1 Introduction

Decentralized Finance or “DeFi” is the latest Fintech trend [1] that takes the world of Finance by storm [2]. Owing to its decentralized nature, it juxtaposes traditional finance [3] or centralized finance, also referred to as TradFi and CeFi, respectively [4, 5]. Owing to its combinatory possibilities, some authors have referred to it as “Money Lego” [6] or the “Lego of Finance” [7]. However, what exactly is in the box of financial click bricks? What precisely constitutes DeFi, and what can be built from it? Does the creation of business models from such interlocking pieces warrant a warning label? This article applies a two-pronged methodology. First, it scrutinizes the existing body of scientific literature on DeFi and distills its main characteristics, the business models that are built upon them, and the potential risks they bring along. In the second step, it verifies the theoretical findings in an empirical survey among DeFi experts.

As an immediate response to the highly centralized financial sector that brought about the 2008 global financial crisis, distributed ledger-based cryptocurrencies were introduced to facilitate financial transactions without dependence on trusted middlemen [1], thereby shifting the power from financial services firms toward the end user [8]. Throughout the decade that followed, the original blockchain concept was enhanced, and new layers of infrastructure were built, leading to the gradual emergence of a DeFi infrastructure layer on top of existing blockchain protocols [9].

Three aspects are particularly noteworthy in the context of the most recent Defi developments. First, the pace at which novel concepts and technologies are being developed and deployed is higher than ever before [10]. Second, the dissemination of DeFi concepts occurs in a decentralized manner. Third, as the youngest offspring of the Fintech family, much of the DeFi-induced interference is imposed from the outside onto the financial services industry [9] as young projects and start-ups predominantly from the technology area are attempting to disrupt the incumbents (Tables 1, 2, 3, 4, 5, 6).

Table 1 Importance of DeFi characteristics according to literature
Full size table
Table 2 Dominant DeFi business models according to literature
Full size table
Table 3 The most important risks imposed by DeFi
Full size table
Table 4 Importance of DeFi characteristics according to experts
Full size table
Table 5 Dominant DeFi business models according to experts
Full size table
Table 6 Most important risks imposed by DeFi
Full size table

As it is not only a novel subject, but one that develops in leaps and rebounds, DeFi is still an area that is not well understood in practice or academia. Therefore, this article lays some groundwork that is believed to be useful for researchers and practitioners alike seeking to engage in the DeFi world. It sets out to develop a taxonomy of the different elements of DeFi and its applications, and to derive a workable definition for the term DeFi itself. This definition and taxonomy can be used by scholars seeking to address research questions on DeFi and by practitioners who want to actively use the DeFi ecosystem.

A taxonomy is a system for classifying and organizing subjects. In doing so, it helps to understand a set of subject-specific concepts and creates a nomenclature for those concepts. A taxonomy facilitates the generation of knowledge as it creates a common language and thus reduces misunderstandings [11]. Therefore it enhances the efficiency of dealing with a particular subject. No such system exists for DeFi yet, hampering the generation of tacit knowledge of the subject. This paper at hand therefore seeks to address the following three research questions (RQ):

RQ1: What are the characteristics of DeFi?

RQ2: What are the business models built around DeFi?

RQ3: Which risks does DeFi entail?

This is done by applying a three-staged research approach. First, a comprehensive literature review is being conducted and the findings are reported. In the second stage, insights from theory are refined in an empirical study among experts from academia as well as industry. This is done to align the expressions derived from scientific texts with the jargon used by practitioners. As jargon is commonly applied in specific communication settings and may not be well understood outside those contexts, it is deemed necessary to bring the theoretical view to congruence with the academic view [12]. Third, the discoveries from the theory and the empirical survey are synthesized and discussed before recommendations are derived on how to classify and organize the constituents of DeFi.

Hence, the objective of this theory-building work is to provide a taxonomy that standardizes the characteristics and perils of DeFi and identifies the business models it renders possible. These insights will support future research efforts on this subject and will help design DeFi strategies for Fintech firms and incumbent players alike.

Thus, this study contributes to the advancement of knowledge on this novel topic by further fleshing out its ontology and most notably investigating its characteristics, the business models built around DeFi, and the risks implied by DeFi. Akin to the developments in the field of Fintech, the results of this study will help researchers to work and communicate more efficiently on the topic of DeFi, and it provides practitioners with hands-on advice on potential caveats arising from the field of DeFi [13].

The remainder of this paper is organized as follows. In the second section, the background of the topic and various definitions of DeFi are provided. This is followed by a literature review in section three that takes a qualitative stock of the characteristics of DeFi and provides an ordinal orientation of their usage along with DeFi’s applications and the risks it poses, as described in theory. section four consists of an empirical study in which the findings from the literature review are validated by subject matter experts. The results of the literature review as well as the empirical study are jointly assessed in the discussion part of the paper, which constitutes section five. Here also a taxonomy will be developed. The paper ends with a concluding section six which also presents limitations as well as future research avenues.

2 Background

Despite its success, it must be recognized that no clear definition exists yet for the term DeFi. Numerous authors have approached the topic via the phenomenology of DeFi and described for instance components of it, e.g. Gąsiorkiewicz, Monkiewicz [14] who proffer that “[a] vital component of decentralized finance is crypto-currencies” (p. 145) or Hughes [15] who suggests that “DeFi is a broad term that refers to decentralized finance generally” and that “[i]t can include issuances that are designed to be decentralized despite using a permissioned ledger, along with truly decentralized issuances on public blockchains.” (p.902). Other authors describe the functionalities it provides, as did Bailey, Rettler [16] by saying that “[i]n DeFi, we find applications for borrowing and lending, derivatives trading, synthetic financial products, generating yield, and so on.” (p.2). Yet other researchers focus on the objectives of DeFi, such as Caldarelli and Ellul [17] who state that “Decentralized Finance (DeFi) takes the promise of blockchain a step further and aims to transform traditional financial products into trustless and transparent protocols that run without involving intermediaries” (p.1).

Only a few authors have attempted to define the word itself. Katona [6] did so by stating that DeFi is tantamount to blockchain-based financial services (“Decentralized Finance (DeFi), an umbrella term connoting blockchain-based financial services, provides compelling opportunities in electricity projects financing, which have historically been burdened by bureaucratic processes.”, p.2). Similarly, Zetzsche, Arner [2] propose that DeFi is a specific way of providing financial services (“Hence, in this article we understand DeFi to comprise, at its core, what its simple name suggests: the decentralized provision of financial services through a mix of infrastructure, markets, technology, methods, and applications.”, p.173). Babu and Abraham [18] suggest similar when they describe DeFi as a specific form of finance (“Decentralized Finance (DeFi) DeFi is a form of cryptocurrency-based finance that uses automated Smart Contracts to manage the lending/recovery/liquidation process.”, p.121).

In contrast, Halden et al. [19] suggest that DeFi is a type of financial technology ("More recently matured digital tools such as DLT and related financial technologies such as Decentralized Finance (DeFi) and FinTech options have the potential to minimize the transaction time and introduce close to real-time markets.”, p.138). In a similar vein, Caldarelli and Ellul [17] propose that DeFi is a use-case for public blockchains (“Decentralized Finance (DeFi) is proving to be one of the most significant use-cases for public blockchains, with over fifty billion dollars of value locked and growing”., p1.)

Other authors differ when they either posit that DeFi is an infrastructure as Schär [9] did (“Decentralized finance (DeFi) is a blockchain-based financial infrastructure that has recently gained a lot of traction.”, p.153), or a set of decentralized applications as Stepanova and Eriņš [20] put it (“DeFi is a set of decentralized applications (dApps) that automate financial services based on the blockchain technology without any centralized control.", p.328), or a financial system as Tien, Wang [21] proffered (“However, because of Decentralized Finance (DeFi), a more transparent and interoperable financial system on the Ethereum blockchain, there is indeed a chance that cryptocurrency can be paid for a service and generate extra profit at the same time.”, p. 503). Kutsyk et al. [22] even suggested that Defi is an ecosystem (“Many projects are now being implemented in the field of decentralized finance (DeFi), which is a decentralized, public and unreliable ecosystem that combines various financial services based on public blockchains, mainly Ethereum.”, p.102).

Given this broad ontological spectrum provided for DeFi, including financial services, provision of financial services, finance, financial technology, use-case, infrastructure, set of decentralized applications, financial system, and ecosystem this paper will follow the basic notion of Arusoaie [23] who described DeFi as a specific paradigm: “DeFi is a peer-to-peer financial paradigm which leverages blockchain-based smart contracts to ensure its integrity and security.” (p.6).

For this research, the phenomenon of DeFi is purposely treated as a paradigm, as it is the widest form of consensus within a science, yet it sets apart one scientific group from another. Simultaneously, a paradigm incorporates, defines, and connects models, theories, methods, and tools that exist within its domain [24]. However, even when using this rather large common denominator, another obstacle emerges before a unifying definition can be provided when looking at the above-mentioned definitions of DeFi: As varied as the objects that describe the term DeFi are the characteristics ascribed to them. They include attributes such as “decentral” [2, p.173], “decentralized” and “disintermediated”, [18, p.121], “transparent and interoperable” [21, p.503], and “public and unreliable” [22, p.102]. This further underscores the need for a sound taxonomy delineating the characteristics of DeFi, the business models it enables, and the risks it entails.

3 Literature review

In the first stage of the scientific investigation, a literature review was conducted. The methods applied to conduct the review and to derive results are described in the following sections.

3.1 Literature review methods

Following the protocol suggested by Tranfield, Denyer [25] a systematic literature search was conducted across all available databases about social sciences, business administration, management sciences as well as economics and related fields of information technology for all papers published until September 30th 2021, using the keywords “DeFi”, “Decentralized Finance” and “Decentralised Finance” anywhere throughout the publication. These databases comprised EBSCO, Business Source Premier, ABI Inform/Proquest, Directory of Open Access Journals (DOAJ), JSTOR, SAGE, Science Direct, Springer Link, Taylor Francis, and the Wiley Online Library. To ensure that no source was overlooked, the results were cross-checked with SCOPUS and Web of Science. As the term DeFi was only coined in recent years, the start date was fixed to January 1st, 2018. To control for the quality of the articles, a delimiter was set that only scholarly publications (i.e., peer-reviewed journal articles and conference papers) would be included in the results. Book reviews were excluded because the assessed books were not peer-reviewed. Moreover, the relevant language was set to English. It was ensured that the searches were not case-sensitive. After excluding double entries, this search yielded 118 scientific papers on DeFi. These articles constitute the current scholarly body of knowledge on this subject matter.

As part of the research endeavor to establish a taxonomy of DeFi, particular interest was paid to the definition of the term, its characteristics, the business models built around it, and the risks attached to it. However, not all 118 texts shed light on these items. The reasons for this are described in the following paragraphs.

An entire series of previously identified texts, for instance, had to be excluded from the sample, as they use decentralized finance in a different context than the one used here, such as decentralized budgeting and financing processes [26,27,28,29,30,31,32,33,34,35,36,37].

Numerous additional articles were excluded because they did not yield any further information on the subject matter. For instance, one article does mention the term decentralized finance in the article abstract but does not elaborate on it anywhere throughout the text [see e.g. 38]. Another article references the term in a section header without further explicating it in the text: “4.4. Financial Technology and Decentralized Finance Applications and Integration” [39, p.11].

Other scholars do not refer to the term throughout the full text body, but in appendices, such as the reference section [40], in footnotes; see, for example, Kokkinis and Miglionico [41], Guseva [42], Bhushan, Sahoo [43, 44], Hinkes [45] or simply as an article keyword as was done by Majuri [46] without elaborating further on the term itself.

Still other articles mention DeFi as a data source [47] “This study analyses the determinants of interest rates in the cryptocurrency lending market using a unique database from the Decentralised Finance platform.” (p.1), or as part of a company’s description in an empirical sample [48].

Other papers do mention the term DeFi or Decentralized Finance throughout the text, yet just as a side note without elaborating on it any further, for instance, Huang, Nya [49, p.3] “The decentralization of telecommunications in terms of decentralized finance and mobile phones can be regarded as already accomplished”. Wang, Gou [50], Wong and Eng [51], and Wang, Jiang [52] treat the term in a similarly inconsequential manner.

Yet, other authors do refer to the term DeFi, but just as one trend among others that are currently shaping the field of finance similar to blockchain technology (see, for example, [53], smart contracts [54], cryptocurrencies [55,56,57,58], stablecoins [59], CBDC [60], or cryptocurrency-based credit cards [61]. Further, scholars mention it briefly as yet another technological development in the context of decentralized systems [62], decentralized applications [63], IoT [64], Artificial Intelligence [65], and collective intelligence [66].

Other scholars briefly mention the term in sociological frameworks, such as ethnography [67] or universal basic income [68].

Another group of researchers uses DeFi in the context of increasing or retreating cryptocurrency prices (see). e.g. Liang [69, p.5] “With the price of Bitcoin rising this year, the focus on decentralized finance will not be cool, but Bitcoin will never become a currency with monetary functions.” Similarly, Carvalho and Karimi [70], and Carvalho [71].

Yet, in another set of papers, the term Decentralized Finance could not be found anywhere in the text body, but only in the author’s biography section (for example, Crouch, Olefir [72] or Wang, He [73] or Gururajan and Bai [74].

In one case, the term decentralized finance even came into existence by using the past participle of the verb “to decentralize,” see Ewers, Dicce [75]: “Yet, in contrast to these agglomerative processes, globalization has simultaneously decentralized finance.” (p.6.). This article also had to be removed from the sample of scientific texts on DeFi.

For articles that elaborate on the term DeFi, its characteristics, related business models, and/or risks, it was distinguished between original works and papers largely citing the works of other scholars. For example, Alt [76] cited Schär [9] when ascribing the term “notion” to DeFi. Similarly Deng, Zhang [77] reference Chen and Bellavitis [78] when describing the characteristics of DeFi: “Chen and Bellavitis (2020) studied blockchain-based finance technology, which can reduce transaction costs, produce distributed trust, and empower new business models, thus making the financial system more decentralized, innovative, interoperable, borderless, and transparent.” (p.129). The same goes for Longo, Podda [79, p.3] who also cited Chen and Bellavitis [78] as well as Carayannis, Christodoulou [80] who, too, cited Chen and Bellavitis [78]. In all cases where secondary literature solely cites original peer-reviewed work without making a significant contribution relevant to the research questions at hand, the referenced original articles were scrutinized in great detail and the referencing article was excluded from the sample. In cases where authors cite non-peer-review sources such as websites [e.g. 81], the texts were also excluded from the sample.

3.2 Literature review results

Of the 118 scientific texts scrutinized initially, a subset of 41 papers yielded original insights into DeFi’s characteristics and/or derived business models and/or risks attached to the object under investigation. All sources with corresponding text passages can be found in the Appendix. The following paragraphs report these findings from the literature study. The descriptive results are presented in the order of the three research questions derived above.

RQ1: What are the characteristics of DeFi?

Of the 118 surveyed texts, 22 mention one or more characteristics of DeFi originally. After subsuming highly similar characteristics under one identical label (e.g. “combinatory” and “seamless” were added to the characteristic “interoperable”) 27 sufficiently distinct characteristics could be identified. They are listed in the following table in order of their mentions:

The table is led by the characteristics disintermediated (14 mentions), decentral (12 mentions), and transparent (9 mentions). Automatic, interoperable, and secure share the fourth place, with six mentions each. Deterministic, inclusive, and permissionless were each mentioned five times and thus followed in the fifth spot. Borderless and cost-efficient were listed four times in the extant body of literature as DeFi characteristics and thus occupy spot six. Innovative, open, and technology-focused were each cited three times and shared rank seven, whereas democratic, efficient, immutable, and public came on rank eight. Autonomous, anonymous, digital, empowering, entrepreneurial, fast, trustless, unregulated, and verifiable shared spot nine.

RQ2: What are the business models built around DeFi?

Among the entirety of the surveyed scientific texts, 27 pertained to business models enabled by DeFi. To categorize the aforementioned business models, wherever possible, the items contained in the stylized financial services value chain were used, as suggested by Schueffel [82] for categorizing DeFi activities. It contains the following nine items: “Accounts”; “Origination & Issuance”; “Payments”; Deposits & Loans; “Exchanges” “Derivatives”; “Wealth & Asset Management”; “Insurance”; and “Data” (p.IV). The item “Governance” was added, as mentioned in one text, but it is not part of the suggested value chain. As one or multiple business activities were mentioned in the papers the following values were derived:

With 27 mentions most often “Deposits & Loans” were mentioned as business models that are enabled by DeFi. This is followed by “Derivatives” with 19 mentions and “Origination & Issuance” as well as “Exchanges” with 17 and 10 mentions, respectively. “Wealth and Asset Management” is on rank five with 9 mentions and “Payments” follows on place six with 4 mentions. Data received 3 mentions and the business activities “Governance” and “Insurance” share spot eight with one mention each.

RQ3: Which risks does DeFi entail?

Across all the DeFi articles scrutinized, 12 yielded insights into the risk aspects of DeFi. Once filing similar risks into one risk bracket e.g. network congestion” was put with “Immature technology” 24 satisfactorily distinct risks could be identified. The following tables provide a ranked overview of these risks, along with the number of mentions.

With 5 mentions “Regulatory uncertainty” was the most frequently mentioned risk. A “Flawed oracle” and “Immature technology” followed with 4 mentions in second place. “Fraud” and “Lack of knowledge” were both mentioned three times and ended up in third rank. Ten other risks shared spot four whereas nine additional ones were mentioned once and thus ended up on rank five.

4 Survey

The second stage of this study rests on a survey of the DeFi experts. The method applied to poll the information as well as the results obtained are laid out in the subsequent paragraphs.

4.1 Survey methods

To evaluate these theoretical findings regarding DeFi, a survey was conducted with experts. Expert surveys are an effective method for assessing concepts that would otherwise be difficult—if not impossible—to capture through alternative research methods [83].

Given the idiosyncratic nature of research topics and the associated obstacles in finding many willing expert participants, an increasing number of group studies have reverted to using convenience rather than representative sampling to recruit participants [84]. In general, such non-probability samples do not meet the basic assumptions of inferential statistics. Nevertheless, research building on convenience samples can yield highly insightful results for establishing the plausibility of relationships among variables which is a conditio sine qua non for theory-building work [85].

Besides, these samples are oftentimes based on the organizational and personal characteristics of respondents rather than archetypical attributes [86]. In addition, and to mitigate one of the major shortcomings of expert panels, i.e. that they lack diversity, great store was set by inviting experts from a wide range of geographies, industries, and professions [87], thus creating a sample stratified for these three strata. The latter criteria ranged from university professors and scientists to bankers and entrepreneurs with in-depth knowledge of DeFi. The experts were invited to answer the survey by e-mail or online messaging service to provide them with the highest possible convenience.

Possible panelists were identified through four sources: first, panel candidates were identified as managers working for companies that are active in the DeFi space. Second, they were identified as attendees of conferences relating to the topic of DeFi. Third, participants were found using the business network LinkedIn, and, fourth, they were discovered using snowball sampling, i.e. by referrals of experts who were already participating. As the objective of the expert survey was to validate the findings from the extant body of scientific knowledge, any authors identified in the literature review were excluded from the panel.

In total 36 personalized invitations were sent out through e-mail, LinkedIn, or WhatsApp. Multiple personalized follow-ups were sent out in cases of non-respondents. In the end 34, experts gave their feedback in the period from Jan 1st 2022 to February 28th 2022, representing a panel of sufficient size and spanning all inhabited continents of the world.

The participants were asked to evaluate the suitability of the characteristics, business activities, and risks attached to DeFi. For this purpose, a five-point Likert scale was developed dividing pairs of opposite statements. Respondents could choose from “Totally irrelevant” to “Totally relevant” for the DeFi characteristics, from “DeFi does not enable at all” to “DeFi enables strongly” for the business models, and from “No risk at all” to Highly serious risk” for the jeopardies that DeFi imposes on users.

4.2 Survey results

In total, 34 experts answered the survey, thereby not only exceeding the minimum threshold of four participants for focus groups [88] but also five to ten panelists for homogeneous populations of expert panels and the threshold of 15 to 30 for heterogenous populations of such groups [89,90,91]. As only two experts did not respond to the survey but were not members of any identical strata (geography, industry, and profession), it was deemed futile to conduct a non-response analysis as those two missing responses would have not significantly and systematically skewed the ordinal survey results.

Most of the respondents were entrepreneurs (N = 10) followed by bankers (N = 8), finance experts (N = 7), educators (N = 6), researchers (N = 5), IT specialists (N = 4), legal practitioners (N = 4) and managers (N = 4). Yet, also a blogger (N = 1), an author (N = 1), and a crypto strategist (N = 1) participated in the survey. It is noteworthy, however, that the field of expertise was open, so each expert could name more than one occupation (Fig. 1).

Fig. 1
figure 1

A taxonomy of DeFi. This figure outlines the key characteristics, business models, and risks associated with DeFi with the width of the boxes denoting the item’s importance in an ordinal fashion

Full size image

While a Shapiro–Wilk test for the normality revealed that not all variables were distributed normally, Cronbach’s Alpha tests were executed to assess the interrater reliability of the groups of items assessed [92]. This was done because Cronbach's Alpha construct is based on the mean of a set of Pearson's correlations and has thus limitations resulting from non-normal distributed data. Yet, strong departures from normality will merely limit the maximum value of the calculated Cronbach Alpha [93].

The Cronbach alpha for the set of characteristics was 0.906, for the range of business models, 0.801, and for the risks under observation 0.837. As those figures are well above the necessary reliability threshold of 0.60 to 0.70 [94] they can be considered reliable [95].

Moreover, the interclass correlation coefficients of all three sets of variables were calculated, i.e. for characteristics, business models, and risks. They were 0.906, 0.801, and 0.837 respectively at a significance level of alpha < 0.01. Hence, 90.6% of consistency exists among the raters of DeFi characteristics, 80.1% of consistency exists among the respondents giving feedback on business models and 83.7% of consistency exists among the respondents on the risks. These values point towards a high reliability of the measured constructs.

As was done for results of the literature research the outcomes of the expert survey are presented in the order of the three research questions that were developed above.

RQ1: What are the characteristics of DeFi?

On the question of which characteristic they deem highly important with regards to DeFi the surveyed experts named “borderless” as the most crucial feature (Mean 4.35). It was followed by “decentral” (Mean 4.5) and verifiable (4.45). Secure, cost-efficient, and digital were ranked four, five, and six with Mean values of 4.44; 4.39; and 4.38 respectively. The subsequent spots were taken by the characteristics “efficient” on place seven with a mean of 4.35, “immutable” on place eight with a value of 4.26, “innovative” with a mean of 4.26 on spot nine and “trustless” with a mean of 4.21 on place 10. The values are listed in the table below. Due to space limitations and their minor importance, the characteristics ranking in the latter two-thirds of the table are not elaborated in detail.

RQ2: What are the business models built around DeFi?

According to the findings of our survey, DeFi experts consider “Payments”, “Deposit & Loans” as well as “Exchanges” as the most important three business models emerging in DeFi (Means values 4.35; 4.29, and 4.29 respectively). “Derivatives” and “Wealth & Asset Management” follow on four and five with mean values of 4.09, and 3.91 respectively as can be seen from the following table:

Again, due to constraints on space and their relatively lower significance, the business models listed in the bottom half of the table are not described in depth.

RQ3: Which risks does DeFi entail?

When it comes to the inherent risks that DeFi brings along, “Coding Errors” are viewed as the most severe ones with a mean of 4.24. “Fraud” follows in place 2 with a mean of 4.12, in place 3 follows the “Flawed Oracle” and closely followed by “Illicit Activity” in four, both with a mean of 3.88. Regulatory uncertainty follows ranks 5 with a mean of 3.82. Compromised privacy and Concentration of governance came in on spots 6 and 7, both with means of 3.71. On place eight the risk “Immature markets” is listed and on eight “Immature technology with means of 3.50 and 3.47 respectively. With a mean of 3.47 lack of knowledge is placed on spot 10. Here, too, given space constraints and their relative insignificance, the characteristics positioned from rank 11 to 24 are not discussed in depth.

4.3 Summary of results

The investigation delved into the topic of DeFi, guided by three pivotal research questions: the distinctive characteristics of DeFi, the emerging business models it fosters, and the spectrum of risks it encompasses. Through an exhaustive review of the literature paired with an empirical survey amongst a diverse panel of DeFi experts, the study sheds light on the nuances of this nascent financial phenomenon.

In dissecting the characteristics of DeFi, the study brought to the forefront 27 distinct attributes, among which “borderless”, “decentral”, and “verifiable” stood out as the cornerstones. This trio of traits encapsulates DeFi's essence: its disregard for geographical bounds, its shift of control from centralized powers to the hands of the individual, and its commitment to transparency through verifiable transactions. Such an amalgamation of features distinctively positions DeFi as an innovative force capable of reshaping the financial landscape.

Venturing into the realm of business models catalyzed by DeFi, the analysis identified nine principal models, with “Payments”, “Deposits & Loans”, and “Exchanges” distinguished as the forefront runners. This finding highlights DeFi’s role as a catalyst of innovation, challenging and redefining traditional financial mechanisms through the facilitation of direct, peer-to-peer transactions.

Furthermore, the probe into the associated risks unveiled a nexus of 24 distinct concerns, pinpointing “Coding Errors”, “Fraud” as the two foremost challenges. This insight underscores the embryonic state of DeFi technology and underscores the critical need for fortified security measures to shield against potential vulnerabilities.

5 Discussion

When comparing the findings from the literature review to the results from the expert survey, several outcomes are noteworthy. As far as the DeFi characteristics are concerned it is striking that neither the scientific experts nor the survey respondents deemed “decentral” the most important characteristic, but “disintermediated” and “borderless” respectively. In both groups, the attribute “decentral” only came in second place, which is interesting as it is the very characteristic constituting part of the name. Being “transparent” was ranked at place three in literature but dropped to spot 15 in the empirical study which is substantial even when considering that the finer-grained survey method did not produce shared ranks. Place three is now occupied in the expert surveys by the feature “verifiable”. The elements “automatic”, “interoperable” and “secure” followed rank four of the literature survey. The first two properties dropped to ranks 13, and 12 respectively, while “secure” moved up to rank four. Other big drops in importance could be observed for the characteristics “deterministic”, “inclusive” and “technology-focused” whereas the survey respondents majorly upvoted the properties “digital”, “efficient” and “immutable”.

Founded on the extant body of literature and tested through a survey with a group of DeFi experts, a DeFi taxonomy is proposed comprising 27 DeFi characteristics. Furthermore, 9 distinct business models were identified, and 24 risks were derived that are imposed on to the user by DeFi. This taxonomy is depicted in the following graph with the width of the boxes indicating the importance of the items as per the empirical study. More specifically, the width reflects in relative terms the mean value of the items mentioned in the previous three tables. The row on the left-hand side of the graph depicts the characteristics of DeFi from the most important property mentioned, “borderless”, to the least important one “unregulated”. The middle column depicts the significance of business models mentioned in the context of DeFi, from “Payments” (most significant) to “Data” (least significant). The stack on the right contains the risks brought about by DeFi from “Coding Errors” indicated by the survey participants as the most severe one to “Immutability” nominated as the least grave.

Based on these results it can be concluded that most DeFi experts agree that DeFi is first and foremost borderless, decentral, verifiable, secure, and cost-efficient. Specifically, this means that, first, the borderless nature of DeFi substantially widens access to financial services worldwide, overcoming geographical barriers and fostering inclusivity. This not only diversifies its user base but also amplifies the flow of capital across borders, thus extending the global reach of financial innovations. Second, DeFi’s decentralized structure transfers governance from traditional centralized financial bodies to a widespread network of users, cultivating a democratic and fair financial landscape. This shift significantly mitigates the risks of systemic disruptions often seen with centralized entities and complicates the implementation of censorship and external oversight. The third key characteristic is the verifiability of transactions through distributed ledger technology, which ensures transparency and trust among users. In a system lacking conventional regulatory structures, the ability of users to independently confirm the legitimacy of transactions is crucial, enhancing confidence in the system’s integrity. Fourth, security in DeFi is fundamentally strong, underpinned by advanced cryptographic methods and consensus algorithms that protect against fraudulent actions and security threats. Fifth, DeFi stands out for its cost efficiency, achieved by eliminating middlemen and leveraging smart contracts to automate processes. This results in lower transaction fees and fewer operational inefficiencies, making financial services not only more accessible but also more competitive relative to traditional financial institutions.

In addition, and grounded on the findings of this study it can be asserted that Payments, Deposits & Loans, and Exchanges are the dominant business models in the DeFi sphere. When it comes to risks, coding errors, fraud, flawed oracles, illicit activities, and regulatory uncertainty play the most important roles.

Furthermore, and resting on the entire body of scientific literature on DeFi as on the empirical survey conducted among DeFi experts, DeFi can be defined as a paradigm applied to create and deliver financial services first and foremost in a borderless, decentral, verifiable and secure fashion, yielding business models such as Payments, Deposit and Loans as well as Exchanges, but also exposing stakeholders to risks predominantly resulting from coding errors, fraud, flawed oracles and illicit activities.

These findings also correspond with the results of a literature survey more recently conducted by Ozili [96] who identified inclusive, permissionless, disintermediated, immutable, censorship-resistant, efficient, and borderless as prominent characteristics of DeFi as well as execution risks, data theft, interconnectedness, and external data risk as salient risks. Ozili [96] furthermore pointed out legal liability as another important risk. Being a topic that received increasing attention in recent times, it had not yet received significant attention in the literature that served as a basis for the study at hand. Besides, the study results resonate with the findings of Gangwal, Valluri [97] who established in a recent study that oracle information fed to the requesting party deviates considerably from the data reported by the sources.

What is noteworthy, however, is that there is a substantial difference between the presumed importance that scientists attached to specific characteristics, business models, and risks in the past and the ones now perceived by the surveyed group of experts. Most notably the characteristics “disintermediated”, “transparent” and “interoperable” are not deemed as important by the group of experts as literature had suggested whereas the business models Derivatives and Origination and Issuance dropped in prominence. The risks of regulatory uncertainty, immature technology, and lack of knowledge also dropped majorly in emphasis. The latter trend could well be explained by the fact that not only have regulations evolved around DeFi, but that technology has also matured and knowledge has been built up in the meantime.

Furthermore, this study also yields an important methodological insight. Taylor and Laver [98] suggested that literature reviews can be interpreted as ‘expert judgments ‘. This study indicates that there are indeed some overlaps among the group of scientists publishing on DeFi, yet that are also differing opinions among that group of scholars and the heterogenous group of DeFi experts comprising practitioners as well as academics. Scholars are thus not always to be seen as the only experts existing.

6 Conclusion

By identifying the most foundational characteristics of DeFi, the business model predominantly built on the DeFi paradigm, and the risks entailed, this paper contributes not only to an improved understanding of the DeFi ontology but also helps researchers to communicate more efficiently on that topic due to the avoidance of misunderstandings and it helps practitioners to better navigate the opportunities and risks inherent to that topic.

6.1 Limitations and future research avenues

No research comes without limitations; hence it is important to recognize the key constraints of this paper. One of the main limitations of this research arises from its sample. Due to the qualitative nature of the study, a convenience sample was used yielding a rather small sample size. Hence, despite the sound interrater reliability of the data polled, no conclusions can be drawn about the external validity of the sample. Moreover, the small sample size also limited the statistical methods to purely descriptive ones. Future research may therefore sample a representative population and apply structural equation modeling to obtain richer causal insights from the parametric and non-parametric data gathered [99]. Alternatively, a sufficiently large and representative sample could be used to produce inferential estimates using hierarchical regression analysis.

Another possible avenue for future research is to investigate identical research questions yet apply a longitudinal approach. The study at hand is limited as it only gathered a snapshot of empirical data. Hence a repetition of the study incorporating newly generated knowledge could be of interest, for instance including legal liability as a main risk in DeFi settings. Moreover, a longitudinal study could be highly interesting to analyze how expert feedback on DeFi changes over time.

Besides, the study at hand used semantic scales that lack an objective anchor for measuring the construct in questions as is often the case with perceptual items [100, 101]. Accordingly, the respondents may have varied notably in their perceptions of the items survey. Being of a predominantly qualitative nature the study furthermore yields purely descriptive figures which can merely be interpreted in an ordinal fashion. Investigating the research questions using objectively measurable data and analyzing this data using computer-assisted qualitative data analysis software could therefore be another future research avenue.

6.2 Contribution

Despite its limitations, this study has contributed by examining the entire body of extant scientific literature on the topic of DeFi. At the time of conducting the study no such published work existed. Using an expert survey, the study further refined these purely theoretical findings by adding the insights of practitioners. In this way, it distilled the most important characteristics of DeFi, the key business models that are built around these characteristics, and it identified the major risks that DeFi may pose to its users.

By bringing to light DeFi’s inherent nature—its ability to operate beyond borders, its decentralized control, and its commitment to transparent transactions—the study highlights the ground-breaking potential DeFi holds for reshaping finance on a global scale. These identified characteristics paint a picture of a financial system that breaks away from traditional confines, democratizes access to financial services, and builds a foundation of trust through transparency.

The study’s insight into DeFi’s business models underscores the sector’s innovative spirit. From enabling global payments without middlemen to pioneering new forms of lending and borrowing, DeFi is setting the stage for a financial ecosystem that’s not only more efficient but also more inclusive.

However, the examination of DeFi’s associated risks, particularly coding errors and fraud, serves as a cautionary note. It highlights the need for stronger security measures and thoughtful regulation to navigate the challenges inherent in this new technology.

In other words, this investigation has opened up the DeFi box and examined the pieces that it contains. The characteristics of the building blocks have been identified and it has been established which models are most often built from these bricks. Moreover, the pitfalls that come along when building those models have been investigated.

This theory-building piece of work has thus addressed a significant gap in the body of knowledge on DeFi by scientifically identifying its characteristics, the business models it facilitates, and the risks that it imposes. It has established the current scholarly status quo, verified it with a circle of international experts, and derived a taxonomy along with a concise definition of DeFi which can be used by academics and practitioners alike.