group norm leading to avoid illegal attitudes. In fact, as demonstrated by Huang and Lu (2017), consumers with a stronger collectivist orientation are more likely to regard questionable consumer activities as unethical, mainly because of the fact that they treat group norms as the standard by which to measure behavior. However, while the above stated relationship seemed to be valid in public life, the same has not been found for private businesses. Indeed, as demonstrated by the work of Chan et al. (1998) investigating Chinese consumer unethical practices, Chinese consumers framed buying counterfeit products as a “questionable action” but not as an unethical one. If these illegal acts are committed in private life and there is no harm for the businesses, they will tend to do illegal activities; the same has been recently found in the study of Sardanelli et al. (2019) that for investigating the adoption of on-line streaming services, considered an Italian sample of consumers (closer to Americans in terms of cultural dimensions; Hofstede, 2001). However, if people of collectivist cultures feel that these illegal activities will finance organized crime or they have an attitude to business, as assumed for cryptocurrencies (Foley et al., 2019), they will not act illegally (Chang et al., 2019). Stemming from that, and from how questions within the questionnaire have been posed, it is possible to state:

According to Hofstede (2001), collectivist societies, like China, are featured by a long-term orientation: people feel free to postpone decisions till they are comfortable with their expected consequences, giving such people more control over their actions. As found by Pavlou and Chai (2002), Chinese users try to be more in control of their electronic commerce transactions than the US users and rely on this control in their online transaction behavior. This has been also confirmed by following studies (Han and Kim, 2018).

5.1 Measure validation

About the discriminant and convergent validity of the scales, they were initially investigated through an exploratory principal component factor analysis – with a Varimax rotation (Table 2). Component factor analysis is a statistical approach usually implemented for data reduction by creating one or more index variables (often referred to as factors, components and dimensions) from a larger set of measured variables (Field, 2013).

All the items of the questionnaire significantly loaded on the constructs; using the 0.40 rule-of-thumb, all cross-loadings are low, while the resulting solution explained 78% of the total variability. Then, through the implementation of the PLS, a confirmatory factor analysis was implemented to determinate discriminant validity of the principal constructs; thus, to test whether the data fit the hypothesized measurement model (Field, 2013). As usually done, discriminant validity is shown when the square root of each construct’s average variance extracted (AVE) is larger than its correlations with other constructs.

Tables 3 and 4 report that the square root of the AVE is much larger than all other cross-correlations for both the USA and China samples. Tables 3 and 4 report the:

• reliability coefficients and AVE values;

• correlation matrix and descriptive statistics of the study’s principal factors for the USA and China; and

• Cronbach’s alpha coefficient for each construct as to measure reliability (for both countries they are above 0.70).

5.2 Hypotheses testing

The hypotheses developed in the prior sections have been tested by recurring to a PLS. PLS is able to specify the relationships among the principal construct, as well as with their underlying items. The possibility to insert several measurements for each construct allows accurate estimates of the paths among constructs. Figure 2 depicts the results for the USA and China, widely supporting the role of the TPB and other behavioral variables for the intention to use cryptocurrencies in electronic commerce. The model is valid in both the countries presenting the following R² values: 0.82 for U.S. and 0.75 for China.

To test the moderating effect of cultural differences on TPB and other related behavioral variables, the test for moderators was implemented (Baron and Kenny, 1986). As reported in Table 5, and similarly to Pavlou and Chai (2002), two regression models have been compared:

• the simple regression model with seven independent variables; and

• the moderated regression model that is additionally comprehensive of the interaction of the seven behavioral variables with culture.

In particular, simple regression predicts a linear relationship, Y = α + βX + ε, between a dependent variable Y and an explanatory variable X, with the error term ε encompassing omitted factors. The moderated regression model, instead, is usually adopted to identify factors that change the relationship between independent and dependent variables (i.e. moderators) (Field, 2013).

Culture moderates the relationship, as emerging from the significant interaction effect. Indeed, as stated in H1 and H8a, attitude positively influences the adoption of cryptocurrencies and, related to that, collectivist rather than in individualist cultures lead to have a stronger attitude toward the adoption of cryptocurrencies in electronic commerce. These hypotheses are both verified, as it can be seen in Figure 2 (China, b = 0.51, p < 0.01; the USA, b = 0.21, p < 0.05), also because the moderated regression model, depicted in Table 5, shows that the interaction of culture with attitude is significant (b = 0.43, t = 5.12, p < 0.001). Similarly, H2, H8b and H9a, which states that subjective and societal norms positively influence the intention to use cryptocurrencies for electronic commerce, and this relationship is stronger in collectivist/high power distance cultures rather than individualist/low power distance ones, received support. Indeed, it resulted that Chinese users of cryptocurrencies (b = 0.61, p < 0.01) are influenced by their society and social norms, while their counterparts in the USA are not affected at all by them (b = 0.11, p > 0.05); the moderated regression model also shows that the interaction of norm with culture is significant (b = 0.59, t = 2.12, p < 0.01), providing additional support for H8b and H9a. H3 argues that perceived behavioral control positively influences the intention to use cryptocurrencies for electronic commerce, and this hypothesis has been verified as well (China, b = 0.12, p < 0.05; the USA, b = 0.36, p < 0.01); however, culture does not seem to moderate this relationship (b = 0.20, t = 1.43, p > 0.05); thus, H11a is rejected. About the hypothesized positive influence of illegal attitude toward the intention to use cryptocurrencies for electronic commerce, i.e. H4, results shown in Figure 2 depict that this relationship is positive for the USA (b = 0.49, p < 0.01), but negative for China (b = −0.36, p < 0.01). This relationship seems to be strongly moderated by culture, as advance by H10a and H10b and shown in Table 5 (b = −0.19, t = −1.54, p < 0.01). H5 advanced that herding behavior positively influences the intention to use cryptocurrencies for electronic commerce; this has been found as verified for China (b = 0.46, p < 0.01) and the USA (b = 0.17, p < 0.05) users. Moreover, this result seems to advance that culture plays a role within this relationship, indeed the moderated regression model also shows that the interaction of herding behavior with culture is significant (b = 0.21, t = 2.11, p < 0.05), providing additional support for H9b. H6 advanced that perception of risk positively influences the intention to use cryptocurrencies for electronic commerce, this relationship is verified and is stronger for the USA (b = 0.34, p < 0.01) than China (b = −0.23, p < 0.01), where this relationship is negative; this is also supported by the moderated regression model (b = −0.36, t = 1.98, p < 0.01), verifying H11b. Finally, H7, which states that financial literacy positively influences the intention to use cryptocurrencies for electronic commerce, did not find support (China, b = 0.32, p > 0.01; the USA, b = 0.24, p > 0.05).

As reported in Table 6, most of the research hypotheses at the basis of this study are supported. Culture has been found to moderate the relationships between attitude, subjective and social norm, illegal attitude, herding behavior, perceived risk and the intention to use cryptocurrencies for electronic commerce; in other words, the role of cultural differences was found to be a noteworthy moderator in the proposed cryptocurrencies adoption model. However, within this model, financial literacy seems not to be a driver of the intention to use cryptocurrencies for electronic commerce; this outstanding result, together with the others, will be discussed in the following section.