Adoption of blockchain digital asset management systems: An empirical study

MSc Information Studies

Track: Business Information Systems

Master Thesis

Adoption of blockchain digital

asset management systems: An

empirical study

by

Jeroen Meijaard

10611002

September 11, 2016

First Examiner UvA:

dr. D. Heinhuis

Second Examiner UvA:

prof. dr. T.M. van Engers

Preface

This thesis on the adoption of blockchain digital asset management systems, is written in order to fulfill the graduation requirements of the master Infor-mation studies: Business inforInfor-mation systems Program at the University of Amsterdam (UvA).

During a graduation internship at the risk department of Deloitte Accoun-tants and Advisory, I have performed the research needed to complete this study. While the exploratory nature of the research made it occasionally difficult, conducting an extensive investigation has allowed me to answer the question that was identified at the start.

I would like to thank my supervisor dr. Dick Heinhuis for his guidance and support during the process of writing my master thesis. Furthermore I would like to thank all my colleagues at Deloitte, especially Frank Cederhout, who supported me along the way. I also wish to thank all of the respondents of the survey, my family and in particular my girlfriend.

I hope you enjoy reading this thesis on probably one of the most disruptive innovations until now.

Jeroen Meijaard

Abstract

Goal - The goal of this research is to identify the success factors of consumer blockchain digital asset management systems by testing factors that are in-fluencing consumers to adopt.

Background - Since the development of Bitcoin in late 2008, the underlying technology blockchain has been seen as ”the next big thing”. While cryp-tocurrencies like Bitcoin are no longer seen as an application being adopted by the average consumer, blockchain digital asset management systems are. In order to check whether the predicted positive future of blockchain digital asset management systems is true, and to close the recognized gap of aca-demic literature on blockchain technology, this study explored and tested the factors influencing the adoption of these systems.

Methods - Based on the characteristics of blockchain digital asset manage-ment found in the literature review, a conceptual model was formed con-taining factors of the UTAUT2 and WOSP models. Distilled from systems theory, the predicting factors extendability, connectivity, flexibility, privacy, security and reliability are added to the existing UTAUT2 model and tested by using a combination of factor analysis and multiple regression.

Results - The analysis found that there is a lack of blockchain knowledge, since only 60% of the respondents have heard of blockchain and 20% have used an application incorporating it. The influence of the opportunity based predictor variables as extendability, connectivity and flexibility were found to be of a significant influence on the adoption of blockchain digital asset management systems, explaining 43% of the adoption intention.

Conclusion - To conclude, this study designed a conceptual model for mea-suring the adoption of blockchain digital asset management systems con-taining opportunity oriented variables which can predict the adoption in-tention of consumers. Finally, it provides a solid basis for further research on blockchain digital asset management systems for consumers and other (blockchain) socio-technical systems.

Contents

1 Introduction 1 1.1 Issue . . . 1 1.2 Practical relevance . . . 2 1.3 Scientific relevance . . . 3 1.4 Research Questions . . . 3 2 Theoretical framework 5 2.1 Approach . . . 5

2.2 Blockchain digital asset management characteristics . . . 6

2.2.1 Introduction blockchain digital asset management sys-tems . . . 6

2.2.2 Blockchain . . . 6

2.2.3 Digital assets . . . 9

2.2.4 Stakeholders . . . 11

2.3 Adoption and technology acceptance Models . . . 12

2.3.1 Overview of models . . . 12

2.3.2 Chosen adoption model . . . 15

2.3.3 Model extension and refinement . . . 19

3 Methodology 23 3.1 Research strategy . . . 23 3.1.1 Conceptual model . . . 23 3.2 Study design . . . 26 3.3 Studied object . . . 27 3.4 Pilot study . . . 28 3.5 Participants . . . 29 3.6 Data analysis . . . 30

4 Results 31

4.1 Descriptives . . . 31

4.2 Reliability . . . 32

4.3 Validity . . . 34

4.4 Significance relations . . . 36

5 Discussion and limitations 40 6 Conclusion 44 A Digital asset 52 A.1 Warranty example . . . 52

B Survey 53 C Results output 55 C.1 PP Plot . . . 55

C.2 Descriptives . . . 56

Chapter 1

Introduction

1.1

Issue

Recently there has been a lot of interest in blockchain developments by companies in and outside of the financial services industry. This is due to the mathematical exactitude of blockchain which can perform transac-tions (transfer value) without the usage of traditional channels (Bheemaiah, 2015). Hayes and Tasca (2016) stated that “Blockchain applications span-ning a number of sectors promise to change the way companies and people transact, send payments, sign contracts, transfer ownership of things, and much, much more.” (p. 1). Blockchain has also been called the equivalent of Web 2.0 as a transformative technology (Guadamuz and Marsden, 2015).

While most interest is going out to blockchain for organizational usage, it ”can potentially restructure the power relationship between consumers and intermediaries online” (Fairfield, 2014, p. 37). This is enabled by the shift of control from intermediaries to the consumer due to use of the Bitcoin blockchain protocol, which is distributed and therefore challenges the place of intermediaries (Fairfield, 2014). While companies invest heavily in explor-ing blockchain business models, the consumer adoption of currently available applications is slow. Possibly due to associated risks, but until now the fac-tors influencing consumer adoption of blockchain based systems are unknown (Godsiff, 2015).

a large community, in order to be resistant against criminal attacks like in the mini-Blockchain (Bruce, 2014). As well as giving support, as in the case of the cryptocurrencies of which most of them are built on the Blockchain paradigm, the government has to allow it by providing legal support (Godsiff, 2015).

Being the first blockchain application, Bitcoin and other cryptocurrencies are the first blockchain applications on the market. Their adoption rate is still low and are not preferred as payment methods by consumers, possibly due to the lack of awareness and understanding of the blockchain technology (Tsani-dis et al., 2015) and are therefore questionable to function as an alternative currency (O’Dwyer, 2015). In order for the network to function cost effective, secure and efficient, a certain critical mass of nodes is needed to adopt the blockchain (Grewal-Carr, 2016). However it is stated that blockchain ”has the potential to decentralize the way we store data and manage information” (Wright and De Filippi, 2015, p. 1). Recent developments have shown po-tential to combine it with the internet of things, based on the transactional privacy and the value of digitized assets on the blockchain network by asset tracking, transferring etc. (Christidis and Devetsikiotis, 2016). Digital asset management is predicted to be one of the most promising applications of blockchain (Bitfury Group, 2016). Therefore this study tries to discover the factors influencing the adoption of consumer blockchain asset management software.

1.2

Practical relevance

Mentioned before is the enormous potential of products and services under-lying blockchain protocols. Besides the large potential benefits from product innovations, due to the disruptive nature of these products, these potential consumer blockchain products and services can be seen as really-new prod-ucts which accompany risks of failure for the company (Chao et al., 2013). Therefore this research can contribute to explore the consumer factors that determine new product adoption in order to mitigate the risks of failure for companies, by researching consumer blockchain products and services, which are as far as we know unknown at this point in time.

Due to the imbalance of large investments in future developments in blockchain of companies on the one hand and the lack of adoption of current consumer applications on the other, it is vital for application builders and investors to known which are the factors for the users and how the adoption of blockchain applications can be increased.

1.3

Scientific relevance

Research on individual-level technology adoption is one of the most mature streams in the information system research. The current literature is most focused on minor tweaking of theoretical replication. Venkatesh et al. (2007) therefore call for a focus on intervention, contingencies and alternative the-oretical perspectives which do not focus solely on social psychology. First of all, this adoption research studies the applicability of adoption theories into a new field, namely consumer blockchain products adoption. As already iden-tified by Bheemaiah (2015), although the financial and governmental bodies are acknowledging the potential of blockchain, research at business schools are lacking behind. Furthermore, the lack of research on the user perspective of blockchain technologies in scientific research can be identified as a gap in current literature (Baur et al., 2015). Also it is concluded that only experts are interviewed, which are not able to represent the average consumer. As well as the need for more research on the possible transaction fees, security of funds, and scalability of the protocol. Finally, in a study on adoption of blockchain based cryptocurrencies,Spenkelink (2014) concluded that ”future research is to test the model with a larger sample size on a more quantitative basis.” (p. 66).

1.4

Research Questions

The goal of this research is to identify the success factors of (future) consumer blockchain digital asset management systems by testing factors that are in-fluencing consumers to adopt. The results of this study can contribute to the understanding of firms and entrepreneurs who want to deploy blockchain applications or to the current knowledge level of academic literature, govern-ment bodies, and consumers.

The main research question is: “Which factors influence consumer adoption of blockchain digital asset management systems?”.

This reseach question will be answered by the following sub-questions: 1. What are the characteristics of blockchain digital asset management

systems?

2. Which adoption model can measure the adoption of blockchain digital asset management systems?

3. Which blockhain digital asset management specific factors can be added to the model?

4. To what extent does the conceptual model explain the consumer adop-tion of blockchain digital asset management systems?

This thesis will be structured as follows. In chapter 2 (Theoretical frame-work), the characteristics of blockchain digital asset management systems are explored. Second, the adoption theories that have been used to measure adoption in related fields will be discussed. Third, based on the characteris-tics of digital asset management systems, specific predictive factors are added in order to form a conceptual model. In chapter 3 (Methodology) The con-ceptual model will be presented and the chosen methods will be outlined. In chapter 4 (Results) the conceptual model will be tested. Finally, in chapter 5 (Discussion and limitations) and chapter 6 (Conclusion) the findings will be discussed and their implications will be concluded.

Chapter 2

Theoretical framework

This chapter will first introduce the approach used for searching existing related literature. In order to answer subquestion 1 ”What are the charac-teristics of blockchain digital asset management systems?”, the blockchain digital asset management system characteristics are outlined. Subquestion 2, ”Which adoption model can measure the adoption of blockchain digital as-set management systems?”, will be answered by exploring different adoption frameworks. Finally, specific blockchain digital asset management system predictor variables will be discussed to provide a sufficient result for sub-question 3 ”Which blockchain specific factors can be added to the model?”.

2.1

Approach

In order to ensure the quality of the research a critical literature review has been conducted. The parameters of a literature study by Bell (2005) are used for defining the literature search (Mark et al., 2009). The language of publication is English and the subject area is focused on technology adoption of blockchain digital asset management systems by consumers. Due to the exploratory and distributed nature of blockchain digital asset management systems, there was no specific geographical or publication period selected. The literature review is constructed from primary and secondary sources. These sources consisted out of high quality academic journals in the informa-tion systems field, conceived from the widely used Businesssource premier, sciencedirect and and IEEE database. When this search did not concede enough literature, google scholar, theses, and reports are self-assessed on

quality before being used.

Below are the keywords and selection criteria outlined, which are used for extracting the needed literature. For blockchain digital asset management systems the following keywords were used: ”blockchain”, ”digital asset”, ”asset management system” and ”colored coin”. For literature on technol-ogy acceptance models, the following keywords were used: ”Adoption the-ory”, ”Adoption Research”, ”technology”, ”Technology adoption model” and ”Technology acceptance model”. For literature on systems theory, the follow-ing keywords were used: ”Systems theory”, ”Systems Research” and ”System performance”.

2.2

Blockchain digital asset management

char-acteristics

2.2.1

Introduction blockchain digital asset management

systems

As stated before, blockchain digital asset management has been named as one of the most promising applications of blockchain technology (Bitfury Group, 2016). The support of blockchain enables self-sufficient assets, which could be seen as a transformative technology. This provides for disintermediation between digital asset consumers, asset issuers and developers. Furthermore it allows for decoupling all tasks related to asset management (establish-ing identities, secur(establish-ing funds, transaction process(establish-ing, etc.) (Bitfury Group, 2016). Therefore the following paragraphs will first outline blockchain. Af-terwards it’s possibilities for digital assets will be discussed.

2.2.2

Blockchain

Definition

Originating from the paper of Nakamoto written in 2008, the blockchain structure is a chronological database of transactions and the first to solve the byzantine generals problem (achieving consensus in a network with-out requiring identities or trust relationships) (Peters and Panayi, 2015).

Swan, author of the book Blockchain, defines blockchain technology as fol-lows: ”Blockchain technology is the secure decentralized computing ledger that underlies Bitcoin and cryptocurrencies, and more profoundly is a next-generation global-scale decentralized infrastructure and mechanism for se-curely updating distributed computing nodes with ongoing consensus truth states.” (Swan, 2015, p. 42). Grewal-Carr (2016) see blockchain as ”a tech-nology that allows people who don’t know each other to trust a shared record of events” (p. 2). Another paper define blockchain as ”a distributed peer-to-peer network where non-trusting members can interact with each other without a trust intermediary, in a verifiable manner.” (Christidis and Devet-sikiotis, 2016, p. 2292).

Furthermore it is regarded as a ”distributed ledger”, which is seen as ”an-other type of database for recording transactions” in which data is stored in fixed structured blocks (Grewal-Carr, 2016, p. 5). Wright and De Filippi (2015) state that ”it is an encrypted and decentralized database” (p. 12). Furthermore ”Blockchain technology has the potential to reduce the role of one of the most important economic and regulatory actors in our society—the middleman” (Wright and De Filippi, 2015, p. 2). This is done by giving par-ticipants in the blockchain the ability to transfer digital data or property in a secure, immutable and safe way, while also being pseudo-anonymous (Wright and De Filippi, 2015).

Characteristics

McConaghy et al. (2016) state that blockchain can be characterized by im-mutability, decentralized control, and creation and transfer of assets. Grewal-Carr (2016) described the following commonalities between the types of blockchains:

1. ”A blockchain is digitally distributed across a number of computers in almost real-time” (Grewal-Carr, 2016, p. 7).

2. ”A blockchain uses many participants in the network to reach consen-sus” (Grewal-Carr, 2016, p. 7).

3. ”A blockchain uses cryptography and digital signatures to prove iden-tity” (Grewal-Carr, 2016, p. 7).

4. ”A blockchain has mechanisms to make it hard (but not impossible) to change historical records” (Grewal-Carr, 2016, p. 7).

5. ”A blockchain is time-stamped” (Grewal-Carr, 2016, p. 7). 6. ”A blockchain is programmable” (Grewal-Carr, 2016, p. 7).

According to Grewal-Carr (2016) this leaves us with the following charac-teristics for all blockchains: Decentralized processing network, distributed ledger, digital signatures, programmable logic, and private vs. public.

Differences in blockchains

When looking at possible differences between multiple blockchains, a dis-tinction can be made between private vs public blockchains. In private blockchains (which are permissioned) all participants are known on fore-hand in the network and are allowed to update the ledger. In case of public blockchains, anyone, who has the software, can read or write data to and from the ledger. This is also known as permission-less blockchains (Grewal-Carr, 2016). The public blockchains are based on a consensus mechanism to validate transactions and permissions. Private blockchain transactions and permissions are monitored by a centralized manner of decision-making. There is thereby a difference in decentralization and restrictions between the two distinctions. An option for future applications is to be the partially decentralized, so called consortium blockchains, which are a hybrid form be-tween public and private blockchains (Pilkington, 2016).

As mentioned before blockchain allows for trust between parties. While trans-actions can only be done by someone who owns the private key, considering that other blockchain participants see them as valid in combination with a valid signature of the adress’ public key, the major obstacle is to reach con-sensus in the whole peer-to-peer blockchain network. This concon-sensus on the current state of all the earlier blocks in the ledger is provided by the complex mechanisms that arbitrate dispute and protect data integrity of the trans-actions in a particular block (Grewal-Carr, 2016). Thereby also eliminating the double-spend problem (Pilkington, 2016).

The consensus reaching mechanisms can vary per blockchain as well as whether they are private, public or a hybrid. The public Bitcoin blockchain is the

most well-known example of blockchains. Bitcoin’s proof-of-work mechanism is a complex mathematical problem which participants on the network are trying to solve in order to validate transactions on the ledger in exchange for Bitcoins. These participants are called ”miners”. Kraft (2016) state that proof-of-work is ”A particular cryptographic hash involving the block’s con-tent is formed, and must be below a threshold value.” (p. 2). Essentially it is performing a brute-force search for a hash collision. When a block is vali-dated, a new block is generated and added to the blockchain (Kraft, 2016). Proof-of-stake is an alternative which is used by altcoins. Rather than split-ting blocks across relative mining power of miners as with proof-of-work, it distributes stake blocks according to the wealth of miners. This leads to a decreased chance of a fifty-one percent attack and possible faster blockchains (Pilkington, 2016).

2.2.3

Digital assets

As seen in 2.2.2, the characteristics of blockchain provides a solid basis for digital asset management systems, by for instance being a decentralized en-crypted database (Wright and De Filippi, 2015). This is due to the fact that trustless public ledgers allow parties to hold and transfer value on their own terms (Fairfield, 2014). Therefore any digital asset could be stored and trans-ferred in a blockchain (Swan, 2015). The advantages of a blockchain based asset system instead of a traditional asset system, due to the decentralized nature, are system related factors such as the resistance against pirate at-tacks and rights management (Kishigami et al., 2015).

Showing the potential for digital assets, is the latest increase of incidents of security breaches compromising the privacy of users, the model in which third parties gather huge amounts of personal data is being questioned. Blockchain allows users control over their own data via a trusted and secure management system of personal data, using a decentralized network without interference of a middle man (Zyskind et al., 2015). For sustaining a blockchain based ledger for digital assets auditability, counterfeit resistance and user security are the general requirements (Bitfury Group, 2016).

Bitfury Group (2016) defines a digital asset as ”a floating claim of a certain service or goods ensured by the asset issuer, which is not linked to a particu-lar account, and is governed using computer tech- nologies and the Internet,

including asset issuance, claim of ownership, and transfer.” (p. 2).

Digital assets could be deployed in various ways, namely on the same or sep-arate blockchain, via colored coin protocols, metacoins, and on multi-asset blockchains (Bitfury Group, 2016). Each variation comes with her own costs and benefits.

The colored coin protocol are a new concept which uses the current Bitcoin blockchain and allows for storage of digital assets in the blockchain (Rosen-feld, 2012). On the basis of the Open Assets Protocol, colored coins allow for the possibility to add assets in the blockchain by interpreting the added meta-data in transactions as assets(Hillbom and Tillstr¨om, 2016). However colored coins do not support digital assets native on the blockchain and therefore lack efficient payment verification. Metacoins are colored coins which use a mid-dleware layer (dedicated servers) to verify colored coin transactions. These are overlay protocols using another blockchain as a basis. Alternatively na-tive supported digital asset based blockchains are sidechains and Multi-asset blockchains. Multi-asset blockchain has more space-efficient proof of owner-ship but lacks in security due to merged mining and blockchain anchoring (Bitfury Group, 2016). Nasdaq announced on May 11th 2015 a plan to create a blockchain based expansion of it’s Nasdaq Private market platform based on colored coin (Bystr¨om et al., 2016). Hawk is an example of a private blockchain for storing financial transactions via private contracts (Kosba et al., 2015). An example of a digital asset on public and permissionless blockchain via the colored coin and blockchain network is the E-warranty JSON object of Warranteer in Appendix A.1.

Smart contracts can add to the power of digital assets by allowing for the implementation of a contractual agreement or self-executing transactions and perform them automatically without interference of human operators. Al-though out of scope of this research, smart contracts provide the possibility to connect it with physical assets via Internet of Things will allow those as-sets to function as digital asas-sets in the blockchain (Wright and De Filippi, 2015).

2.2.4

Stakeholders

Digital assets on blockchains could be used for example in the context of financial assets, decentralized exchanges, smart property, electronic money, business-to-consumer assets, event tickets, digital subscription, peer-to-peer economy, or for a digital democracy (Bitfury Group, 2016). Within all dig-ital asset management systems the a range of different stakeholders can be defined: application developers, blockchain notaries, asset issuers, regulators and end users (Bitfury Group, 2016).

Combining the possibility to leave out a central authority for organization, the digital asset and various stakeholders, a fundamental shift in organiza-tion is going on, while transiorganiza-tioning from centralizaorganiza-tion to decentralizaorganiza-tion (Wright and De Filippi, 2015). Furthermore multiple self-enforcing smart contracts, mentioned in 2.2.3, can be bound together to form the rules and procedures needed for functioning as a decentralized autonomous organiza-tion. Therefore this leaves decision making to the users instead of letting it up to a select group of people at the executive level, by asking for multiple votes. This will result in bypassing inefficient large hierarchical organizations (Wright and De Filippi, 2015).

Socio-technical systems are defined as ”systems that involve both com- plex physical–technical systems and networks of interdependent actors.” (De Bruijn and Herder, 2009, p. 1). This complexity is shown in the levels of re-quirements that a sociotechnical system has to incorporated, namely hard-ware, softhard-ware, cognitive or social system (Whitworth, 2009). Due to the blockchain digital asset managements impact on society, economic and tech-nology, it could be seen as a socio-technoligical system. The internet, of which blockchain has been compared with, is also regarded as such (Fuchs, 2005).

Conclusion

Based on the section above, subquestion 1 ”What are the characteristics of blockchain digital asset management systems?” can be answered. It is found that blockchain digital asset management systems are characterised by being a decentralized database for digital assets, which has technical advantages above other systems and can be defined as a socio-techinical system.

2.3

Adoption and technology acceptance

Mod-els

2.3.1

Overview of models

In order to provide a basis for creating an adoption model, which includes the factors that are related to blockchain digital asset management adoption, the adoption literature is explored in order to find a suitable model. Based on the nature of blockchain digital asset management systems found in section 2.2, the search is specified to adoption models for information systems.

IDT

Rogers introduced the innovation diffusion theory (IDT), which describes how technology is adopted as a life cycle (Rogers Everett, 1995). Further-more Rogers (1995) describes five categories of influencing factors of adoption decision making, namely; relative advantage, compatibility, complexity, tri-alability and observability (Rezvani et al., 2015). Critique is given on the model due to the imprecise definitions and effects included in the model, while the justification of the five adopter categories are also disputed (Bayer and Melone, 1989).

TAM

The most referred model in technology adoption research is the technology acceptance model (TAM), due to the robustness of the model, scales, and the relationships that were tested (Venkatesh et al., 2007). The model iden-tifies perceived ease of use and perceived usefulness as predictors of behavior (Cowan and Daim, 2011). The model is later extended to TAM2 with social influence processes (subjective norm, voluntariness, and image) and cognitive instrumental process (job relevance, output quality, result demonstrability, and perceived ease of use), which can be seen in figure 2.1. The model was tested on longitudinal studies and explained a large variance of usefulness perception and usage intentions (Venkatesh and Davis, 2000).

Figure 2.1: TAM2 (Venkatesh and Davis, 2000)

In 2008 Venkatesh et al. proposed the TAM3 model, which is based on ear-lier models and focuses on organizational technology adoption and the role of interventions (Venkatesh and Bala, 2008). Due to the focus of this study on consumer adoption, this model is not considered as a option.

UTAUT

Builded upon previous adoption frameworks like TAM, Venkatesh presented the Unified Theory of Acceptance and Use of Technology (UTAUT), for mea-suring consumer acceptance and use of technology (Venkatesh et al., 2003). UTAUT consist out of four constructs, namely; performance expectancy, ef-fort expectancy, social influence, and facilitating conditions which have an effect on behavioral intention to technology usage (Venkatesh et al., 2012). UTAUT is extend to UTAUT2 based on the constructs of hedonic motivation, price value and habit, which are focussed on consumer usage, as can be seen in figure 2.2. This extention has led to an improvement in predicting technology use of 40% to 52% and from 56% to 74% for behavioral intention (Venkatesh et al., 2012).

Figure 2.2: Unified theory of acceptance and use of technology 2 (Venkatesh et al., 2012)

CAT

A model which merges the roles of cognition and affect into one model is the Consumer Acceptance of Technology model (CAT), which is based on the technology acceptance model and the pleasure, arousal and dominance paradigm of affect, with a resulting explained variance of 50% (Kulviwat et al., 2007).

ECM

A model that lays more focus on the continuity of usage of information sys-tem adoption is the Expectation Confirmation Model (ECM), which predicts that the confirmation, satisfaction and perceived usefulness will influence the intention of continued use of the information system (Liao et al., 2009). A key difference in constructs with the leading adoption model TAM is that ECM uses satisfaction instead of user’s attitude, which explains the difference be-tween an overall evaluation of the product (attitude) and post-consumption evaluation (satisfaction). In the study of Liao et al. (2009), it is found that ECM performs even well on predicting initial adopters. While for Initial adaptors expectations are important determiners, short-term users intention

of usage will be mainly influenced by their satisfaction with the product or service based on the predictor confirmation of expectations (Liao et al., 2009).

TCT

Lastly, the Technology Continuance Theory (TCT) integrates the TAM, ECM and Cognitive model (COG), and predicts continuance intention as the dependent variable (Liao et al., 2009). While the Cognitive model is as well effective in forecasting intention of usage, it is missing some of the key variables of the other models and is therefore not explored individually. The attitude and satisfaction are both included in the model based on the cog-nitive model assumption that satisfaction could impact the overall attitude. By also including perceived usefulness and ease of use, the TCT model is capable of predicting long-term as well as initial adoption, which can be seen in figure 2.3. While it could be argued to include perceived usefulness, since it is only significant at initial adoption, it is needed for initial adoption and much tested in the TAM model (Liao et al., 2009).

Figure 2.3: Technology continuance theory (Liao et al., 2009)

2.3.2

Chosen adoption model

When looking for an adoption model that can predict adoption for blockchain digital asset management systems, there is a wide variety of models applica-ble. In order to determine which model suits the blockchain system, adoption models applied in related fields are discussed. In this section, the adoption model which is going to be used is outlined.

Adoption models in related fields

Even though there is a lack of adoption research on blockchain, especially quantitative, it can be argued on the basis of the characteristics mentioned in 2.2 of the blockchain digital asset management systems, that other informa-tion systems can be related to the studied object. To narrow it down further, information systems that are transactional based, rely on a complex technical infrastructure and function in a trust-based environment are discussed. TAM has been cited more than other models due to the fine-tuned aspects of the model through the times the model has been applied in research (Venkatesh et al., 2007). Trust as a factor has been added to the model when researching the adoption of Internet banking, since it is reasoned that it has a larger effect on customer acceptance than perceived ease of use (Hoehle et al., 2012). Recent research combined the TAM and TPB model with the added constructs of security, privacy, self efficacy, government support and technology support in order to predict Internet banking adoption in Tunisia (Nasri and Charfeddine, 2012).

In the field of pocket personal computer and hedonic information systems, the CAT model is tested. It is found that there is a non-significant influence of attitude on intention (Wang and Scheepers, 2012). There has been a lack of studies validating the model.

In comparison research on long-term adoption based on e-learning systems, it is found that the TCT model has more explanatory power than TAM and ECM. TAM also had less variance than the tested ECM. According to the researchers, the TCT model therefore made a substantial improvement over the TAM and ECM models, by synthesizing adoption constructs and adoption stages (initial, short-term, long-term). It is argued that it could be used both qualitatively and quantitative due to the well-established con-structs involved. Although largely investigated in prior literature, it could be argued to leave out the relation between perceived usefulness and intention, and perceived usefulness and satisfaction, since this connection only holds for initial adopters. Removing these effects did not reduce the overall explained variance by much (Liao et al., 2009).

One of the trust based environment of information system related fields in which UTAUT is applied, is the Internet banking adoption. A recent study

showed the statistical power as perceived risk as a determinant predicts adop-tion (Martins et al., 2014). Since the UTAUT2 model is evolved out of all the heavily cited and applicable TAM models, the model is seen as the best fit to measure the blockchain digital asset management adoption (Venkatesh et al., 2007).

UTAUT2

Because UTAUT2 is a theory that is specifically constructed for predicting consumer adoption, it aligns with the target group of this study. Based on one of the most replicated and tested adoption models (TAM), it can be argued the UTAUT2 model is a solid basis for further refinement of the adoption model for blockchain digital asset management systems. Table 2.1 describes the UTAUT2 variables, which will be included in the conceptual model.

Table 2.1: UTAUT2 Constructs (Venkatesh et al., 2012)

Construct Definition Root

Performance Expectancy

The degree to which using a tech-nology will provide benefits to con-sumers in performing certain activi-ties.

(Venkatesh et al., 2012)

Effort Expectancy The degree of ease associated with consumers’ use of technology .

(Venkatesh et al., 2012)

Social Influence

The extent to which consumers per-ceive that important others (e.g., family and friends) believe they should use a particular technology.

(Venkatesh et al., 2012)

Facilitating Conditions

Consumers’ perceptions of the re-sources and support available to per-form a behavior.

(Venkatesh et al., 2012)

Hedonic Motivation The fun or pleasure derived from us-ing a technology.

(Venkatesh et al., 2012)

Price Value

Consumers’ cognitive tradeoff be-tween the perceived benefits of the applications and the monetary cost for using them

(Venkatesh et al., 2012)

Habit

The extent to which people tend to perform behaviors automatically be-cause of learning

(Venkatesh et al., 2012)

Experience

Reflects an opportunity to use a tar-get technology and is typically opera-tionalized as the passage of time from the initial use of a technology by an individual.

(Venkatesh et al., 2012)

Conclusion

Based on this section, the subquestion 2 ”Which adoption model can mea-sure the adoption of blockchain digital asset management systems?” can be answered. By exploring different adoption frameworks, it is found that

UTAUT2 is the most suitable basis of the model, since it is focused on con-sumers, includes all of the latest research and is tested in related fields.

2.3.3

Model extension and refinement

To specify the UTAUT2 model that was taken as a basis for further develop-ment of a specific adoption model for blockchain digital asset managedevelop-ment systems, a more system and technical approach is taken to explore new con-structs. Based on the characteristics of blockchain digital asset management systems found in 2.2, system performance and design models from systems theory are explored in this section, in order to add variables to the adoption model. While models like the Total Performance System model (Meacham et al., 2003) and waterfall method (Khalifa and Verner, 2000) are found using the approach mentioned in 2.1, the models are mainly guidelines in order to reach a predefined system performance alongside predefined user requirements. The only model found applicable to blockchain digital asset management was the Web of system performance model and was therefore further explored.

Web of system performance

Starting from a point of view that previous models provide a limited view on system performance, the Web Of System Performance model (WOSP) was created by Withworth et al. (2003) as an extension of the most used Technology Acceptance Model (TAM). The theoretical framework is used for balanced design and evaluation of advanced information systems by measur-ing system performance and can be applied to each level of the system, e.g. hardware, software, cognitive or social system level (Whitworth and Zaic, 2003). These levels are dependent on the level of the system, so only a socio-technical system has a system level as well as all the other levels (Whitworth, 2009).

In previous work, the WOSP model has been tested in regards to socio-technical systems (browsers) although not validated significantly on a large scale (Whitworth et al., 2008). Since blockchain digital asset management systems can be seen as an advanced information system based on their char-acteristics as a distributed based database with a trustless proof mechanism seen in section 2.2, the model could add to the UTAUT2 model explaining

the intention to adopt blockchain digital asset management systems. System performance in the WOSP model is defined as ”How successfully it inter-acts with its environment, especially that it continues to interact with the environment.” (Whitworth and Zaic, 2003, p. 259). The interaction of the system can be seen as the intention of use of the system by users (Whitworth and Zaic, 2003).

The WOSP model analyses system performance via four system elements, namely: effector, boundary, structure and receptor purposes. Each element which can be steered towards opportunity creation or risk reduction and has an inter-item tension between them. Each of the elements contain two vari-ables, which is shown in figure 2.4 (Whitworth and Zaic, 2003).

Figure 2.4: WOSP model

The effector purposes are functionality and usability, which are also included in the original TAM model. These effectors act upon the environment of the information system. In the WOSP model, functionality is defined as ”To act directly on the environment to produce a desired change” (Whitworth et al., 2008, p. 781), while the given definition for usability is ”To minimize the relative resource costs of action” (Whitworth et al., 2008, p. 781). These factors are already included in the chosen UTAUT2 model and are therefore not added to the model. When looking at the boundary purposes, secu-rity and extendability can be distinguished. The boundary is the separation of the system from the system’s surroundings. The definition given in the WOSP model for security is ”To protect against unauthorized entry, misuse

or takeover” (Whitworth et al., 2008, p. 781). Extendability is ”To use out-side elements as part of the system” (Whitworth et al., 2008, p. 781).

The structure purposes are focused on the support and coordination of sys-tem activity within the syssys-tem itself. It consist of reliability and flexibility variables. Conform the WOSP model, reliability is ”To continue operating despite internal failure” (Whitworth et al., 2008, p. 781). The flexibility vari-ables is seen as ”To adapt the system’s operation to environment changes” (Whitworth et al., 2008, p. 781). Instead of acting on the environment like ef-fector purposes, the receptor purposes gather environment information in or-der to analyze it. It consists of connectivity, which is defined as ”To open and use channels to communicate with other systems” (Whitworth et al., 2008, p. 781) and privacy ”To manage the release of self information” (Whitworth et al., 2008, p. 781). All these variables influence the system performance, which in this study will be defined as the intention to use (Whitworth and Zaic, 2003). A distinction can be made between active or opportunity cre-ation variables (flexibility, functionality, extendability and connectivity) and passive or risk reduction variables (usability, security, reliability and privacy).

While the WOSP model is derived from systems theory, which aligns with the characteristic of a distributed database of the blockchain digital asset management systems with a trustless proof mechanism, it also aligns with user requirements of the system. As already mentioned in section 2.2.4, there are various stakeholders, which all can be seen as consumers depending on their role in the autonomous organization, do to the innovative nature of the blockchain digital asset management system. For application develop-ers the ease of application development, entry permissions and the reach are defined as important concerns in the paper by Bitfury Group (Bitfury Group, 2016). These requirements can be captured in the constructs of secu-rity, flexibility and extendability (Whitworth et al., 2008). When looking at blockchain notaries, the cost of operation, the rules regarding transaction, as well as the entry permissions can be of influence. These requirements can be found in constructs as security, usability and connectivity (Whitworth et al., 2008). The asset issuers require a counterfeit resistance system with regards to their openness to third party applications, while also keeping in mind the cost of operation and the entry permissions. The regulators ask for audibility, transaction finality and immutability of the ledger, which can be related to the reliability construct (Whitworth et al., 2008).Finally, the end users see

ease of use, user security, reach, entry permissions, legality confidentiality and transparency as important factors in order to adopt a blockchain digi-tal asset system, found in constructs such as privacy, flexibility and security (Bitfury Group, 2016).

Due to ability of the WOSP factors to capture system performance factors of socio-technical system, and the alignment with the characteristics and user requirements of blockchain digital asset management systems , these variables will add to the explanatory power the conceptual model.

Conclusion

The factors of the WOSP model are added to the model on the basis of the system characteristics (distributed database for digital assets where trust is not a factor and reliance on system performance) derived from the outcome of subquestion 1, and thereby answering subquestion 3 ”Which blockchain specific factors can be added to the model?”.

Chapter 3

Methodology

In this chapter the methodology needed for answering subquestion 4 ”Which factors are found to have a significant effect on the consumer adoption of blockchain asset management systems?” is outlined.

3.1

Research strategy

In this reasearch, a deductive approach is used to test a conceptual adoption framework in a survey on the sociotechnical blockchain warranty solution. Furthermore by adding more technical variables to the overly social based adoption literature, this research tries to discover other variance explained for consumers intention to adopt technology. This strategy suits the exploratory nature of the research question in this study. Due to the limited resources, the decision was made to only test the newly added variables. Also, earlier research shows that the UTAUT2 model has been tested and validated on multiple occasions and within various technological contexts.

3.1.1

Conceptual model

Figure 3.1: Conceptual Model

UTAUT2 hypotheses

Due to the lack of resources and to the large amount of questions, the UTAUT2 variables are not tested. Since the UTAUT2 model originated from TAM and is widely validated across various of information systems, the normal hypotheses are expected to be valid (Venkatesh et al., 2012).

Reliability

Derived from the characteristics of blockchain digital asset management, mentioned in section 2.2, audibility, transaction finality and immutability of the system are important for the adoption of the system (Bitfury Group, 2016). It can be logically argued that these characteristics align with the reliability construct of the WOSP model.

H1. Reliability positively influences the intention to adopt blockchain digital asset management systems.

Privacy

In section 2.2, legality, confidentiality and transparency of the system are mentioned to postively influence the adoption of the digital asset system (Bitfury Group, 2016). It can be logically argued that these characteristics are captured in the construct of privacy in the WOSP model.

H2. Privacy positively influences the intention to adopt blockchain digital asset management systems.

Security

Following the characteristics of blockchain digital asset management, men-tioned in section 2.2, entry permissions, user security and counterfeit re-sistance of the system are key factors for the adoption of the asset system (Bitfury Group, 2016). It can be logically argued that these characteristics are part of the security construct of the WOSP model.

H3. Security positively influences the intention to adopt blockchain digital asset management systems.

Flexibility

Based on the characteristics of the digital asset system shown in section 2.2, the reach of the system is important for the adoption of the system (Bitfury Group, 2016). It can be logically argued that these characteristics are in-cluded in the flexibility construct of the WOSP model.

H4. Flexibility positively influences the intention to adopt blockchain digital asset management systems.

Connectivity

From the stakeholders point of view, mentioned in section 2.2, the rules re-garding transactions of the system are factors influencing the adoption of the digital asset system (Bitfury Group, 2016). It can be logically argued that these characteristics can be seen as a part of the connectivity construct of the WOSP model.

H5. Connectivity positively influences the intention to adopt blockchain digital asset management systems.

Extendability

Two of the factors for blockchain digital asset management adoption, men-tioned in section 2.2, are openness to third party applications and the ease of application development (Bitfury Group, 2016). It can be logically argued that these characteristics are part of the extendability costruct of the WOSP model.

H6. Extendability positively influences the intention to adopt blockchain digital asset management systems.

Construct operationalization

The measures used for this study were adopted from a study testing the WOSP factors on also socio-technological software (Whitworth et al., 2008). Therefore the construct operationalization is based on prior research. Since the measures are tested but not validated, this is taken into account in the study design. The operationalized measures can be found in appendix B.

3.2

Study design

The hypotheses arising from the conceptual model will be tested by deploying a survey using Qualtrics software. Since the operationalized measures are not validated and the newness of the software, a pilot survey will be deployed first

in order to determine the quality. Second, the main online survey will target the general population in order to find out which technical factors influence the adoption intention of blockchain based digital asset management systems.

The online survey method is chosen in order to reach a large group of partic-ipants. The items will be measured on a human interaction level, since this also the level in which the original UTAUT2 model applies.

3.3

Studied object

The software tested in this research is a design for a blockchain digital asset management system, namely a warranty asset system, focusing on product warranties. Starting from a consumer problem, handling product warranty, the warranty solution was born. Van de Ven (2016) proposed a couple of sub-problems that should be handled, i.e. making activating, using and changing product warranty easier, more secure and tamper-proof. The current prod-uct process works as follows; When the user receives a receipt, a uniquely QR-code is on the receipt. This QR-code contains all product purchase in-formation, like item, serial number, date and timestamp etc. Scanning the QR-code referrers to the warranty bot on Facebook Messenger. Using an instant message, the picture of the receipt is put in the warranty bot. Af-terwards, the system will unwrap the code and store all product information on the (bitcoin based) blockchain. Therefore, the consumer is able to store, exchange, and change receipts (Van de Ven, 2016).

Figure 3.2: Components product warranty system (Van de Ven, 2016)

The product consists of a couple of concepts. Foremost, the Bitcoin blockchain is the main element of the warranty product. Furthermore the design choice for not creating a mobile app or website, but integrating it into a messenger service as an interface is also key to this product. Another component is the interaction that happens via a chatbot. Lastly the warranty QR-code, the transferable information, is conveyed on the receipt (Van de Ven, 2016).

3.4

Pilot study

The chosen type of pilot study is participatory by informing respondents that the survey is in a pre-test phase. This type of pilot study allows for capturing their reactions, comments and suggestions towards the warranty system. Since blockchain based asset systems are new to the market, gather-ing deeper insights in the respondents view on the survey and it’s questions is key in order to derive quality results in the survey. For the pilot study are five people selected, which can be divided into the different segments of con-sumers. They differed in age, eduction level, general and socio-technological IT knowledge and blockchain knowledge.

Besides minor syntax changes, the pilot study validated the statements. By two out of five respondents, it was questioned if blockchain should be

men-tioned in order to evaluate the variables measured in the statements. The first reason to include this in the survey is due to the key role blockchain has in the functioning of the system. Furthermore it should be noted that the goal of the survey is to measure the influence of the predictor variables on the intention to adopt a system instead of only an application. Lastly, blockchain is the key differentiating factor and the enabler of such a system. Therefore it is concluded to leave the blockchain term into the survey, but define the technology in the introduction of the survey. Based on the un-explored, innovative and disruptive nature of the blockchain technology this compromise seemed reasonable in order to be able to explain the warranty system to a broad audience.The second point of feedback were minor changes to the balance in the introduction of the warranty system in order to make it understandable for the general consumer, without a loss on accurate mea-surement of the variables.

Finally, the statements based on a communal level were found to be hard to understand by all five respondents, which would result in inaccurate mea-surement. Therefore the decision was made to exclude the statements from the survey. However two of the respondents mentioned that blockchain expe-rienced innovators would be a better target audience for the communal based questions. They are more experienced with thinking on communal level of socio-technical systems.

3.5

Participants

This research targeted the general population in the Netherlands. The reason for a choosing a broad target audience is to get a good overview of the whole range of possible users of the blockchain applications, which are needed for the success of current and future blockchain based applications. The partici-pants will be selected on convenience sampling basis, due to time constraints and in order to reach a sufficient sample size. The survey was distributed via mailing lists at a consultancy firm, various sport clubs, university and via posts on social media (Linkedin and Facebook). To calculate the sample size needed in order to measure a medium effect (0,13) with a minimum power of 0,8 for 6 predictor variables for an multiple regression, the Gpower tool is used. This resulted in a minimum sample size of 98 respondents.

3.6

Data analysis

The methods used for testing the results derived from the survey will be discussed below. The software used for doing these analysis were SPSS and AMOS. First, the reliability of the used measures will be explored via cal-culation of the Cronbach’s Alpa of each set of items per construct and the correlation between the items. Second, the validity will be checked by per-forming a confirmatory factor analysis. The confirmatory factor analysis is used due to the exploratory nature of the factors tested in this study. Third, a multiple regression is performed in order to examine the relations between the factors and the intention to adopt the tested blockchain digital asset manage-ment system. Using a combination of factor analysis and multiple regression is in line with previous UTAUT2 adoption research in related fields: mobile services (Carlsson et al., 2006), information systems (Segura and Thiesse, 2015) and internet banking (Arenas-Gait´an et al., 2015). Structural equa-tion modelling (SEM) is used in similiar studies (Malik Bader Alazzam et al., 2016), as a powerful method for performing the analysis of multiple regres-sion equations of all variables simultaneously (Alavifar et al., 2012). However due to testing the added variables and not the whole proposed model, SEM analysis is overpowered and sequential testing with multiple regression will be sufficient. Finally, the Pearson correlation test is used to spot differences between possible moderating factors as blockchain and ICT knowledge, age and education.

Chapter 4

Results

In this section the results of the survey will be analysed and subquestion 4 ”Does the conceptual model has a siginificant effect on the consumer adop-tion of blockchain asset management systems?” will be answered. First, the descriptives will be discussed. Second, the reliability of the constructs are examined. Third, the construct validity will be evaluated. Finally, the hy-potheses proposed in 3.1.1 will be tested.

4.1

Descriptives

The number op respondents of the survey was 203. Of the 203 surveys, 36 responses were only partially completed and were removed from further analysis. Therefore the survey had a completion rate of 82%. This is in line with the amount of valid responses in related research: 157 (Carlsson et al., 2006), 182 (Yang, 2013) and 157 (Wagner and Hess, 2013). Of the 167 valid responses were 28% female and 72% male respondents. The respon-dents age, measured in groups, were from 18 till 29 years (65%), 30 till 44 years (19%), 45 till 59 (15%) and 60 years and older (1%). As expected the largest group of respondents were in the highest completed education group (87%), followed by middle (10%) and low educated (3%). All the respondents used internet applications daily (10%) or multiple times a day (90%). Of the 167 respondents 60% has heard of blockchain before, which leaves a group of 40% that did not hear of blockchain until the survey. Only 20% out of all respondents had used a product or service supported by blockchain before.

The descriptive table, which can be found in appendix C.2, shows that pri-vacy (µ = 5,92) and connectivity (µ = 5,88) have the highest means. Fur-thermore it shows that the behavioral intention has a µ of 4,85 on a seven point Likert scale. The highest standard deviation is of behavioral intention (σ = 1,474), while the lowest is of flexibility (σ = 1,203).

4.2

Reliability

In order to test the reliability of the results from the Likert scale based ques-tionnaire, the correctness of the measurement for each construct is inspected. In this research the Likert data has been treated on a interval level even though there is an ongoing debate regarding the treatment of Likert data as categorical data or interval-level data (Norman, 2010). Since recent stud-ies show that Likert data can be used for parametrics test, this seems valid (Norman, 2010). The PP plot in appendix C.1, shows that normality can be assumed. Based on the confirmed parametric assumptions, the analysis can be continued.

The Cronbach’s Alpha is used for testing the reliability and the individual items are checked whether they contribute to the overall reliability of the construct. In social science research a Cronbach’s Alpha higher than 0,70 is considered acceptable and 0,80 is found to be reliable (Brah and Ying Lim, 2006).Table 4.1 shows that all of the constructs have a Cronbach’s Alpha of 0,7 and higher, which is acceptable as stated earlier. The constructs secu-rity, reliability, privacy and behavioral intention have a Cronbach’s Alpha of above 0,8.

Table 4.1: Cronbach’s Alpha of constructs and items Construct/ Item Corrected Item-Total Correlation Cronbach’s Alpha (if item Deleted) Extendability 0,773 Extendability1 0,527 0,802 Extendability2 0,666 0,636 Extendability3 0,651 0,651 Security 0,815 Security1 0,651 0,760 Security2 0,630 0,785 Security3 0,721 0,690 Reliability 0,853 Reliability1 0,727 0,795 Reliability2 0,695 0,821 Reliability3 0,755 0,767 Flexibility 0,743 Flexibility1 0,526 0,710 Flexibility2 0,609 0,611 Flexibility3 0,573 0,653 Connectivity 0,758 Connectivity1 0,635 0,629 Connectivity2 0,709 0,545 Connectivity3 0,454 0,849 Privacy 0,826 Privacy1 0,732 0,722 Privacy2 0,652 0,798 Privacy3 0,678 0,765 Behavioral Intention 0,936 Behavioral Intention1 0,842 0,926 Behavioral Intention2 0,866 0,907 Behavioral Intention3 0,895 0,885

As can be derived from table 4.1, all items, except connectivity3, have a correlation higher than 0,5 and therefore it can be assumed that the items

are measuring the same construct. Even though the Cronbach’s alpha of extendability could be improved by dropping item extendability1, the item is not removed, since the minimal improvement of the reliability of this con-struct when leaving out extendability1, does not out weight the loss of data and the Chronbach’s Alpha is well above the needed threshold. However, the connectivity construct shows a significant improvement of reliability, when connectivity3 is dropped. This improves the Cronbach’s alpha from 0,758 to 0,849, while also dropping a low correlated item (0,454). To conclude, all items,except connectivity3, will be included for further analysis.

The adjustments made on the basis of the analysis on the reliability by drop-ping item connectivity3, results in the following overview of the Cronbach’s Alpha on the constructs.

Table 4.2: Overview Cronbach’s Alpa

Construct Cronbach’s Alpha Extendability 0,773 Security 0,815 Reliability 0,853 Flexibility 0,743 Connectivity 0,849 Privacy 0,826 Behavioral Intention 0,936

4.3

Validity

The construct validity is tested by performing a confirmatory factor analysis. Since the nature of this study is highly exploratory, the congruence of the new added constructs is important. Therefore, the factor loadings of the items that make up the constructs are examined. As can be seen in table 4.3, the standardized factor loadings of the items, obtained from the factor model shown in appendix C.3, are on average above 0,7 and therefore contribute

to the unobserved latent construct, as this guideline is also used in previous information system research (Doll et al., 1995).

Table 4.3: Standardized factor loadings

Item Stand. factor loadings Extendability1 0,558 Extendability2 0,787 Extendability3 0,863 Security1 0,767 Security2 0,714 Security3 0,843 Reliability1 0,734 Reliability2 0,830 Reliability3 0,856 Flexibility1 0,758 Flexibility2 0,674 Flexibility3 0,647 Connectivity1 0,846 Connectivity2 0,872 Privacy1 0,908 Privacy2 0,709 Privacy3 0,735 Behavioral Intention1 0,882 Behavioral Intention2 0,904 Behavioral Intention3 0,949

In order to determine the convergent validity, the average variance extracted (AVE) is checked to be above 0,5 (Premkumar and Bhattacherjee, 2008). The results in table 4.4 show that all values are above 0,5, except flexibility. Since the average variance extracted of flexibility is really close to 0,5 it is included in further analysis. Therefore it is argued that the convergent validity of the measures is acceptable.

Table 4.4: Overview average variance extracted Construct AVE Extendability 0,559 Security 0,603 Reliability 0,653 Flexibility 0,482 Connectivity 0,738 Privacy 0,622 Behavioral Intention 0,832

When looking at the discriminant validity, it is found that the square root of the AVE of most constructs, except behavioral intention, is lower than the correlation of other constructs. This is due to the high overall correlations between all the constructs, but should be taken into account when analyzing the results and formulating conclusions.

4.4

Significance relations

On the basis of the performed validity and reliability analysis, the constructs are potential determinants of the behavioral intention to adopt blockchain digital asset management systems. In order to test the predefined hypothe-ses, which predict a significant relation between the constructs and behav-ioral intention, a multiple regression analysis is used. The results, which can be found in figure 4.1, show that extendability (ρ = 0,002), flexibility (ρ = 0,014) and connectivity (ρ = 0,012) have a significant relation (ρ < 0,05) with behavioral intention. Also the results show that security, reliability and privacy did not have a significant relation with behavioral intention to adopt, since ρ is higher than 0,05.

Figure 4.1: Regression coefficients

Figure 4.1 also shows that connectivity (β = 0,349) has the largest impact on behavioral intention, followed by flexibility (β = 0,321) and extendability (β = 0,301). Furthermore the VIF values, which are lower than 10, reject the possibility of multicollinearity, which means that each constructs explains a different part of the behavioral intention to adopt.

Figure 4.2 shows that the means of the different constructs are significantly (ρ < 0,05) different. Together with the knowledge of the VIF factor, it can be argued that the constructs are measuring a different part of adoption intention.

The results of the R2 _{explain how much the model can explain the behavioral}

intention and how well the data fits the model, which is shown in figure 4.3. With a R2 of 0,43, the explained variance of the model is relatively high in social science research.

Figure 4.3: R-square

In order to measure the influence of age, gender, education and blockchain knowledge on the relation between each of the constructs and the behavioral intention to adopt blockchain digital asset management systems, a Pearson correlation is performed. The results are shown in figure 4.4, which reveals that internet usage, heard of blockchain and blockchain usage only influence the relation between extendability and behavioral intention.

Chapter 5

Discussion and limitations

This research has focused on designing and testing a conceptual model for which the adoption of a blockchain digital asset management system can be measured. The first subquestion, “What are the characteristics of blockchain digital asset management systems?”, showed that blockchain digital asset management systems are based on a distributed database, wherein trust plays no longer a role, which allows for the management of digital assets, with a large technical basis and a social impact. The second subquestion, “Which adoption model can measure the adoption of blockchain digital as-set management systems?”, was answered by providing evidence that the UTAUT2 adoption framework was most applicable to blockchain asset man-agement, due to it’s usage in related fields and based on the most replicated and tested TAM model. The third subquestion, “Which blockhain digital as-set management specific factors can be added to the model?”, was explained by adding factors, derived from systems theory, of the WOSP model. These factors were extendability, flexibility, connectivity, reliability, security and privacy. The fourth subquestion was “To what extent does the conceptual model explain the consumer adoption of blockchain digital asset management systems?”. The results show that hypotheses 1, 2 and 3 are found not to be supported. However, the results showed that hypotheses 4, 5 and 6 are sup-ported.

The first hypothesis that stated that reliability positively influences the in-tention to adopt blockchain digital asset management systems, was found not to be supported. Other research has found that technicalities of the system (in which system reliability is included), has a negatively effect on the

adop-tion of another informaadop-tion system named mobile internet (Kim et al., 2007). The difference between the studies is that in the study of Kim et al., relia-bility is posed as a barrier to adoption instead of as a contributing factor to the adoption of an information system. Hypothesis 2 proposed that privacy positively influences the intention to adopt blockchain digital asset manage-ment systems, was not supported by the results of this research. In recent research on information systems in healthcare, this relation of privacy with the adoption of the information system is studied. The research tried to ex-plain the effect, but concluded that many empirical studies showed different results concerning the influence of privacy on adoption (Angst and Agarwal, 2009). In Hypothesis 3 it is proposed that security has a positive influence on the intention to adopt blockchain digital asset management systems. This is not supported by the results of this study. Although it seems that it does not relate to adoption, a recent study on e-banking services showed that se-curity as well as privacy are the major causes of dissatisfaction of consumers (Poon, 2007). Hypothesis 4 proposed that flexibility has a positive influ-ence on the intention to adopt blockchain digital asset management systems. The results show that the hypothesis is supported. Studies in related fields confirm that flexibility can be of influence on the adoption in the context of mobile systems (Anckar and Walden, 2003). Hypothesis 5 proposed that connectivity positively influences the intention to adopt blockchain digital asset management systems. This was found to be supported by the results of this research, which confirms results of studies on mobile internet, that showed that connectivity has an important influence on the adoption and continuance of consumers (Kim and Kim, 2003).

Finally, hypothesis 6 proposed that extendability positively influences the intention to adopt blockchain digital asset management systems, which was supported by the results of this study. This outcome aligns with a recent study on management information system adoption, which also concludes that extendability has an effect on adoption (Ngai et al., 2009). An interest-ing result of this study is that all the supported hypotheses are opportunity based, while all the unsupported hypothesis tend to focus on risk reduction. This is in contrast with the research performed on browsers as socio-technical system using the factors of the WOSP model, in which all factors are found to be significant (Whitworth et al., 2008).

Blockchain digital asset management systems could be defined as a new in-novation, which is currently in the technology development cycle and going to be adopted first by innovators and early adopters (Rogers, 2010). Early adopters are in general more favorable to change, are less risk averse and more adventurous. Therefore it can be argued that consumers are more focused on the potential of the blockchain innovation, rather than the risks. Another explanation could be that due to the newness of the system, the focus of con-sumers is in the early stage of the life cycle of the system and thus the focus of consumers lays more on the advantages. In latter stages of the product life cycle, originating from the diffusion theory, when the innovation has reached maturity, the focus is expected to shift to more risk reduction based fac-tors in the usual S-shaped diffusion curve (Mahler and Rogers, 1999). The interesting finding of only opportunity based factors influencing the adop-tion, contradicts the studies that found the risk averse attitude of consumers influencing the adoption of technological innovations (Bauer and Hein, 2006).

This study provides a conceptual framework for measuring adoption of future blockchain digital asset management systems. The added technical factors explaining a large portion (43%) of the intention to adopt blockchain digital asset management systems, provided by the variance explained by the op-portunity based variables extendability, flexibility and connectivity. Using the Gpower tool to calculate the power with a 43% R2_{, resulted in a power}

of 1.

With the evidence found on the influence of extendability, flexibility and con-nectivity on the adoption intention of consumer on digital asset management systems, practitioners could focus more on these factors when developing a similar product. Furthermore the found knowledge on blockchain en digital asset management systems, could benefit society in order to understand the capabilities of the technology studied in this research. Furthermore it should be noted that the adoption of blockchain digital asset management has a large societal impact to their ability to function as decentralized organiza-tions. This leaves practitioners with a wide variety of additional factors that are possibly related to the adoption intention, which should be explored in the future.

Recommended by Venkatesh et al. (2007), this study brings in constructs from a new stream of science, namely systems theory, to the overly dominated