The early bird catches the worm: The role of regulatory uncertainty in early adoption of blockchain’s cryptocurrency by fintech ventures

ABSTRACT Regulatory uncertainty about a technology confronts new technology-based firms (NTBFs) with questions of whether or not to adopt this technology. Following institutional theory, NTBFs are expected not to adopt a technology until regulatory uncertainty has been reduced. However, following the resource-based view, NTBFs are expected to adopt the technology under regulatory uncertainty due to limited or no regulation, as this provides them with an opportunity to secure competitive resources. We investigate whether regulatory uncertainty enabled or inhibited 108 fintech ventures in adopting blockchain’s core application, cryptocurrency, in a time when governments were still considering potential regulation. Our findings indicate that regulatory uncertainty has a positive effect on NTBFs’ adoption of technology. We extend our knowledge on the role of regulatory uncertainty in technology adoption and we shed light on the boundary conditions of both the resource-based view and institutional theory. Further, we reflect on regulating the “winds of change.”


Introduction
Since bitcoin was introduced in 2009 (Bitcoin Project, n.d.), many different uses have been created for blockchain technology (Casino et al., 2019).Blockchain technology's most-used application is cryptocurrency (Bellavitis et al., in press).The enormous growth and investment hype in cryptocurrencies have caused several governments around the world to consider the possibilities and need to regulate this technology and its application, resulting in mixed messages on future regulation of cryptocurrencies.Recently, El Salvador was the first country to make bitcoin legal tender (Renteria et al., 2021) and Cuba is likely to follow soon (Al Jazeera, 2021).Most (inter) governmental institutes, however, send mixed messages regarding potential regulatory restrictions.While the European Parliament acknowledges benefits of cryptocurrencies (Dabrowski & Janikowski, 2018), at the same time, it is in the process of regulating initial coin offerings (ICOs), the process of CONTACT Arjan J. Frederiks a.j.frederiks@rug.nlP.O.Box 72, Groningen 9700AB, Netherlands introducing a new cryptocurrency (Fox, 2018).The US Department of Treasury has offered interpretative guidance on regulating use of cryptocurrencies (Financial Crimes Enforcement Network, 2013) that are said to be "choking Bitcoin entrepreneurs" (Matonis, 2013).At the same time, the Federal Reserve Board and Federal Advisor Council appear to have adopted a "watchful waiting" policy (Federal Reserve, 2014;Murphy et al., 2015, p. 13), leaving many fintech companies in uncertainty regarding possible regulations and policies.These mixed messages lead to regulatory uncertainty (Hoffmann et al., 2009).Business research has identified uncertainty in decision-making as a core challenge for businesses and for decades has studied its effects extensively (Lawrence & Lorsch, 1967;Miles & Snow, 1978;Thompson, 1967).However, the literature presents ambivalent conclusions regarding regulatory uncertainty's effect on adopting a new technology (compare, Hoffmann et al., 2009).Based on an institutional theory perspective, some scholars argue that uncertainty of external institutional factors, such as regulation, causes risk for individual companies (Meyer & Rowan, 1977), including small businesses (Fauzi & Sheng, in press).Due to elevated risk levels, technology adoption investments might be postponed to a time when risk has been reduced and hence uncertainty is lower (Yang et al., 2004).However, based on a resourcebased view (Barney, 1991;Kraaijenbrink et al., 2010), others argue that proactive firms try to prevent the organization's valuable resources from becoming vulnerable through possible future regulation (Aragón-Correa & Sharma, 2003).A technology adoption investment made under regulatory uncertainty, can secure competitive resources for companies (Hoffmann et al., 2009) by giving them a first-mover advantage (Lieberman & Montgomery, 1988), leveraging complementary resources (Hoffmann et al., 2009), influencing the direction of new regulation (Rip & Groen, 2001), and building a proprietary learning curve (Spence, 1981).
The regulatory uncertainty regarding a technology adoption decision is especially threatening to new technology-based firms (NTBFs; Autio, 1997;Little, 1977).NTBFs are new firms that are established based on a technologybased venture idea (Costa et al., 2018;Frederiks et al., 2019).When successful, NTBFs often grow rapidly (Autio, 1997;Grilli & Murtinu, 2018), they often internationalize (Cahen et al., 2016;Costa et al., in press), and they play an important role in economic development (Autio & Yli-Renko, 1998;Fontes & Coombs, 2001;Santos et al., 2017).As technology is at the core of such firms, regulatory uncertainty regarding the technology is an important aspect to NTBFs' activities.
Our aim in this study, therefore, is to shed light on the effect perceived regulatory uncertainty has on the decision by NTBFs to adopt a new technology (Jack et al., 2015;McGrath, 1997).Thereby we hope to gain insight on whether institutional theory or the resource-based view is a better theoretical lens for understanding this phenomenon.To understand perceived effects of regulatory uncertainty on the technology adoption decision by NTBFs, we will develop and test two competing hypotheses: one based on institutional theory and one based on the resource-based view.Then we will test these hypotheses using a sample of 108 fintech entrepreneurs' ventures' adoption of cryptocurrencies.As mentioned above, cryptocurrencies are facing potential regulation, which leads to regulatory uncertainty.The fintech industry is an important and growing group of NTBFs (Eckenrode & Friedman, 2017;Mnuchin & Phillips, 2018) and is one of the biggest adopters of blockchain technology (Treat & Brodersen, 2017).Therefore, the fintech industry is an important industry for cryptocurrency adoption (Bellavitis et al., in press), and it is facing the potential effects of regulatory uncertainty.Knowledge on these effects in the fintech industry could be informative to other NTBFs in industries in which blockchain adoption rates are currently not as high.
Our study makes three main theoretical contributions.First, we add to the entrepreneurship and small business literature by demonstrating the importance of regulatory uncertainty as a form of environmental uncertainty that affects NTBFs' technology adoption.Whereas most studies on the effect of regulatory uncertainty focus on existing corporations' strategic responses (for example, Bui & Villiers, 2017;Engau & Hoffmann, 2009, 2011a, 2011b;Engau et al., 2011;Fremeth & Richter, 2011;Rothenberg & Ettlie, 2011) and can lead to opportunities for them (Kolk & Mulder, 2011), a small business and entrepreneurship perspective on the effects regulatory uncertainty has on technology adoption is, to the best of our knowledge, lacking in the literature.Understanding the role that regulatory uncertainty has on the technology adoption decision of NTBFs is important.Not only because technologies are at the core of NTBFs, but also because compared to large, existing corporations, small, entrepreneurial firms often have to "put their eggs in one basket" due to their liabilities of newness (Stinchcombe, 1965) and smallness (Gimenez-Fernandez et al., 2020;Lefebvre, 2020), and may not have the resources to recover from a wrong decision on such an important strategic matter.
Second, we add to the literature that the role of regulatory uncertainty on technology adoption might be different in the context of large firms then it is in the context of small firms, such as NTBFs.Larger organizations facing regulatory uncertainty regarding the technology are more likely to postpone investment decisions until uncertainty, and therefore risk, has been reduced (Bittlingmayer, 2001;Marcus & Kaufman, 1986;Porter & van der Linde, 1995).However, the NTBFs in our sample who reported higher levels of regulatory uncertainty regarding the technology are more likely to adopt that technology.Our findings suggest that for NTBFs the trade-off between opportunity-seeking and risk-reduction under regulatory uncertainty is different to that of larger organizations.The role of regulatory uncertainty on technology adoption in small firms, therefore, may be opposite to what we see for large corporations.
Third, by developing and testing two competing hypotheses derived from two different theoretical lenses, we shed light on the boundary conditions of institutional theory.Our finding that NTBFs are more likely to adopt technology under regulatory uncertainty is consistent with the resource-based view, but inconsistent with institutional theory.Whereas institutional theory is suitable to explain technology adoption decisions under regulatory uncertainty of large companies, as existing studies show (for example, Bittlingmayer, 2001;Marcus & Kaufman, 1986;Porter & van der Linde, 1995), our findings imply a boundary condition for this theory as it does not explain this same phenomenon so well for small, technology-based entrepreneurial firms such as NTBFs.Under these conditions, the resource-based view, which more explicitly allows for entrepreneurial agency by the NTBFs, better explains technology adoption under regulatory uncertainty.

Theoretical framework
To better understand whether regulatory uncertainty affects the technology adoption decisions of NTBFs, we will first explore the broader literature on technology adoption and derive several of these known factors.Then we will look at the role of uncertainty in small firms in general and the role of regulatory uncertainty more specifically.Finally, we build upon the literature on institutional theory and the resource-based view and develop our two competing hypotheses regarding the effects regulatory uncertainty has on technology adoption decisions.

Technology adoption
To better understand the role of regulatory uncertainty on technology adoption decisions, we will first briefly describe technology adoption.For a technological innovation to be successful in the market and diffused through a population, it needs to be adopted by a large part of that population.Technology adoption has been studied from a wide variety of perspectives, including technology management and marketing, and in a wide variety of sectors and firms, including small businesses (Eiriz et al., 2019;Nguyen et al., 2015).Decades of research in these fields has found that the decision to adopt a new technology depends on various factors (for example, Neumeyer et al., 2018).These factors can often be grouped into three contexts that influence the decision to adopt a technology: the technological context, the organizational context, and the environmental context (DePietro et al., 1990).The technological context concerns the technology's own attributes that affect its adoption, such as its relative advantage, complexity, and compatibility (Hoti, 2015;Oliveira & Martins, 2011;Rogers, 1962Rogers, , 2003;;Thong, 1999;Tornatzky & Klein, 1982).The organizational context is concerned with processes and structures of the firm itself that constrain or facilitate the adoption and implementation of innovations (DePietro et al., 1990), such as technology competence, support from top management, and organizational readiness (Iacovou et al., 1995;Wang et al., 2010).The environmental context concerns the factors of the company's external environment that can influence technology adoption (DePietro et al., 1990), such as pressure from trading partners, competitors, and governments.It is within this last context that we explore the role of regulatory uncertainty.
In the technology adoption literature, the regulatory environment has been identified as a known factor which either facilitates (for example, Kuan & Chau, 2001;Li, 2008;Zhu & Kraemer, 2005) or hinders (for example, Gibbs & Kraemer, 2004;Hsu et al., 2006) technology adoption.In reality, however, laws and regulations tend to follow social change (Dror, 1958), and technological innovation is one of the major sources of social change (Barnett, 1953).Therefore, there are situations in which regulations governing a technology are simply nonexistent (such as face recognition technology, see for example, Garvie et al., 2016) or they have not yet matured (such as car sharing technology, see for example, Mitchell, 2015).In these situations, the regulatory environment itself is not a barrier or facilitator, but the perceived uncertainty about future regulations could be.In this study, we are therefore interested in the effect of regulatory uncertainty on technology adoption decisions.Regulatory uncertainty is a part of the environmental context and not as an additional context next to the technological, organizational, and environmental contexts (DePietro et al., 1990).

Regulatory uncertainty
In this study, we are interested in the effect of regulatory uncertainty.Regulatory uncertainty is "an individual's perceived inability to predict the future state of the regulatory environment" (Hoffmann et al., 2009(Hoffmann et al., , p. 1229)).It is a form of environmental uncertainty (Bylund & McCaffrey, 2017).Environmental uncertainty is a strong (Dequech, 1997), fundamental (Dequech, 2000), more substantive form of uncertainty (Dosi & Egidi, 1991).Environmental uncertainty is characterized by an open set of outcomes and a closed set of decisions (Packard et al., 2017).NTBFs facing regulatory uncertainty are faced with an open set of outcomes: the regulation regarding their technology may take shape along a continuum between a complete ban on the technology to fully unregulated use of the technology.Most likely, however, the form of regulation takes place somewhere between these two extremities.Regardless of how the outcome will be, these NTBFs are faced with a closed set of decisions: they can either adopt the technology, or not adopt the technology.
As the decision to adopt a new technology is made by the management of an organization, this decision is highly dependent on the management's perceptions (Finkelstein & Hambrick, 1996;Hambrick et al., 2005).Hence, the technology adoption decision also depends on the management's perception of regulatory uncertainty (Hoffmann et al., 2009).The literature, thus far, shows no clear effect of perceived regulatory uncertainty on investment decisions such as a technology adoption decision.The two main literature streams interested in the effect of regulatory uncertainty on investment decisions are institutional theory and the resource-based view (Hoffmann et al., 2009).Whereas institutional theory describes how firms behave due to institutional pressures external to the focal firm (Meyer & Rowan, 1977), the resource-based view explains firm behavior by looking at internal factors (Barney, 1991) and is therefore closely related to agency theory (Mahoney & Pandian, 1992).Although institutional theory and the resource-based view have different points of departure, they can explain the same phenomena under different circumstances.Therefore, multiple studies combine both theories (for example, Oliver, 1997) and, as we do here, develop competing hypotheses (for example , Eisenhardt, 1988;Zhang et al., 2018).In the next two sections we will theorize whether regulatory uncertainty leads to more or less technology adoption by formulating two competing hypotheses.Our overall conceptual model is shown in Figure 1.

The effect of regulatory uncertainty on technology adoption from an institutional theory perspective
Institutional theory is concerned with how organizations conform to rules and norms of the institutional environment to better secure their positions and legitimacy (Bruton et al., 2010;Meyer & Rowan, 1977;Scott, 2001).Institutions define what is appropriate and what is not, and therefore have a strong influence on organizations and their actions (DiMaggio & Powell, 1983).Regulations are important as instruments through which institutions affect companies (Greenwood & Hinings, 1996).Although institutional theory has often been researched in a setting of larger organizations (Bruton et al., 2010), institutional theory has shown to be an important theory to explain phenomena in small and entrepreneurial firms (for example, Davidsson et al., 2006;Fauzi & Sheng, in press;Hessels & Terjesen, 2010;Yousafzai et al., 2015).
The regulatory institutional environment puts pressure on companies' decision-making, including investment decisions such as adopting a technology (Greenwood & Hinings, 1996;Urbano et al., 2019).Regulation of a technology puts pressure on organizations to only work with that technology under the conditions allowed by the regulation.When organizations perceive regulatory uncertainty, the pressures put on the organization are ambivalent and organizations may not know under which conditions the technology may or may not be used in the future.Adopting a technology under such conditions is challenging and may lead to irreversible sunk costs (for example, Goldsby & Hanisch, 2022).Whereas larger firms with deeper pockets may be able to recover from such failed investments, for smaller firms such as NTBFs such a decision may be an "all-in" investment.Under such conditions, the option to wait-and-see how regulation may unfold is particularly valuable (Pindyck, 1991).Following institutional theory, we therefore expect NTBFs experiencing high regulatory uncertainty to report lower levels of technology adoption.
Prior research has shown empirical evidence supporting that expectation.Several studies have shown that regulatory uncertainty causes firms to postpone making investments and to adopt a wait-and-see attitude (Bittlingmayer, 2001;Roose, 1954), especially when they expect clarification in the short or middleterm (Hoffmann et al., 2009).Roose (1954), referring to the 1930s depression, found that regulatory uncertainty affected investment timing, and Bittlingmayer (2001) showed that regulatory uncertainty had caused a waitand-see attitude in investments between 1947 and 1991.Marcus and Kaufman (1986) found that regulatory uncertainty caused companies to hesitate to invest, and Porter and van der Linde (1995) found that greater certainty encourages investment, because greater certainty increases the chances that investment values will grow.This is illustrated by a recent example on the effect regulatory uncertainty has regarding the European Union's genome editing classification in crops and its effect on innovation in this field: "The lack of legal clarity regarding the regulation [. ..] undermines confidence in the technology, and therefore stifles investment and innovation" (Jones, 2015, p. 3).
Summarizing, the literature following an institutional theory perspective shows that regulatory uncertainty causes companies to adopt a wait-and-see attitude (Bittlingmayer, 2001) since such uncertainty increases investment risk (Porter & van der Linde, 1995) and irreversible sunk costs (Pindyck, 1991).
This decreases the probability that companies will decide to invest (Luo, 2004).Following a similar line of argument, a higher degree of perceived regulatory uncertainty regarding a technology could cause NTBFs to hold back or postpone making investments such as adopting new technologies.Hence, based on institutional theory, we hypothesize: Hypothesis 1: NTBFs experiencing higher levels of perceived regulatory uncertainty regarding a technology are less likely to adopt that technology.

The effect of regulatory uncertainty on technology adoption from a resource-based view
The literature following a resource-based view, in contrast, challenges the assumption that regulatory uncertainty causes firms to postpone investment decisions such as those toward technology adoption (Hoffmann et al., 2009).According to the resource-based view, to gain a sustainable competitive advantage, firms need resources that are valuable, rare, inimitable, and non-substitutable (Barney, 1991(Barney, , 2001;;Barney et al., 2001;Kraaijenbrink et al., 2010).Following the resource-based view, firms faced with uncertainty "take greater risks, and use more innovative strategies than [. ..] in less turbulent environments" (Aragón-Correa & Sharma, 2003, p. 77).They do so, because they try to "anticipate events" and "implement preventive actions" (Aragón-Correa & Sharma, 2003, p. 77) to minimize the vulnerability of the organization's resources (Hoffmann et al., 2009).
To ensure its long-term sustainable competitive advantage, firms take action to secure competitive resources (Barney, 1991).These actions include making investments in such resources (Hoffmann et al., 2009).Following the resource-based view perspective, regulatory uncertainty gives firms the opportunity to make investments in valuable, competitive resources that have not yet been regulated.Taking into account that it may become more difficult to access said resources after regulation has unfolded, firms may decide to take action now that the resources are still accessible.
In addition, investment decisions cannot be viewed from a regulatory uncertainty perspective alone.Firms operate with multiple resources and strategies and an investment decision is just one of many decisions they need to take.When adopting a technology may be leveraged as complementary resources to existing resources or strategies, firms may be more likely to make that decision, despite regulatory uncertainty (Hoffmann et al., 2009).They do so, in line with the resource-based view, because it makes their existing resources more valuable (Barney, 1991).
As technologies are at the core of NTBFs, it is likely that technology adoption decisions will complement the existing technological resources of the firm and that NTBFs can leverage complementary resources.Following the resource-based view, we therefore expect NTBFs experiencing high regulatory uncertainty to report higher levels of technology adoption.
These expectations are supported by empirical findings in the field (for example, Carrera et al., 2003;Rugman & Verbeke, 1998).Periods with increased uncertainty in the regulatory environment caused a wave of investments by firms attempting to expand their resource portfolio (Carrera et al., 2003).In this way, the firms could spread the risk of uncertainty by distributing risk over a larger portfolio.Thereby firms appeared to become less dependent on the effects of regulatory uncertainty (Carrera et al., 2003;Hoffmann et al., 2009).Under regulatory uncertainty, decision-makers feared irreversible investments in inflexible resources and tended to make reversible investments in flexible resources with a high probability of increasing performance (Rugman & Verbeke, 1998).These findings support the argument that investment decisions made under regulatory uncertainty are aimed at securing competitive resources (Hoffmann et al., 2009).
Following the abovementioned line of argument, a higher degree of perceived regulatory uncertainty regarding a technology might prompt NTBFs to make investments such as those in adopting that technology in order to secure competitive resources and leverage complementary resources.Hence, based on the resource-based view, we formulate the following competing hypothesis: Hypothesis 2: NTBFs experiencing higher levels of perceived regulatory uncertainty regarding a technology are more likely to adopt that technology.

Background information on the empirical setting
Cryptocurrencies are based on blockchain technology.The origins of blockchain are often traced back to its first application, cryptocurrency bitcoin (BTC), and its pseudonymous inventor Satoshi Nakamoto. Nakamoto (2008) used a cryptographic principle to develop a currency and transaction system called bitcoin, which eliminated the need for a trusted third party, such as a bank or financial services company, in the online transaction system.In essence, blockchain technology is a public ledger that records all transactions since the first transaction (the so-called "genesis").Every transaction that happens between two transactors is recorded in this ledger.The currency that is transferred comes from all previous transactions, which are all visible to every participant in the network.Therefore, every transaction can be traced back to the currency's origin, proving its authenticity without the need for third party validation.Updating and verifying this ledger is done by the cooperative effort of the whole network, which bundles participants' computer power to solve cryptographic hashes, for which they are rewarded with Bitcoins.This is the only way new bitcoins are created since genesis.The system is tamperproof because a malicious actor would require more computing power than the whole network combined to create a new longest version of the bitcoin blockchain, which would then be accepted as the new "true" bitcoin blockchain (Buterin, 2013).Based on the blockchain technology, many NTBFs launch their own tokens and coins (Bellavitis et al., in press).
The most important feature of blockchain technology is its elimination of trust in a third party (for example, a bank), because the information contained in the blocks on the blockchain represents the consensual truth.Soon after bitcoin was launched, the underlying blockchain technology's potential to eliminate trust in third parties was applied in more areas than just currencies.Despite its novelty, blockchain technology has already found a wide range of applications in financial and nonfinancial areas and it is likely to continue to expand its range and the quality of its applications (Crosby et al., 2016).Examples of industries where cryptocurrencies and token solutions are currently being developed are payments and settlements (Ripple, 2017), data storage and exchange (Wilkinson et al., 2016), software programming (Lisk, 2016), social media (Steem, 2017), personal identification (Civic, 2017), and medical data (Albeyatti, 2018).In conclusion, although still in the very early stages of development and adoption, blockchain technology has great potential and a wide range of possible applications.
Blockchain's core strength, its detachment from the existing financial or regulatory system, is also its biggest weakness.As many applications of blockchain offer the possibility for its users to remain anonymous via pseudonyms, banks, governments, and tax offices cannot properly track cryptocurrency transactions on the blockchain (Woolf, 2016).This opens up possibilities for unwanted usage, such as facilitating illegal trade on the black market, money laundering, or tax evasion (Foley et al., 2019).Many governments have announced intentions to regulate cryptocurrencies (Borrás & Edler, 2020), but they are still in the process of developing appropriate legislation (for example, Bellavitis et al., 2021;Federal Reserve, 2014;Fox, 2018), leaving many (potential) blockchain companies to face regulatory uncertainty, as they wonder whether their (future) company will still be in business after the introduction of the new, yet still unknown, regulations.This makes blockchain a suitable setting to study the effects of regulatory uncertainty on technology adoption.

Data collection
We collected the data for this research project in early 2017.Blockchain adoption was still in its early stages of diffusion and cryptocurrency exchanges were hardly regulated (Bellavitis et al., 2021).At the time, regulation had been planned in several countries, but was not implemented (Bellavitis et al., 2021).Hence, many fintech ventures faced regulatory uncertainty regarding the use of cryptocurrencies.Data collection therefore happened at a very timely moment, since as of the second half of 2017, many countries have implemented regulations for cryptocurrencies (Bellavitis et al., 2021).To the best of our knowledge, around that time in 2017 no universal databases of blockchainbased companies were available in the fintech sector, which made data collection difficult.
We used an online survey in fintech companies to collect the data.Fintech companies were chosen because the financial industry had one of the highest blockchain adoption rates (Treat & Brodersen, 2017) and could show best which factors contribute to the successful adoption of blockchain technology.Although the unit of analysis in our study is at the company level, one individual within each firm answered the survey on behalf of the company.The unit of observation is therefore at the individual level, which is not uncommon in small business research (for example, Ngoasong & Kimbu, 2019).As the NTBFs in our sample are small organizations, the individual respondents were very likely to be involved in the strategic decisions of the firms, and their perceptions strongly influence the organizations' perceptions (Finkelstein & Hambrick, 1996;Hambrick et al., 2005;Pryor et al., 2021).
We made use of Bureau Van Dijk's ORBIS data set to select and contact 40,000 fintech companies around the world.This resulted in 64 complete responses (0.2% response rate).Due to the low response rate, we tested for nonresponse bias by comparing differences between the first, second, third, and last quartile of respondents (Armstrong & Overton, 1977).Having found no differences (F(1,62) = 0.20, p = .66),we conclude that a nonresponse bias is unlikely.
To increase sample size, we searched for directories that provided overviews of fintech start-ups with publicly available e-mail addresses.We scanned hundreds of websites that list fintech companies from a variety of countries, and then in total sent out 933 surveys.We received 28 complete responses (3.0% response rate).Once again, we tested for nonresponse bias (Armstrong & Overton, 1977).As no significant differences were found (F(1,26) = 1.34, p = .26),nonresponse bias is unlikely.
In addition, we distributed a link to our survey to internet communities where fintech start-ups come together.This resulted in an additional 19 responses.Again, the nonresponse bias test (Armstrong & Overton, 1977) indicated that nonresponse bias is unlikely (F(1,17) = 0.93, p = .35).Our total sample size therefore is 111.
As the data was collected from different sources, we conducted one-way analysis of variance tests to determine differences between the three data sources on all variables in the data set.We found that the companies from the ORBIS data set had significantly less employees (p < .01)than the companies from the other two data sources, and they scored significantly lower on trading partner pressure than the companies in our self-built database (p < .001).The companies in our own built database scored significantly higher on organizational readiness compared to the other two data sources (p < .05).We found no significant differences between the three data sources on any of the other variables, including the dependent variable.Since we found only small differences on a few variables in the data set, we merged the data from the three data sources into one data set.As the data was collected through online surveys, we tested for careless responses (Meade & Craig, 2012).We found one instance of straightlining which we removed from the data set, resulting in a data set of N = 110.Table 1 gives an overview of the sample composition.

Measurements
We measured the main variable of interest, regulatory uncertainty, using a self-developed three-item scale (Cronbach's α = .91),inspired by Teo et al.'s (2006) concept of unresolved legal issues.We asked respondents to what extent they agreed or disagreed to the following statements: (1) "My company is concerned about potential future tax regulations on Blockchain activities;" (2) "My company is concerned about potential future regulatory restrictions on Blockchain activities;" and (3) "My company is concerned about potential future regulatory prohibition of Blockchain activities."All items were measured on a 5-point Likert scale ranging from "strongly disagree" to "strongly agree." The dependent variable in technology adoption studies is commonly measured as "adopted" vs "not adopted" (for example, Kuan & Chau, 2001;Zhu et al., 2003).However, as blockchain technology is not widely adopted yet, there could be a considerable number of companies interested in the technology and in the process of evaluating its adoption, without having adopted it yet.To be able to capture these different levels of adoption, we followed Oliveira et al. (2014) and asked the participants to answer the question "In what stage of blockchain technology adoption is your organization currently engaged?"Answering options were (1) not considering, (2) have evaluated but do not plan to adopt, (3) have evaluated and plan to adopt, (4) are currently adopting, and ( 5) have already adopted it.
We also included several control variables known to affect technology adoption decision.We modified items of these measures to fit the blockchain technology context.Relative advantage was measured using a five-item scale (Cronbach's α = .90)adapted from Oliveira et al. (2014).Measures for complexity (two items, α = .85),compatibility (four items, α = .75),technology competence (three items, α = .87),top management support (four items, α = .95),trading partner pressure (three items, α = .96),and competitive pressure (two items, α = .78)were adapted from Wang et al. (2010).Organizational readiness was measured using a two-item scale (Cronbach's α = .85)that we adapted from Gibbs and Kraemer (2004).All items can be found in Appendix A. As previous research indicated that firm size has an effect on technology adoption (Li, 2008;Pan & Jang, 2008;Wang et al., 2010;Zhu et al., 2003), we included number of employees (Zhu & Kraemer, 2005) and revenue during the last year (Wang et al., 2010) as additional control variables.Table 2 gives an overview of the mean values, standard deviations and correlations of the variables.

Instrument validation
To assess the construct validity of the measures, we conducted a principal component analysis (PCA) prior to testing the hypotheses.We calculated the overall Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy (Kaiser, 1970) to determine whether the PCA would yield distinct and reliable factors.The KMO score was 0.855, which is classified as "meritorious" (Kaiser, 1974) or "great" (Field, 2009;Hutcheson & Sofroniou, 1999), meaning a PCA would lead to reliable factors.Bartlett's test of sphericity (Bartlett, 1937;Field, 2009) was statistically significant (p < .0005),which further indicated that PCA is a viable validation technique.The KMO-measure for most items was above the threshold of 0.5, except for the two items measuring "complexity," which had a KMO score of 0.416 and 0.425, respectively (Kaiser, 1974).Accordingly, we  tested the assumptions again without the "complexity" items, resulting in an overall KMO score of 0.881, which is only a slight improvement.We therefore decided to keep the "complexity" items in our analysis.
As visual inspection of the scree plot did not show a clear inflection point (Cattell, 1966), we used the Kaiser's criterion (Kaiser, 1960) to determine what number of components to extract (Field, 2009).PCA using all 28 items revealed seven components that had eigenvalues greater than 1 (Field, 2009;Kaiser, 1960).These seven components explained 40.2%, 9.7%, 8.9%, 6.8%, 6.0%, 4.5%, and 3.8% of the total variance, respectively, and hence explained 79.9% of total variance.We employed a varimax orthogonal rotation to aid interpretability (Field, 2009), but it did not result in simple structure (Child, 2006;Thurstone, 1947).In the rotated component matrix, one "compatibility" item showed cross-loadings and another "compatibility" item had no loadings above 0.5 (Pituch & Stevens, 2016), therefore we removed these items.One "technology complexity" item loaded on the wrong component, while the other two items showed cross-loadings.Therefore, we completely removed the "technology complexity" items.Lastly, one of the two "competitive pressure" items showed cross-loadings, which prompted us to remove "competitive pressure" from our analysis.
Accordingly, these items were deleted and we performed the PCA again to extract seven components.Table 3 shows the resulting component loadings and their Cronbach's alpha coefficients.All variables' Cronbach's alpha values are above the threshold value of 0.7 (Nunnally, 1978).Simple structure was present, meaning that all items within a construct load on the same component, without cross-loadings (Thurstone, 1947).
As all independent variables and the dependent variables were measured in the same survey, there is a risk that common-method variance is present in the data (Lindell & Whitney, 2001;Podsakoff et al., 2003).Previous research has suggested both ex ante and ex post solutions to deal with common-method variance.We followed ex ante procedural recommendations by Podsakoff et al. (2003) and guaranteed anonymity to the respondents, ensured itemwordings were simple and concise, and measured independent and dependent variables using different response scales.
If common-method variance would be present in the data, then one latent dimension should account for the variance in the independent variables (Podsakoff et al., 2003).Several ex post statistical techniques have been developed to detect the presence of common method variance (Richardson et al., 2009).There are several trade-offs for each of these approaches (Podsakoff et al., 2003;Richardson et al., 2009).Generally, the CFA marker approach is recommended, but only if that a marker construct is included that is theoretically unrelated to both the independent and the dependent variables (Richardson et al., 2009;Williams et al., 2010).As we followed ex ante recommendations to keep the survey as short as possible prevent respondent fatigue, and increase response rate, no such ideal marker was included.We therefore used the Harman's single-factor test to determine whether the common-method variance was present in our data set (Podsakoff et al., 2003).This approach has recently been used in research in entrepreneurship (Bagheri et al., 2022), small business (Hernández-Linares et al., 2021), and technology adoption (Chatterjee et al., 2021).

Analyses
The size of our sample (N = 110) is too small to confirm or reject the hypotheses using linear regression.According to G*Power (Faul et al., 2009), the minimum number of required cases for this study would be 130.Therefore, due to the relatively low sample size compared to the number of items we measured, the data has the characteristics of high-dimensional data (Johnstone & Titterington, 2009).This led us to analyze our data following a hierarchical principal component regression 1 (Bro & Smilde, 2014).Principal component regressions are very well suited to analyze highdimensional data (Hubert & Verboven, 2003).Principal component regression is a form of regression analysis that is based on principal component analysis.Whereas with linear regression the dependent variable is regressed directly on the independent variables, with principal component regression the dependent variable is regressed on the principal components of the independent variables.An important disadvantage of the principal component regression is that when all components are used, the model explains all of the variance (Hadi & Ling, 1998).Therefore, often a subset of components is used in the regression (Jolliffe, 1982).Since we reached simple structure in the final PCA (see Table 3), we will use those seven components to perform our regression analyses to test the hypotheses (Liu et al., 2003).Although not used very often, principal component regression has been used before in studies on innovation management (Laursen & Foss, 2003), entrepreneurship and small business (Mahadea & Khumalo, 2020), and technology adoption (Kuntashula et al., 2015).Before conducting the principal component regression, we tested its assumptions.We found no significant autocorrelation (Durbin & Watson, 1950), no heteroscedasticity, no evidence of multicollinearity (Dormann et al., 2013), no studentized deleted residuals greater than ±3 standard deviations (Chaloner & Brant, 1988), and no values for Cook's distance above 1 (Stevens, 1984).There were two high-leverage outliers with leverage values greater than 0.2 (Fung, 1993) which were excluded accordingly.This resulted in a final sample size of 108.With all assumptions tested, we ran the principal component regression.

Results
Table 4 shows the results of the principal component regression.First, we regressed the control variables on the adoption stage.This model fitted well with the data (Model 1; R 2 = .60,F(8, 99) = 18.91, p < .001).In this model, relative advantage had a significant positive effect on technology adoption (B = 0.292, p < .01).Complexity had a nonsignificant effect on technology adoption (B = −0.01,p = .92).Compatibility had a marginal positive effect on technology adoption (B = 0.17, p < .10).Top management support (B = 0.85, In addition, we ran the multiple linear regression models using ordinary least squares (OLS).Using OLS regression, the predictors had effects in the same direction as using principal component regression, but due to the small sample size the effects were not significant.This supports our choice for principal component regression.p < .01),organizational readiness (B = 0.54, p < .01),and trading partner pressure (B = 0.45, p < .01)all had significant positive effects on technology adoption.
Second, we added regulatory uncertainty to our model.This model had a good fit with the data (Model 2: R 2 = .63,F(9, 98) = 18.87, p < .001),and adding regulatory uncertainty led to a significantly better fit with the data due to a statistically significant increase in variance explained (∆R 2 = .03,∆F(1, 98) = 7.95, p < .001).In this final model, no significant changes occurred to the already included predictors.Regulatory uncertainty was found to have a significant positive effect on technology adoption (B = .26,p < .01).These findings refute Hypothesis 1 and support Hypothesis 2.

Discussion
In our study, we shed light on the effect of regulatory uncertainty on NTBFs' technology adoption by testing two competing hypotheses, each based on a different theoretical lens.We find that NTBFs who perceive higher levels of regulatory uncertainty regarding the technology, are more willing to adopt this technology; therefore, we provide evidence that under the circumstances of the empirical setting, the resource-based view is a better theoretical lens to understand technology adoption by NTBFs under regulatory uncertainty than the institutional theory.
Our findings are contrary to some previous findings regarding the effect of regulatory uncertainty on technology adoption following an institutional theory approach (for example, Bui & Villiers, 2017;Engau & Hoffmann, 2009, 2011a, 2011b;Engau et al., 2011;Fremeth & Richter, 2011;Rothenberg & Ettlie, 2011).These studies find that corporations are less likely to adopt new technology when they perceive the regulatory uncertainty to be high.Their findings are in line with the logic of institutional theory, where companies wait until the uncertainty regarding regulations has been reduced before they take action that is line with the regulatory framework.
Although contrary to some previous findings, they seem to be in line with recent exploratory work on the role of regulation on new technology ventures (Amankwah-Amoah & Hinson, 2019;Giones et al., 2019).In their study on how companies deal with regulation for the technology in the drone industry, Giones et al. (2019) found that corporations and start-ups take quite different approaches.Corporate incumbents used their resources to lobby in favor of their interests.By forming associations of like-minded corporations, they lobbied for favorable regulation.New entrants, however, were found to have a more reactive approach to regulation (Giones et al., 2019).As these new entrants were so focused on the quality of their offering, they were often unaware of the disruptive impact of the technology, and only when potential usage restrictions were discussed, did they start to see the need for engaging with regulators.Their study seems to indicate that the corporations tried to reduce uncertainty first by lobbying for regulation (that would be favorable to them) before adopting the technology.This is in line with the institutional theory and the abovementioned previous findings on the effect of regulatory uncertainty on technology adoption by corporations.Moreover, their study seems to indicate that start-ups do not worry much about the regulatory uncertainty regarding the technology, as they see the opportunity to put a better offering into the market with the technology.These start-ups focus on gaining control over the valuable resources to do so, in line with the resource-based view.
Our results are also in line with exploratory findings on the contextual influences on the development of new technology ventures by Amankwah-Amoah and Hinson (2019).Based on interviews with technology entrepreneurs in Ghana, they find that technology entrepreneurs innovate from regulatory uncertainty as a strategy to overcome constraints such as the absence of regulations.Their finding is in line with the resource-based view, because by innovating these technology entrepreneurs take action and gain control over valuable resources.These findings seem to further illustrate why regulatory uncertainty is associated with an increase in technology adoption by NTBFs.

Theoretical implications
Our study has three theoretical implications.First, we add to the literature by highlighting the importance of regulatory uncertainty regarding the technology in technology adoption by NTBFs.Most studies on the role of regulatory uncertainty focus on the strategic responses of existing corporations (for example, Bui & Villiers, 2017;Engau & Hoffmann, 2009, 2011a, 2011b;Engau et al., 2011;Fremeth & Richter, 2011;Rothenberg & Ettlie, 2011) or study how regulatory uncertainty can lead to opportunities for these existing corporations (Kolk & Mulder, 2011).By showing that regulatory uncertainty affects small firms' technology adoption decision, we learn more about how regulatory uncertainty influences entrepreneurs in this decision.This is important because the rate at which NTBFs introduce new technologies in society is increasing (McGrath, 2013), leading to stronger "winds of change." Second, we add to the literature that the role of regulatory uncertainty on technology adoption might be different for larger organizations than for small firms, such as NTBFs.Previous studies have shown that larger organizations facing regulatory uncertainty regarding the technology are more likely to postpone investment decisions until uncertainty, and therefore risk, has been reduced (Bittlingmayer, 2001;Marcus & Kaufman, 1986;Porter & van der Linde, 1995).However, we find that the NTBFs in our sample who reported higher levels of regulatory uncertainty regarding the technology are more likely to adopt that technology.Our findings suggest that for NTBFs the trade-off between opportunity-seeking and risk-reduction under regulatory uncertainty is different to that of larger organizations.
Third, our study sheds light on the boundary conditions of institutional theory.Whereas institutional theory can explain large organizations' decisions regarding technology adoption under uncertainty, our study suggests it is less suitable to explain these same decisions by NTBFs and that the resource-based view is a more suitable theoretical lens under those conditions.This can be explained by following one of the main critiques on institutional theory, namely that there is little to no room for individual agency in institutional theory (DiMaggio, 1988).NTBFs' interests in potentially securing valuable resources outweigh the institutional pressures to postpone the risky investment decision.

Practical implications
Our research has several practical implications.First, for regulators and policy advisors, our study shows that there could be downsides to regulating new technologies.Whereas large firms tend to wait for regulatory uncertainty to subside before they start making investments, entrepreneurs tend to use regulatory uncertainty as a way of securing valuable resources, such as new technologies.Regulating new technologies early on, reduces NTBFs' chances of make these investments, which diminishes the role entrepreneurs play in bringing innovation and creative destruction to the market.This does not mean that new technologies should not be regulated at all, as there are important reasons for regulating new technologies, including cryptocurrencies (Borrás & Edler, 2020).However, when regulatory decisions are taken, the negative effects of reducing regulatory uncertainty should be taken into account.
Second, for NTBFs, our study suggests that regulatory uncertainty regarding a new technology provides opportunities to secure valuable resources by adopting this new technology.This could give these NTBFs a first-mover or fast-follower advantage (Lieberman & Montgomery, 1988, 1998).Especially in web-related technologies, such as cryptocurrencies, where winner-takes-all markets are becoming more and more common, these kinds of advantage have critical value.

Limitations of this study
This study has several limitations that could be interesting to address in future research.Here, we focus on two main limitations: the use of a cross-sectional design and using nonrandom samples.First, the limitations of cross-sectional designs are well-documented in the literature (for example, Bowen & Wiersema, 1999).Cross-sectional designs assume that parameters of the research model are stable across firms and over time.These assumptions very likely do not hold for our research.Nevertheless, we opted for a crosssectional design because we were interested in studying the effects of regulatory uncertainty on technology adoption and at the time of designing our study and collecting the data, several countries were thinking about providing legal clarity regarding the use of cryptocurrencies, which would reduce regulatory uncertainty severely in these countries at a later point in time.Measuring the dependent variable at a later point in time therefore seemed risky.Recent research has shown that many countries indeed implemented regulation from the second half of 2017 on (Bellavitis et al., 2021).Still, crosssectional research has severe limitations and the conclusions from this research should, for example, not be generalized to situations in which regulatory uncertainty has been reduced.Although our tests showed no empirical evidence that common-method variance affected the results of our study, it still remains a possibility.Reversed causality is another threat caused by the cross-sectional design.Although the theory of technology adoption clearly argues that the decision to adopt a technology depends on factors such as regulatory uncertainty, we cannot fully rule out the possibility that the current stage of adoption affected the respondents' levels of perceived regulatory uncertainty.
Second, the participants in this study were drawn from three nonrandom samples, and although we found only minor differences between the samples, combining the responses from three different populations might have affected the external validity of our results.Although we did not find statistical evidence for a nonresponse bias, the low response rates on our survey indicate that we need to be careful in extrapolating the results from this sample to a larger population.Therefore, we would like to comment on the generalizability of our findings (Simons et al., 2017).To the best of our knowledge, this is the first study where the relationship between regulatory uncertainty and technology adoption is tested with small firms.As small firms are such a heterogeneous group of companies (Wennberg et al., 2010), we cannot claim our results to hold for all small firms.At the same time, as NTBFs are found to be rather homogeneous (Chan et al., 2006), we are confident our findings will generalize to other (that is, non-fintech) NTBFs working on blockchain-based solutions.We also expect our findings to generalize to NTBFs working on other new, possibly groundbreaking technologies of which both its positive and negative implications are not yet fully understood.An example would be artificial intelligence technology, which has both tremendous possibilities (Heaven, 2020) and great threats (Hao, 2019), and for which the European Union has proposed, but not yet approved, regulation (European Commission, 2021), creating a situation of regulatory uncertainty.

Recommendations for future research
Despite the abovementioned limitations, our data gives important insights into the effects that regulatory uncertainty has on fintech ventures regarding the use of cryptocurrencies in a time of high regulatory uncertainty.Furthermore, based on our pioneering work, we have two methodological and three theoretical recommendations for future research.First, to overcome the limitation of the cross-sectional design in this study, we call for future research to study the effects of regulatory uncertainty on technology adoption using a longitudinal approach to capture changing levels of regulatory uncertainty and technology adoption over time and across settings.
Second, to overcome the limitation of the use of three nonrandom samples, we call for studies on the effects of regulatory uncertainty of technology adoption using random samples.As databases of fintech companies are currently more widespread than they were in 2017, future research could study similar relationships to the ones explored in this study using a more random sample.As by now regulatory uncertainty regarding cryptocurrencies has been reduced due to more governments having introduced regulations (Bellavitis et al., 2021), we recommend that future studies investigate NTBFs' adoption of other state-of-the-art technologies for which regulations will most likely be under discussion due to the large implications these technologies will have for society.A previously mentioned example of such a technology would be artificial intelligence (European Commission, 2021;Hao, 2019;Heaven, 2020).
To further advance theory, we first would recommend follow-up studies to investigate further how regulatory uncertainty affects the technology adoption decision.Interviewing NTBFs or observing decision-making moments in such companies would create a better understanding of how the effects of regulatory uncertainty on technology adoption are taken into account when a top management team is discussing whether to adopt it (Rothenberg & Ettlie, 2011).
Second, now that regulatory uncertainty regarding cryptocurrencies has been reduced, it would be interesting to study the changes in adoption rates.For example, it would be interesting to find out whether (large) firms have now become more likely to adopt cryptocurrencies, as institutional theory would suggest.
Third, our study suggests that for NTBFs agency effects are stronger than institutional pressures in the early stages of technology adoption.This raises the question whether teams with a flat hierarchy and one influential (technology) enthusiast exercising agency (Neumeyer & Santos, in press) are more likely to adopt the technology early on than teams with more institutionalized structures and hierarchies.

Conclusion
This study has investigated the effect regulatory uncertainty regarding a technology has on NTBFs' adoption of this technology.Institutional theory, on the one hand, suggests that firms would prefer to postpone investment decisions, such as adopting a technology, whereas the resource-based view suggests that firms could benefit from this uncertainty to gather valuable resources and gain a first-mover or fast-follower advantage.By studying the technology adoption decisions of fintech firms facing regulatory uncertainty regarding blockchain-based cryptocurrency, our study finds that these NTBFs are more likely to adopt technology under regulatory uncertainty.This finding suggests that the resource-based view provides a better theoretical lens to understand NTBFs' technology adoption under regulatory uncertainty.For practice this means that regulators need to take the positive effects of regulatory uncertainty on innovation into account when they decide on potential regulation, and that creating regulatory clarity does not necessarily benefit NTBFs.

Figure 1 .
Figure 1.Conceptual model of the competing hypotheses for the effect of regulatory uncertainty on technology adoption.

Table 3 .
Principal component analysis results.

Table 4 .
Hierarchical principal component regression analysis predicting adoption stage.