Token governance in initial coin offerings: Implications of token retention and resale restrictions for ICO success

Abstract

The high degree of asymmetric information in initial coin offerings (ICOs) increases the probability of moral hazard, adverse selection, and outright fraud. Token governance mechanisms may help mitigate such investment risks. Using novel data on token retention in the primary market and resale restrictions in the secondary market, we estimate the impact of these governance mechanisms on the fundraising success and post-funding performance of ICOs. First, we estimate the valuation elasticity of token retention, indicating that 10% fewer tokens sold increases the ICO funding amount by 3.2%. Second, restricting ICO investors’ ability to resell tokens in the secondary market has a detrimental impact on the 12-month buy-and-hold abnormal return. We also discuss contingency effects of the specific implementations of these governance signals and show that the effects are moderated by the quality of the venture’s human capital.

Plain English Summary

The initial coin offerings (ICO) market has faced the issue of bad entrepreneurs and outright fraud since its infancy due to the high level of information asymmetry. We show that the share of tokens retained by the venture is valued as a quality signal by investors to identify good entrepreneurs, leading to an increased funding amount. However, restricting the reselling of the retained tokens in the secondary market has surprisingly a negative impact on the post-funding return. These findings provide clear implications for research, education, and practitioners. Our research contributes to the advancement of the general understanding of the cryptocurrency market and helps to fill the research gap on factors that influence the abnormal returns of token offerings. Entrepreneurs should keep in mind that not all token governance methods are linked to greater fundraising or positive financial returns when creating the offering’s structure.

1 Introduction

The structure and market characteristics of initial coin offerings (ICOs) are characterized by a high degree of information asymmetry between investors and entrepreneurs (Adhami et al., 2018; Fisch, 2019; Howell et al., 2020; Chod & Lyandres, 2021; Momtaz, 2021a, b). Token offerings have long been unregulated in many jurisdictions (Bellavitis et al., 2021, 2022; Huang et al., 2020), with the absence of reporting requirements exposing investors to pronounced risks related to adverse selection, moral hazard, and outright fraud (Momtaz, 2021a; Hornuf et al., 2022, 2023). ICO ventures often lack proven track records and are very early-stage (Howell et al., 2020; Fisch, 2019; Momtaz, 2020). The high degree of disintermediation in the ICO market burdens individual investors with the due diligence, potentially leading to suboptimal information production outcomes in the aggregate (Giudici & Adhami, 2019; Momtaz, 2021a; Cumming et al., 2023). As a result, the contractual relationship between ICO ventures and investors is characterized by (i) complete power of the venture over the contract creation, (ii) no negotiation between both parties, and (iii) investors’ practical inability to enforce legal rights in case the venture is in breach of disclosed investment terms. These characteristics underline the importance of the design of digital assets to de-risk ICOs.

However, the existing literature largely neglects the governance mechanisms of tokens sold in ICOs. Exceptions are studies focusing on the ICO fundraising effects of retained tokens, measured as the percentage of tokens either not distributed (Davydiuk et al., 2023; Giudici & Adhami, 2019; Lyandres et al., 2022) or allocated to insiders (Bourveau et al., 2022). By implication, the resulting research gap includes, inter alia, a lack of knowledge about the anatomy of token governance in ICOs, the effect of token governance on post-funding financial performance, and the moderating effects of human capital on the token governance–performance relationship.

To address this research gap, we explored ICO whitepapers in an inductive way, compiling all token governance mechanisms that ICO ventures might disclose. This exercise resulted in two overarching token governance mechanisms: (i) token retention mechanisms and (ii) token restriction mechanisms. Token retention mechanisms encompass various purpose-bound token beneficiary categories (i.e., stakeholders, venture development, and reserve pool). The disclosed share of tokens reserved for different groups in the primary market allows the venture to communicate different types of goals. First, tokens retained for the founders, the venture team, and other internal and external stakeholders can be summarized as stakeholder allocations and are comparable to insider holdings in traditional financial markets. Allocations to this group aim to foster the long-term focus of the decision-makers and to reduce agency costs for the investors (Leland & Pyle, 1977; Vismara, 2016). Second, venture development retentions include all tokens retained with the main purpose of supporting future growth and operations of the venture, such as research, business development, or company allocations. This group communicates a commitment by the venture to longer-term growth. Third, tokens retained to support the ecosystem through future (incentive) distributions or other stability mechanisms can be summarized as reserve pool retentions. This group includes a variety of disclosed categories, such as reserves, mining pool, rewards pool, and other related categories, and might increase investor confidence in the underlying project. However, ventures also face significant drawbacks based on the retained tokens. Ventures are obligated to retain the token for the disclosed purpose, which leads to decreased flexibility and liquidity of the tokens.

Token restriction mechanisms prohibit the unrestricted transfer of a part of or all tokens in the secondary market. The restriction disclosures in the ICO whitepapers are based on four different types of restrictions: (i) cliff, (ii) lock-up, (iii) vesting, and (iv) reverse vesting. While a cliff prohibits the initial token issuance to the applicable group before the cliff date, a lock-up is preceded by a completed token issuance and prohibits any sales of the applicable tokens before the lock-up end date. Vesting and reverse vesting are both gradually released restrictions, realizing the token allocation and lifting the resale restriction gradually over a defined period, respectively. The implementation of token restriction by the venture amplifies the previously described benefits and drawbacks of token retention. Interestingly, unlike traditional venture capital-backed startups that may also base incentive alignment mechanisms on board seats or voting rights (Cumming et al., 2019), ICO ventures are unique in that their incentive alignment efforts are almost comprehensively described by token allocation and token retention mechanisms.

With the institutional understanding of token governance in ICOs in hand, we derive two overarching hypotheses. The first hypothesis posits that a token allocation resulting in a higher degree of retained tokens in the primary market positively impacts the ICO venture’s funding and the post-funding performance. We refer to the hypothesis as the token retention outperformance hypothesis (TRetO). Leland & Pyle (1977) submit that, due to the implied costs of retaining equity, only good entrepreneurs are expected to retain tokens, leading to a “separating equilibrium” (Spence, 1973), and hence the number of tokens retained reveals the fair value and quality of the underlying venture. Prior research also shows that retained tokens are expected to increase the operational performance of the venture (Davydiuk et al., 2023; Lyandres et al., 2022). The liquid secondary market for ICO tokens provides an ideal setting to analyze the impact of token retention on the financial post-funding performance, measured as the 12-month buy-and-hold abnormal return. We expect that the token price reflects the market expectation about the future financial success of the underlying venture, which should be more positive for ventures retaining more tokens (Leland & Pyle, 1977), and that the forecasted superior operational performance translates to positive financial returns.

The second hypothesis posits that token restrictions that limit the reselling of tokens in the secondary market positively impact ICO ventures’ funding and the post-funding performance. We refer to the hypothesis as the token restriction outperformance hypothesis (TResO). We expect the restriction of retained tokens might also lead to a “separating equilibrium” (Spence, 1973). The restriction choice might be interpreted as a signal about the quality of the underlying venture by potential investors (Leland & Pyle, 1977), mainly due to the voluntary exposure of entrepreneurs to the financial downside and the risk mitigation for investors that the entrepreneurs are selling their retained stakes immediately after token listing. Brav & Gompers (2003) provide empirical support for this theory in the IPO context, but prior research could not identify a clear relationship in other types of fundraising (Davydiuk et al., 2023; Bourveau et al., 2022). Lastly, our paper analyzes the moderating effect of human capital quality signals, such as the team size of the venture, on the effect of token governance signals. We find a supportive theory for both directions of a possible moderating effect. The moderating effect might be negative due to a possible substitution effect of the two signals, given that both signals have related underlying factors; the moderating effect could also be positive given that the retained or restricted tokens should have an increased value when the venture has higher human capital quality.

To assess our hypotheses, we derived a baseline sample of 759 token offerings with a start date before 2021 that included token governance details within their whitepaper. We hand-collected all related data points regarding token governance and restrictions and applied an econometric approach with a baseline regression to estimate the relationship between the token governance variables and fundraising success and post-funding performance under the consideration of a vector of different control variables. Additionally, we applied different two-stage models to control for observed and/or unobserved heterogeneity (e.g., Bertoni et al., 2011; Colombo & Grilli, 2010) and conducted robustness checks of the results with an expanded sample of 986 observations to correct for a potential sample selectivity bias.

The empirical results confirm our TRetO hypothesis. A 1% increase in the share of retained tokens leads, on average, to a 0.318% increase in the fundraising amount. We find additional evidence that the effect is mainly caused by the signal of tokens allocated to stakeholder groups rather than venture development or reserve pool allocations. Our regression models confirm that the increased firm valuation is based on the treatment effect of token retention rather than being confounded by a selection bias. We do not find supporting evidence for a positive relationship between token retention and post-funding financial performance; hence, our results reject the TResO. However, we find empirical evidence for a significant negative relationship between token restriction and post-funding financial performance. The disclosure of a token restriction reduces the 12-month buy-and-hold abnormal return on average by 37.6%, with a p-value below 5%. Our findings suggest that the effect is mainly driven by the impact of the disclosure of an initial restriction rather than (reverse) vesting disclosures. The disclosure of an initial restriction leads, on average, to a 38.7% decrease in the 12-month buy-and-hold abnormal return. We believe that the negative is based on investors’ concern about the restricted flexibility of the venture to unexpected liquidity needs and breaches of the restriction disclosures by the venture (Cohney et al., 2019).

Finally, we find evidence that the team size of the venture moderates the effects of token governance tools, described above. The 0.757% increase in fundraising amount based on a 1% increase in retained tokens is reduced, on average, by 0.039 percentage points per additional full-time employee (FTE) added to the venture team. Our results show a positive moderating effect of human capital on the relationship between token governance tools and the financial post-funding performance. We find evidence that the negative impact of the token restriction disclosure on post-funding performance decreases, on average, by 5.9 percentage points for each additional FTE added to the venture team.

Our study expands the literature on the design of digital assets and analyzes token governance in a novel and more granular way than existing empirical studies. To our knowledge, we are conducting the first research on token governance’s impact on financial post-funding performance and provide the first evidence of how token governance signals are moderated by human capital quality. Thus, our research contributes to the advancement of the general understanding of the cryptocurrency market and helps to fill the research gap on factors that influence the abnormal returns of token offerings. Uncovering several practical implications for entrepreneurs and investors, we show that the direction and strength of the token governance signals depend on the chosen incentive alignment tool and its defined characteristics. Ventures designing the structure of the token offering must also consider the present moderating effect of human capital quality signals on the effect of the included token governance mechanisms to maximize the probability of reaching the fundraising targets and achieving positive post-funding financial performance.

The remainder of the paper is structured as follows. Section 2 describes the institutional and theoretical background and derives testable hypotheses. Section 3 provides details on our empirical setting, including the data collection process, a description of the variables, and the econometric approach. Section 4 presents our empirical results of the main regressions and additional robustness tests. Finally, Section 5 discusses our theoretical contribution and practical implications, highlights limitations, and offers potential avenues for future research.

2 Background and hypotheses

2.1 Asymmetric information in initial coin offerings

Crowdfunding targets the disintermediation of the fundraising process to avoid payments to intermediaries. Individuals typically contribute small amounts of money via the Internet, which are then pooled to support a specific objective (Ahlers et al., 2015). The magnitude and type of objective and hence the structure of crowdfunding can take various forms (Schwienbacher & Larralde, 2010; Mollick, 2014). Based on blockchain technology, a further reduction of the degree of intermediation is possible (Chen et al., 2021; Kher et al., 2021). Token offerings apply smart contract technology that allows the automated execution of pre-defined contracts, which leads to increased efficiency and ultimately decreased costs for the venture in comparison to other fundraising types (Fisch et al., 2022). However, the flip side of disintermediation is that crowdfunding markets are often characterized by high levels of asymmetric information due to the absence of trusted intermediaries (Vismara, 2016), limited due diligence possibilities (Agrawal et al., 2015), inexperienced investor profiles (Ahlers et al., 2015; Boreiko & Risteski, 2021), and little opportunity for interaction between issuer and investor (Vismara, 2016). The asymmetric information issue in the context of token offerings can be amplified for various reasons, including: (i) the underlying projects in an ICO are often early-stage with inexperienced entrepreneurs (Howell et al., 2020; ii) initial coin offerings (unlike, e.g., initial exchange offerings) typically do not involve independent platforms that could potentially certify project legitimacy and quality (Giudici & Adhami, 2019; Momtaz, 2021a; Schueckes & Gutmann, 2021; iii) ICOs are unregulated in many jurisdictions (Bellavitis et al., 2022); and (iv) investors’ payments in exchange for the tokens do not require any intermediating party (e.g., payment agent) (Giudici & Adhami, 2019)). Momtaz (2021a) has additionally highlighted the technology character (e.g., DLT) of the underlying ventures in ICOs as a reason for asymmetric information in the context of token-based crowdfunding. Investors may be required to have a certain level of technological knowledge to be able to perform an appropriate assessment of the venture (Cohney et al., 2019).

The market characteristics and resulting information asymmetry described above portend several consequences for token investors (Alshater et al., 2023; Chod et al., 2020). First, the information asymmetry prior to contract initiation might lead to adverse selection (Chen, 2019). Low-quality entrepreneurs will aim to promote their projects as high-quality, leading to an upward-biased valuation by potential investors (Vismara, 2018). Second, moral hazard might occur after investors provide funding to the entrepreneurs. Momtaz (2021a) argues that the lack of possibilities to verify signals ex-ante and to punish biased signals incentivizes the entrepreneurs to bias signals of the underlying venture quality. Third, the low level of regulation and the elimination of trusted intermediaries, especially in the context of payment, amplify the potential fraud risks in token offerings (Barone, R., & Masciandaro, 2019). A possible solution for the suppliers of finance to ensure that their respective funds are not utilized for unfavorable undertakings or embezzled would be contracts between them and the entrepreneurs or managers of the companies (Shleifer & Vishny, 1997). However, the venture has complete power over the contract creation, and no negotiation takes place between the parties. Additionally, investors usually do not have the opportunity to assert any legal rights in case the entrepreneurs violate any of the disclosed investment terms.

Good token governance mechanisms are required to mitigate these risks and prevent market failure. Token governance addresses the options of how suppliers of finance ensure a return on their invested capital (Shleifer & Vishny, 1997). The underlying objective of all approaches is to align the interests of the agent with those of the principal and therefore limit potentially deviating activities (Jensen & Meckling, 1976). While these objectives are identical to those in the traditional financial markets, the unique characteristics of the token-based crowdfunding market and the expected sensitivity of investors to incentive alignment tools make it a particularly interesting environment for studying the impact of token governance on investor behavior. Prior investigations related to governance mechanisms in IPOs (Brav & Gompers, 2003; Field & Hanka, 2001) are conducted in an environment that strongly deviates from that of ICOs. Investors can usually rely on certain exchange rules that the companies must apply if they want to conduct an IPO. These rules provide not only regulatory clarity for the investor but usually also guarantee certain company characteristics, such as minimum size, revenue, or management quality. Additionally, the IPO process usually includes several third-parties (e.g., investment banks, payment agents, etc.) and marketing events (e.g., IPO roadshow), providing investors additional information and the option to interact with company representatives. These IPO market characteristics sharply reduce the information asymmetry during the investment and limit the transferability of the results to the ICO context, underpinning the importance of our study. In Section 2.2, we provide a detailed analysis of the governance mechanisms in the token-based crowdfunding market and further elaborate on the differences in the deal characteristics among different types of fundraising.

2.2 Token governance in initial coin offerings

We focus on utility token offerings, for which most of the traditional governance tools (e.g., voting rights, board seats) are not available (Alshater et al., 2023; Block et al., 2021). The contractual power between ICO ventures and investors lies fully in the hands of the entrepreneurs, who define the terms of the investment contracts before the ICO launch. There is usually no direct negotiation between the ICO venture and its potential investors. The small minimum investment size, combined with the absence of specific investor requirements, such as a minimum income or wealth, and the convenience of online platforms, provides a large global base of potential investors, empowering the venture at the expense of a dispersed investor base (Cumming et al., 2021; Schueckes & Gutmann, 2021). There is also no regulatory requirement to disclose information, making the provision of information fully voluntary (Momtaz, 2021a; Bourveau et al., 2022). These ICO deal characteristics strongly deviate from the characteristics of an IPO. While we showed above that ICO investors have mostly no legal rights and the only information source is the voluntarily published whitepaper by the venture, IPO investors usually have strong legal rights attached to their investment and can rely on a standardized and compliant IPO prospectus. These differences provide further support for the importance of our study in investigating the impact of token governance mechanisms in the token-based crowdfunding market.

The low contractual power of investors and the unregulated environment could result in a full waiver of token governance mechanisms in initial coin offerings. This would require investors to trust solely in the good intentions of the entrepreneurs. In practice, however, the token issuer faces a trade-off between the costs of token governance commitments and the attractiveness of the offering to investors. If the venture implements incentive alignment tools, it must bear the inherent costs of the chosen type of tool but would potentially increase the attractiveness to investors. On the other hand, if the venture fully removes token governance approaches, it would be less restricted in governing the company and would not be exposed to additional costs; however, this could make the investment less attractive and reduce fundraising. As explained below, our comprehensive manual analysis of all available ICO whitepapers revealed two types of token governance mechanisms: (i) token retention and (ii) token restriction.

2.2.1 Token retention

The token allocation split indicates the number of tokens reserved for different groups in the primary market, providing more extensive information than the sole disclosure of distributed shares in the equity crowdfunding context (Vismara, 2016). Aside from the tokens reserved for allocation to the public (e.g., pre-ICO and ICO sales), there are three main categories: (i) stakeholders, (ii) venture development, and (iii) reserve pool. The sum retained by these groups indicates the total amount retained by the venture in the signaling sense described by Leland & Pyle (1977).

In addition, each individual group has its own purpose with different benefits.

• The stakeholder allocation is comparable to the insider holdings in traditional financial markets and other fundraising types. Allocations to founders and other stakeholders, such as advisers or partners, aim to reduce agency costs and foster the long-term focus of the decision-makers.

• The venture communicates a commitment to longer-term growth through allocation to venture development purposes, such as research or business development.

• The reserve pool is usually a side pool to support the ecosystem of the issued token through (future) incentive distributions and provide stability to the issued token. Allocations falling into this category could increase the confidence of potential investors in the underlying project and the corresponding token.

Token retention within these three categories is one of the two key token governance mechanisms in ICOs. The venture also faces several drawbacks from token retention, such as (i) decreased flexibility if the allocated tokens can be solely used for their disclosed purpose and (ii) decreased liquidity of the token since the publicly distributed portion is reduced, ultimately reducing the number of free-floating tokens.

Fig. 1

Token Allocation and Restriction - SwissBorg (2017)

2.2.2 Token restriction

Token restriction refers to the decision of the venture to prohibit the unrestricted transfer of a part of or all issued tokens in the secondary market. Typically, the ICO whitepaper describes the restriction and the category of tokens it applies to. Even though the restrictions are typically only applicable to one specific group, several types of restrictions are often applied to different categories of tokens within one token offering.

The most common allocation in an ICO that is restricted with one of the types shown below is the one to stakeholders (e.g., Bourveau et al., 2022; Davydiuk et al., 2023), in line with the insider share restrictions during an IPO or venture capital deal. There are four key terms used in the context of token restrictions: (i) cliff, (ii) lock-up, (iii) vesting, and (iv) reverse vesting.

• Cliff: A cliff prohibits any token issuance to the applicable group before an indicated date (known as the cliff date) and is often followed by a vesting period.

• Lock-up: A lock-up prohibits any sales of the issued tokens by the applicable group before an indicated date (the lock-up end date) and is often followed by a reverse vesting period.

• Vesting: The issuance of the allocated tokens to the applicable group is gradually realized over a defined time period (the vesting period).

• Reverse vesting: The issued tokens to the applicable group contain a resale restriction that is gradually lifted over a defined period (the reverse vesting period).

For cliffs and lock-ups, the initial restriction is typically fully lifted on a predetermined date in the future, while vesting and revere vesting restrictions are released more gradually. Generally, the group of restrictions (cliff/lock-up vs. vesting/reverse vesting) is clearly identifiable based on the information in the whitepaper. Sometimes a combination of the different restriction types is used.

One the one hand, the implementation of token restrictions as a token governance tool in ICOs targets the amplification of the previously described benefits of token retention. Ventures seek to align the interests of the entrepreneurs and investors to reduce agency costs. The restriction of the retained tokens could provide incentives to grow the company and keep a long-term focus, as well as show confidence in the project.

On the other hand, the prohibition of receiving and selling the tokens on the market also amplifies the drawbacks of token retention. The venture further limits its flexibility to react to new information over the period of the restriction, such as a change in the market environment or other developments that would require a reallocation of financial resources. In addition, the token’s liquidity will decrease during the restriction period, and the stakeholders holding restricted tokens will face high opportunity costs since they cannot liquidate their holdings (Cummings et al., 2020). Another important drawback is the need for investors to have proof that the ventures are adhering to the restriction. Unlike in regulated markets, the unregulated ICO market typically bears no consequences for breaches of any of these restrictions.

Figure  1 shows an example of the token governance mechanisms disclosed in the whitepaper of SwissBorg (2017). Investors can retrieve information about four different groups, together with the respective percentage of tokens allocated. Additionally, token restrictions are explained in text form below the chart. SwissBorg (2017) discloses that the tokens allocated to the group “Team and Advisors” are subject to a four-year distribution period (i.e., vesting period), which starts directly after the ICO.

2.2.3 A comparison between governance in initial coin offerings and in other financing methods

Table 1 provides a summary of the differences and similarities among the different types of fundraising. Token offerings share some common features with crowdfunding and traditional fundraising types, such as initial public offerings and venture capital, while also having many unique characteristics (e.g., Block et al., 2021; Howell et al., 2020; Momtaz, 2020). The table is based on Momtaz (2020) and has been further expanded with the identified information regarding the governance mechanisms of the different fundraising types. Table 2 summarizes the presented key terms in token governance and provides an overview of the corresponding definitions.

Table 1 Comparison of ICOs to crowdfunding, venture capital, and initial public offerings

Table 2 Token governance mechanisms: key terms

2.3 Hypotheses

The theoretical foundation of signaling in the context of informational asymmetries in financial markets is given by Leland & Pyle (1977). Due to the implied cost of equity retention or token retention in the case of ICOs, only “good” entrepreneurs are expected to retain tokens. This is based on the fact that good entrepreneurs believe in the quality and success prospects of the venture. Bad entrepreneurs, on the other hand, would decide to sell a larger share of tokens to outside investor because they are aware of the venture’s low quality. The immediate sale of all tokens would maximize the entrepreneurs’ return. This leads to a “separating equilibrium” (Spence, 1973), whereby the number of tokens retained reveals the quality and, ultimately, the fair value of the underlying venture.

Downes & Heinkel (1982) provide empirical evidence for Leland and Pyle’s (1977) conjecture in the context of initial public offerings that a firm in which entrepreneurs retain high fractional ownership does indeed have a higher value, based on a large sample of unseasoned new issues. Additional studies provide further evidence for the positive relationship between equity retention as an incentive alignment mechanism and firm value due to the mitigation of principal–agent issues (e.g., Ritter, 1984; Jensen & Meckling, 1976). The theory of signaling also finds application in several fields of entrepreneurial finance, including venture capital (e.g., Busenitz et al., 2005; Conti et al., 2013) and crowdfunding (e.g., Anglin et al., 2018; Barbi & Mattioli, 2018). Prior research shows that the size of investment and equity shares held by the venture team has no significant effect on venture capital funding success (Busenitz et al., 2005; Conti et al., 2013)). Conti et al. (2013) provide evidence that business angels, on the other hand, do consider entrepreneurs’ “skin in the game” as an important value signal of the firm. The increased attraction of business angel investments due to the higher commitment of the founding team is in line with the research of Cassar & Friedman (2009), which shows that entrepreneurs with a greater investment in their ventures are expected to have higher confidence in their ability to perform the related tasks (i.e., entrepreneurial self-efficacy).

Ahlers et al. (2015) and Vismara (2016) provide empirical support that the share of equity retained in an equity crowdfunding campaign is interpreted by investors as a signal of the future success of the venture, in line with the theoretical foundation of Leland & Pyle (1977). Ahlers et al. (2015) argue that the level of uncertainty of investors, driven by the level of reliance on ambiguous information, is negatively related to fundraising success. The study provides evidence that an increased share of equity retention is a key factor in reducing this uncertainty, ultimately leading to an increased funding amount, an increased number of investors, and a faster completion time of the funding project. These findings are confirmed by Vismara (2016), showing that a larger amount of equity offered to outside investors during fundraising via equity crowdfunding decreases the number of participating investors and the total amount of capital raised.

2.3.1 Token retention and the success of token offerings

Our first hypothesis, the token retention outperformance hypothesis (TRetO), relates to the impact of a higher degree of retained tokens in the primary market on ICO ventures’ funding and post-funding performance. Investors in the token-based crowdfunding market are dependent on signals of the quality of the underlying venture to make their assessment before investing. Based on the high level of information asymmetry between insiders (e.g., entrepreneurs and the management team) and outside investors, incentive alignment mechanisms could be particularly valued by potential investors. The importance of these signals in the context of ICOs may be further supported by at least three of the market characteristics described above. First, a product is often unavailable due to the early stage of the underlying venture or project. Information about actual product or service tests is often limited, which makes the assessment of product–market fit highly speculative. The missing information requires investors to search for alternative signals to forecast the company’s future success. Second, no financial data about the underlying venture is available. This is a common issue in entrepreneurial finance, given that the companies receiving investment are in their early stages. Nevertheless, the absence of a reporting standard further limits the availability of quantitative data for assessing ICOs, compelling investors to seek out qualitative signals in conducting their due diligence. Third, the market is mostly unregulated, increasing the probability of disclosure of wrong information and other fraudulent behavior by the firm. A high number of fraud cases in token-based offerings are observable (Andrieu & Sannajust, 2023), which led to a total write-off of the invested capital. Investors may thus be especially sensitive to incentive alignment signals, such as token retention, which would put the own money of the entrepreneurs at risk in case of dishonest actions.

Davydiuk et al. (2023) examine the impact of the share of tokens that cannot be sold directly to outside investors on, among others, the natural logarithm of the funding amount raised and the ratio of the funds raised during the ICO to the defined hard cap. The outcome of the regressions shows a highly significant positive relationship between the token retention variable and the two different success metrics. The prior research of Giudici & Adhami (2019) and Lyandres et al. (2022) has shown a significant negative relationship between the share of distributed tokens during the ICO and the success probability and fundraising amount of the token offering. Despite the different approaches to defining and calculating token retention, we expect our study to confirm a positive relationship between the share of retained tokens and the funding amount based on the following rationales. First, we expect investors to interpret token retention as a quality signal based on the theory of Leland & Pyle (1977) and the high level of information asymmetry in token offerings. Second, we identify stakeholder allocation as one of the three main categories of token retention. Prior research in the context of alternative early-stage fundraising types, such as equity crowdfunding, provides empirical support for a positive relationship between the tokens that are allocated to insiders and the amount raised during the ICO (Conti et al., 2013).

In public equity markets, it has been observed that insider stock retention, in addition to its positive signal during the initial public offering, can have a positive impact on the firm’s post-funding performance and valuation (e.g., Bhagat and Bolton, 2008; Gompers et al., 2003; McConnell et al., 2008). Alternative fundraising methods, such as equity crowdfunding, often lack liquid secondary markets, making a comparable analysis challenging (Lukkarinen & Schwienbacher, 2023; Hornuf & Schwienbacher, 2018). However, ICOs provide the required setting for an empirical study about the relationship between token retention and post-funding success. The existence of a liquid secondary market for tokens issued via initial coin offerings (Adhami et al., 2018) allows us to calculate a post-funding performance return of the tokens, such as the 12-month buy-and-hold abnormal return, and examine its variation based on the disclosure of different token retention amounts. We believe that the described effects of equity retention in the public markets persist in the ICO context.

The separating equilibrium (Spence, 1973) based on the signaling theory of Leland & Pyle (1977) would imply that the entrepreneurs who retain more tokens have superior skills to the ones who retain fewer. Therefore, these ventures should have increased success prospects based on stronger operational performance, which should ultimately translate to superior financial performance of the listed tokens. Despite conducting the first research on token governance’s impact on financial post-funding performance, we can partially build our theory on the findings of prior studies related to operational post-funding performance in the ICO context. Davydiuk et al. (2023) analyze the relationship between the share of all issued tokens of the venture that cannot be sold to outside investors and the probability that an issuer has (i) an active website, (ii) a live product or platform, and (iii) an application available for download in the Apple App Store, at least six months after token listing. The empirical results show positive significant results in all three regressions and imply a positive relationship between the fraction of retained tokens and successful product development. Additionally, Lyandres et al. (2022) have found empirical evidence that an increased percentage of tokens distributed during the token offering has a significant negative impact on the social media activities of the venture, number of code revisions on GitHub, product/platform adoption measured based on the number of wallets containing the issued token and the post-ICO growth in the cumulative number of on-chain transfers of the respective token across wallets. These findings provide further support for a positive relationship between token retention and the post-funding success of the venture.

Hypothesis 1a: The relationship between token retention and the funding amount in token offerings is positive.

Hypothesis 1b: The relationship between token retention and the post-funding performance (12-month BHAR) in token offerings is positive.

2.3.2 Token restriction and the success of token offerings

Our next hypothesis, the token restriction outperformance hypothesis (TresO), pertains to how the ICO ventures’ funding and post-funding performance is impacted by reselling restrictions of tokens in the secondary market. Investors face the risk that entrepreneurs could sell their retained stakes immediately after token listing. This risk is more pronounced than in other types of fundraising for various reasons. First, due to the unregulated environment, the venture has no legal requirement to report any of its transactions or current holding split. This differs strongly from equity retention in public markets, where the regulator usually requires the reporting of transactions and holdings of insiders. Second, a liquid market for the retained tokens exists in many cases, which would allow an immediate sale. This would not be the case for other early-stage fundraising types. These risks are partially mitigated given the public nature of the blockchain, where all investors have access to the transaction history and can identify any insider transaction. However, this requires the know-how to be able to read primary blockchain data. A token restriction implies the voluntary commitment of the venture to hold a part of or the whole amount of retained tokens for a longer defined time period, alleviating the risk of entrepreneurs selling their retained tokens immediately after listing. The restriction of retained tokens is expected to lead to a “separating equilibrium” (Spence, 1973), whereby the decision to restrict the retained tokens reveals the quality of the underlying venture. This impact originates from the sharply increased implied costs for restricted tokens. Entrepreneurs are prohibited from freely transacting with the token, not only increasing the opportunity costs but also leading to a direct negative financial impact in case the venture has no success in the future. Therefore, the decision to voluntarily restrict the issued tokens might provide a quality signal about the underlying venture to potential investors (Leland & Pyle, 1977).

Even though empirical support for this theory exists in the IPO context (Brav & Gompers, 2003), the relationship in other types of fundraising is less clear. While (Davydiuk et al., 2023) show a significant positive effect of the inclusion of a vesting schedule for ICO tokens on the fundraising amount, Bourveau et al.’s (2022) empirical study only confirms a positive relationship between the token vesting disclosure and the probability that the token offering reaches the soft cap requirement. The latter study could not find a significant effect of the token vesting disclosure on the funding amount. The discrepancy between the findings in Davydiuk et al. (2023) and Bourveau et al. (2022) may stem from at least three factors. First, the underlying samples contain different token offerings. Second, Bourveau et al.’s (2022) study includes a larger set of control variables than that of Davydiuk et al. (2023). The results of Bourveau et al. (2022) show significant effects of additionally included variables on the fundraising amount. Third, Bourveau et al. (2022) included additional model specifications, such as the inclusion of country/region and industry indicators, which are not included in Davydiuk et al.’s (2023) model. For our study, this means the results of the relationship between token restriction and funding amount may be highly sensitive to the included control variables and other model specifications.

To our knowledge, our study is the first to also examine the impact of token restrictions on the post-funding token price performance. We show that, in the ICO context, the entrepreneurs have an information advantage compared to the investors. Entrepreneurs who decide to restrict their retained tokens are voluntarily exposing themselves to the adverse token performance during the restriction period and provide an important quality signal to investors (Leland & Pyle, 1977). Therefore, in line with our explanation for the TRetO hypothesis, we assume that entrepreneurs who implement a token restriction should strongly believe in the future success of the underlying venture and have superior skills to those who do not restrict the issued tokens. This assumption is expected to lead to a positive abnormal return among ventures that disclosed token restrictions in their whitepaper.

Hypothesis 2a: The relationship between token restrictions and the funding amount in token offerings is positive.

Hypothesis 2b: The relationship between token restrictions and the post-funding performance (12-month BHAR) in token offerings is negative.

Of course, it is possible that token restrictions impact the post-funding financial performance negatively for a number of reasons. First, the positive effect of token restrictions could already be priced in. If the initial funding valuation already includes the full positive price effect of the token restriction, there might be no further positive impact on post-funding performance. Second, if the venture decides to restrict the retained tokens, it also restricts its flexibility to react to unexpected liquidity needs or market environment changes, which might potentially negatively impact the operational performance. Third, investors face the risk that the restriction disclosures in the whitepaper are not incorporated into the source code (Cohney et al., 2019). Every post-funding experience that calls this accuracy into question might negatively impact the trust of the investor in the project and ultimately lead to increased selling of the token.

2.3.3 Moderator: team size

Finally, we examine the moderating effect of the venture’s team size on the relationship between the token governance signals and the ICO ventures’ funding and post-funding performance. Information about the founding and management team has been proven to be an important decision factor not only for venture capitalists (Gompers et al., 2020), angel investors (Bernstein et al., 2017), and equity-crowdfunding investors (Ahlers et al., 2015; Piva & Rossi-Lamastra, 2018; Barbi & Mattioli, 2018), but also for investors in token crowdfunding (e.g., Momtaz, 2020; Giudici & Adhami, 2019). The average ICO investor considers the size of the venture team as a quality signal of the project, which ultimately leads to a positive relationship between the team size and fundraising amount in token offerings (Roosenboom et al., 2020).

There are several reasons why the moderating effect may be negative. First, investors might consider human capital and token governance disclosures as signals from the same category (e.g., signals that are not directly related to the product/service). These signals are interpreted as substitutes for each other, whereby the disclosure of additional information could lead to diminishing effects. Second, the specific human capital signal of venture team size increases with the number of full-time employees (FTEs) working on the project that is raising funds via token offering. Given that we identified insider token retention as one of the main categories of token retention, an increase in team size may also lead to an increase in retained tokens if the company wants to provide the same level of incentive alignment on a per-FTE basis as firms with smaller teams and therefore supposedly lower human capital quality. Having a common underlying driver may support the substitution thesis between the human capital and token governance signals. Prior studies on signaling in token-based crowdfunding show the first signs that could support our explanation above. For example, Giudici & Adhami (2019) empirically demonstrate that, after the introduction of additional team variables to their initial regression model, the significant effect of token retention on fundraising success disappears.

However, there are also several factors that may lead to a positive moderating effect of team size on the relationship between token governance incentive alignment tools and both funding amount and post-funding performance. First, the token retention and restriction should be more “valuable” the higher the human capital quality of the company is. Investors expect that the retained and restricted tokens may be (partially) used for future operations. The success prospects of these future operations should be enhanced if the personnel responsible for the operations are of higher quality. Second, token restriction and retention may also impact the importance given by investors to human quality signals. Even though a company may have strong founders and teams, the investor faces the risk that this team could leave the company at any time. The additional disclosure of token retention and token restriction may reassure investors that the team will stay with the company for longer due to their upside participation and potential financial losses in case of earlier departure. This would ultimately lead to a positive joint effect of team size and token governance signals on funding amount and post-funding performance. Third, if we assume an investor considers all available disclosures and signals related to the venture in conducting due diligence, signals supporting each other might lead to an overall larger effect in combination than both would have had independently. This would mean investors do not see the human capital quality and token governance signals as substitutes with a diminishing effect in case they appear simultaneously but rather as supportive signals that enhance each other’s importance for the assessment of venture quality.

Hypothesis 3a: The relationships in 1a-2b are negatively moderated by team size.

Hypothesis 3b: The relationships in 1a-2b are positively moderated by team size.

3 Data and methods

3.1 Empirical setting: data sources and sample

Our study sample is retrieved from our hand-collected Token Offerings Research Database (TORD).¹ With more than 6,400 token offerings, the TORD is the most comprehensive publicly available database for token offerings. The database contains ICOs, initial exchange offerings (IEOs), and security token offerings (STOs) with a start date before January 2021. For our empirical analyses, the sample only includes utility tokens issued during ICOs, given the differences in regulation and governance among the tokens of the three types of token-based crowdfunding. The exclusion of IEOs and STOs provides higher comparability of the observations and avoids potential biases within the dataset. The sample is further reduced due to several additional requirements, including: (i) the ICO published a whitepaper, (ii) the whitepaper included the allocation split of the issued tokens, and (iii) all control variables could be retrieved. The control variables are based on the information published in the whitepaper or other public sources, such as CoinMarketCap or GitHub. The exclusion of ICOs without a published token allocation split in their public whitepaper opens the door to sample selection bias. To address this concern, we corrected for the potential sample selectivity in our robustness check in Section 4.2, using a broader sample of 986 observations. Based on all the described requirements above, the baseline sample consists of 759 observations.

3.2 Variables

3.2.1 Dependent variables

This study analyzes two dependent variables, including one market performance indicator, the funding amount of the venture, and one after-market performance indicator, the post-funding financial performance.

Funding valuation.

In line with previous studies on signaling in ICOs (e.g., Bourveau et al., 2022; Mansouri and Momtaz, 2022; Colombo et al., 2023; Momtaz, 2021a), the funding valuation variable is constructed as the logarithm of the total USD funding amount raised during the ICO. Following existing studies (e.g., Momtaz, 2021a), funding amounts reported in currencies other than USD are converted to USD based on the exchange rate quoted on CoinMarketCap at the start date of the ICO. The values represent the gross proceeds before any deduction of fees (e.g., advisory fees) or bounty programs.

Post-funding performance.

Based on the advantages shown in previous studies (e.g., Fisch & Momtaz, 2020; Mansouri & Momtaz, 2022), the post-funding performance of the tokens is derived from the 12-month buy-and-hold abnormal returns (BHARs). The 12-month BHAR is calculated by subtracting the market benchmark’s buy-and-hold return from the specific token’s buy-and-hold return over a 12-month holding period after the token listing. This definition ultimately leads to the exclusion from post-funding performance analyses of all offerings that did not get listed on a secondary exchange. Given the prominent disadvantages in value-weighted benchmarks, specifically the size differences of the listed tokens due to the high market value of Bitcoin and Ether, we used an equally-weighted market benchmark based on all tokens that are tracked on CoinMarketCap, following Momtaz (2021a) and Mansouri & Momtaz (2022).

3.2.2 Independent variables

Token retention variables.

Token retention is usually not directly communicated but can be derived from the information in the whitepaper. Sections with headings such as “Token Allocation,” “Tokenomics,” or related terms often include an overview of the split of the issued token whereby neither the number of groups nor the definition of the groups is standardized. However, our manual in-depth analysis reveals that, despite the variety among the definitions used in a whitepaper, it is possible to categorize the groups based on the purposes of the allocations. All token offerings included in the sample published a split of their issued tokens within their whitepaper. However, the granularity and naming of the categories strongly vary among the ICOs. Our variables related to token retention are solely defined based on the values published in the venture whitepaper, which is published prior to the respective ICO (ex-ante retention). Token retention is calculated as the natural logarithm of 1 + the percentage of total tokens issued that did not get distributed to the public. The variable is log-transformed to account for potential skewness. For this study, we consider all tokens as retained tokens that did not get distributed in (i) the public sale and pre-sale, (ii) private sale and other kinds of offerings, such as direct exchange offerings, or (iii) promotional programs to support the offering (e.g., Airdrop, Bounty, or Promotions). We further divide the retained token into four subgroups: Stakeholder, Reserve, Venture Development, and Others. The stakeholder allocation includes all tokens distributed to the stakeholder groups of the venture, both internal and external. Depending on the token offering, this could include, for example, the founder of the venture, company employees, partners, advisors, or shareholders.

The reserve pool allocations include allocations that are retained by the company for future (reward) distribution to the network or stabilization of the project. Considered allocations are not only the ones explicitly allocated to “Reserve” but also those that are allocated to categories with a meaning that falls within our definition of reserve pool, such as mining pool, rewards pool, stabilization fund, or liquidity pool. Allocations to the development of the business, ecosystem, or strategic partnerships are captured under the “Venture Development” category together with the retained tokens for research, marketing, legal, or related purposes. All allocations to categories whose purpose does not fall into one of the three groups defined above are combined into “Others.” This could, for example, include allocations to social programs, project-type-specific purposes, or offering costs. This group only represents a small share of the distributed tokens and is not included as a separate category in our empirical analyses. All variables for the subgroups are calculated in line with the retained token variable, taking the natural logarithm of 1 + the percentage allocation to the subgroup in the token offering.

Token restriction variables.

Ventures that decide to implement token restrictions in their ICO usually disclose this information in the same chapter of the whitepaper as the token retention. The token restriction variable is a dummy variable, which takes the value of 1 if the whitepaper of the token offerings contains information about a restriction on any of the distributed tokens. Additional dummy variables are included based on the defined restriction types in Section 2.2.2. The initial restriction variable is a dummy variable, which takes the value of 1 if either a cliff or a lock-up is disclosed within the whitepaper. The (reverse) vesting restriction variable is another dummy variable, indicating if any distributed tokens are attached with a vesting or reverse vesting restriction. The assessment of the exact restriction structure is notably demanding since individual token issuers sometimes interchange some of the terms (e.g., cliff and lock-up) in their restriction descriptions. The restriction variables do not consider the allocations to the reserve pool category. Allocations to the reserve pool are restricted by nature and often unlocked over many years without a fixed (reverse) vesting schedule, strongly dependent on the development of the individual projects.

3.2.3 Control variables

Prior ICO studies (e.g., Howell et al., 2020; Mansouri & Momtaz, 2022; Momtaz, 2020) have identified a large number of factors that could impact the success of token offerings. Focusing on the variables that have already shown significant effects in previous studies and omitting the others, we define three categories of control variables for our research: (i) venture characteristics, (ii) offering characteristics, and (iii) market characteristics. Venture characteristics include human capital related variables, such as the total number of team members and the percentage of team members that have a technical background. We consider an employee to have a technical background when the team members’ professional network profile (e.g., LinkedIn) shows a degree from a technical field, such as computer science, information technology, or comparable areas. Furthermore, we included dummy variables to control if the startup already has a minimum viable product available at the ICO date (MVP) and if it discloses the code on GitHub (open source). The whitepaper length variable is based on the natural logarithm of the total words included in the published whitepaper, often used as an indication for the total information available to the investor (e.g., Fisch, 2019; Mansouri & Momtaz, 2022). Related to offering characteristics, we identified important design features of the ICO and set the respective dummy variables to 1 if the features were included and 0 otherwise. Based on prior studies (e.g., Howell et al., 2020; Mansouri & Momtaz, 2022; Momtaz, 2020), we considered if the venture announced a minimum funding amount required to be successful (soft cap) or a maximum total amount accepted (hard cap), conducted a pre-sale event (pre-sale) or offered a bonus structure (bonus) or bounty program (bounty) to its investors. Separately, we controlled for a possible whitelist feature of the token offering and if the underlying technical standard is ERC20 (ERC20). Lastly, we also controlled for the market sentiment during the ICO, including the bull market and bear market dummy variables. The bear market period is from February 2018 until January 2019, while the bull market considers the time prior to February 2018. The full set of included control variables and their definitions are listed in Table 10.

3.3 Summary statistics

The summary statistics, including arithmetic mean, standard deviation, and the 25%, 50%, and 75% quantile, are presented in Table 3.

Table 3 Descriptive statistics

3.3.1 Summary statistics: dependent variables

Table 3 shows that, on average, the startup in our sample raised an amount of $3.80 million (log. = 15.15; log. SD = 1.87) during the token offering and experienced a buy-and-hold abnormal return of –0.663, with a standard deviation of 1.11.

3.3.2 Summary statistics: key variables

The mean of the token retention (log) variable is 3.48, which translates into a 31.5% average share of retained tokens by the startups. The split among the different categories shows that 12.2% (avg. log. = 2.58; log. SD = 0.93) were retained for purposes relating to stakeholder allocations, 2.55% (avg. log. = 1.60; log. SD = 1.52) for reserve pool allocations, and 4.0% (avg. log. = 1.27; log. SD = 1.40) and 0.2% (avg. log. = 0.20; log. SD = 0.66) for venture development and other allocations, respectively. In our baseline sample, all ventures published an allocation split in their whitepaper. Out of the 759 token offerings, in 749 offerings, a defined amount of tokens got retained, and in only 10 cases, 100% of the tokens were distributed during the ICO. The key restriction variable statistics show that, based on our sample, 49.4% of startups have a token allocation restriction disclosed within their whitepaper; 37.3% have an initial restriction of some or all distributed tokens, meaning that these startups incorporated either a lock-up or a cliff restriction; and slightly fewer ventures (30.4%) communicated a vesting period or a reverse-vesting period in their whitepaper relating to their token distribution. In 18.3% of the offerings, we observed a combination of initial restrictions and (reverse) vesting restrictions.

3.3.3 Summary statistics: control variables

The summary statistics in Table 3 indicate that the average venture has a team size of 13.4 FTEs, of which approximately 24.8% have a technical background. On a scale from 1 to 5, with 5 indicating the highest venture quality, industry experts on ICObench rated the average venture 3.4. Among all sample ventures, 19.8% already had a minimum viable product (MVP) available at the ICO start, and 66.1% published their code on GitHub. The average whitepaper length for the startups under study was around 3,484 words (avg. log. = 8.156; log. SD = 0.591). A hard cap was announced for most of the token offerings in the sample (90%), and a soft cap (66.4%) for approximately two-thirds. Regarding the token offerings, 57.0% and 32.4% had an implemented pre-sale period and active whitelist, respectively. With respect to the possible promotion schemes in token offerings, 33.3% of the ventures offered a bounty program, and 0.7% offered a bonus structure. The technical ERC-20 standard was relied upon for 81.6% of the token offerings. Regarding the market sentiment during the ICO, 30.0% of the analyzed offerings took place during a bull market and 71.7% during a bear market.

3.3.4 Correlation

Table 4 presents the pairwise correlation coefficients for all included variables. Given the high correlations among the different restriction variables, we run separate regression models to avoid multicollinearity.

Table 4 Correlation matrix

3.4 Econometric approach

The substantial part of our empirical analyses aims to estimate the causal effects of token retention and token restriction on (i) the funding amount and (ii) the post-funding performance in token offerings. Therefore, in our defined baseline regression, we are interested in the effect of the amount of retained tokens ($RT_i$) and the token restriction dummy ($TR_i$) on the dependent variables $DV_i$ $\in$ {Valuation$_i$, Performance$_i$}, using an ordinary least squares model. The model further includes a vector of control variables($\Omega _i$):

$$\begin{aligned} DV_i= & \beta _1RT_i + \beta _2TR_i + \Omega _i\gamma + \epsilon _i , DV_i \in \nonumber \\ & \{Valuation_i , Performance_i\} \end{aligned}$$

(1)

In the next steps, we address the natural concerns that the estimated regressors from the baseline regression (1) could suffer an endogeneity bias (E[$\Omega _i$, $\epsilon _i$] = 0). Following the techniques used in previous studies in the field of entrepreneurial finance, we applied three different two-stage models to control for observed and/or unobserved heterogeneity (e.g., Bertoni et al., 2011; Colombo & Grilli, 2010), specifically: (i) inverse Mills ratio (IMR model), (ii) generalized residuals (GR) as an estimator (GR model), and (iii) generalized residuals (GR) as instrumental variables (IV model). To address potential endogeneity problems, all of the chosen approaches above rely on a selection model estimation in the first stage. We are specifically interested in the selection of ventures into their token retention share. Section 2.3 shows that better startups should retain more tokens (Leland & Pyle, 1977). Therefore, to estimate the impact of token retention on fundraising success, we must control for this selectivity. Equation (2) models the probability that startup i has a token retention share above the median ($hiRT_i$), based on a vector of exogenous variables that could influence the selection mechanism, $\Omega _i^{(s)}$:

$$\begin{aligned} hiRT_i = \Omega _i^{(s)} \delta + \xi _i \end{aligned}$$

(2)

In our first approach and based on the results of Equation (2), we compute the inverse Mills ratio (IMR$_i$) for the selection of each startup i (e.g., Heckman & Navarro-Lozano, 2004; Mansouri & Momtaz, 2022). The cumulative density function of the standard normal distribution is indicated by $\Phi$(.), and the probability density function by $\phi$(.).

$$\begin{aligned} IMR_i = \frac{\phi (\frac{\Omega _i^{(s)} \delta }{\sigma _\xi })}{\Phi (\frac{\Omega _i^{(s)} \delta }{\sigma _\xi })} \end{aligned}$$

(3)

We then add the computed IMR$_i$ to our baseline regression in the second stage, where $\lambda$ tests the null hypothesis that no selection effect exists:

$$\begin{aligned} DV_i^{IMR}= & \beta _1RT_i + \beta _2TR_i + \lambda IMR_i + \Omega _i\gamma \nonumber \\ & + \nu _i , DV_i^{IMR} \in \{Valuation_i^{IMR} ,\nonumber \\ & Performance_i^{IMR}\} \end{aligned}$$

(4)

For our second and third empirical approaches, we compute generalized residuals (GRs) from the results from Equation (2). In line with Gourieroux et al. (1987), we define the generalized residuals (GRi) for startup i as:

$$\begin{aligned} GR_i= & hiRT_i \times \frac{\phi (-\Omega _i^{(s)} \delta )}{(1-\Phi ) (-\Omega _i^{(s)} \delta )} + (1-hiRT_i) \nonumber \\ & \times \frac{-\phi (\Omega _i^{(s)} \delta )}{\Phi (-\Omega _i^{(s)} \delta )} \end{aligned}$$

(5)

In the second stage of the GR model, we then add the computed $GR_i$ to our baseline regression (Equation (1)), where $\zeta$ tests the null hypothesis that no selection effect exists:

$$\begin{aligned} DV_i^{GR}= & \beta _1RT_i + \beta _2TR_i + \zeta GR_i + \Omega _i\gamma + \nu _i , \nonumber \\ & DV_i^{GR} \in \{Valuation_i^{GR} , Performance_i^{GR}\} \end{aligned}$$

(6)

To control for unobserved heterogeneity, we also use the computed GRs from Equation (5) as instrumental variables for startups’ token retention share in the IV model. We restrict the standard deviation of the error term for startups with above-median token retention shares ($\sigma _{\epsilon , \,hiRT=1}$) to be equal to that of startups with below-median token retention shares ($\sigma _{\epsilon , \,hiRT=0}$), ensuring that $GR_i$ can be added as an instrumental variable to our baseline regression (1) (Mansouri & Momtaz, 2022).

4 Empirical results

4.1 Main results

4.1.1 Token offering success

Table 5 shows the regression results for the effect of token retention and token restrictions on the startup funding valuation. The funding valuation is represented by the natural logarithm of the USD funding amount, which is the dependent variable for models (1), (2), (3), (4), and (5). The baseline (OLS) regression (1) estimates the effects of the natural logarithm of the percentage share of retained tokens and the token restriction dummy variable without considering the potential selection bias. We control for selection-related endogeneity in models (3), (4), and (5), which are based on inverse Mills ratios (IMRs), generalized residuals (GR), and generalized residuals as instrumental variables (IV (GR)), respectively. Model (2) is an additional control model, estimating the effects of all control variables without including the key variables regarding token retention and token restriction. All models include a variety of control variables related to the venture characteristics and specifics of the token offering, as well as country and quarter-year fixed effects. The reported standard errors are robust, and the $R^2$ exceeds 30% in all models, slightly higher than in most existing studies (e.g., Davydiuk et al., 2023). The key variables included in the analysis have variance inflation factors (VIFs) below 3, which is well below the widely accepted threshold of 5 (e.g., Leitterstorf & Rau, 2014; Mansouri & Momtaz, 2022). The additional control variables also have VIFs below 3, except for the two market characteristics controls. These results imply that multicollinearity is not a prominent issue in our study.

Table 5 The effects of token retention and restrictions on the funding amount

Overall, the regressions show different results for token retention and restriction effects. All models suggest that token retention has a significantly positive effect on the fundraising amount in token offerings, with a marginal effect of the natural logarithm of the percentage of retained tokens ranging from 0.308 to 0.318 and a p-value always lower than 10%. Our baseline model (1) shows a coefficient for token retention (log) of 0.318, suggesting that a 1% increase in the share of retained tokens increases the average funding amount by 0.318%. These findings strongly support hypothesis 1a, that token retention has a positive signaling effect in initial coin offerings, and are in line with previous studies showing a negative effect of a larger share of distributed tokens (e.g., Giudici & Adhami, 2019; Lyandres et al., 2022). We thus provide further evidence that the signaling theory of Leland & Pyle (1977) holds in the context of initial coin offerings. Investors might interpret the amount of retained tokens as a quality signal that reveals the fair value of the underlying venture, given that good entrepreneurs should retain more tokens than bad entrepreneurs.

Additionally, the results of all models suggest that the inclusion of a token restriction is not statistically significant, implying that investors do not consider the whitepaper disclosure of token restriction mechanisms as an important signal of the quality of the underlying venture. These results indicate that investors might not give importance to the risk of entrepreneurs selling their retained tokens immediately after listing or do not believe in risk mitigation based on a token restriction disclosure. There are several possible reasons for this contradiction to our defined hypothesis 2a; for example, (i) investors may be concerned about the restricted flexibility of the venture to react to unexpected liquidity needs or market environment changes, or (ii) investors may doubt the accuracy of the disclosure. Cohney et al. (2019) find evidence supporting this second possibility, presenting that most ventures that disclose a token restriction do not embed those rights into the source code. Overall, these results are in line with the study by Bourveau et al. (2022), who also did not find a significant relationship between the disclosure of a vesting schedule and the fundraising amount. However, the findings are contrary to those of Davydiuk et al. (2023), which showed a positive significant relationship. This variation in findings might be due to several factors. First, our study includes several additional significant control variables, such as whitepaper length and expert rating, which are not included in Davydiuk et al.’s (2023) model. Additionally, we include country fixed-effects and robust standard errors resulting in a higher $R^2$ (0.338 vs. 0.202). Second, we consider all types of token restrictions, while the prior study only mentions vesting schedules. It might be possible that our in-depth analysis of the whitepaper includes additional restrictions (e.g., lock-up or cliff) that are not included in the prior study.

Table 5 shows that the results in our baseline OLS model (1) closely correspond to the second-stage models (3), (4), and (5), providing evidence that the results in model (1) are not strongly biased by underlying selection-related endogeneity. This suggestion is further supported by the non-significant IMR and GR coefficients in models (3) and (4), respectively (not tabulated). We also note that the regression coefficients and correlating statistical significance of the control variables in models (1)-(5) are generally in line with those in previous studies (e.g., Fisch, 2019; Fisch et al., 2022; Mansouri & Momtaz, 2022). The length of the whitepaper, the score in the expert rating, and the number of FTEs related to the project show a significantly positive relationship to the fundraising amount. For example, we can observe in the baseline model that a 1-point increase in the expert rating leads to a 49.7% increase in the funding amount, significant at the 1% level. The choice of the ERC-20 standard and the venture’s decision to disclose its code on GitHub negatively impact the amount raised in the token offering, leading to a 33.3% and 40.8% decrease in the funding amount, respectively. Our sensitivity check in column (2) excludes the two key variables. The regression results show a high similarity to those of our baseline model (1) regarding the direction of the effects and their magnitude.

Table 6 presents the results of running the same regressions again, this time using the different token retention categories instead of the overall token retention share. It is observable that the main effect of token retention is driven by the share allocated to the stakeholder group. Model (1) shows a coefficient for stakeholder allocation of 0.162, implying that a 1% increase in the share of retained tokens for stakeholders increases the average funding amount by 0.162%. This finding supports the theory that investors consider an increased number of tokens retained by insiders as a quality signal of the underlying venture and is in line with findings from alternative early-stage fundraising research (e.g., Conti et al., 2013). Allocations to stakeholders might also be interpreted as a further incentive alignment with the decision-makers of the venture, given that those groups are financially participating in the upside of the project. This alignment could moderate the risks investors face in the token offering market, such as moral hazard, adverse selection, and outright fraud. The venture development allocation shows a positive coefficient ranging from 0.081 to 0.101 among the 4 different models. However, when we control for selection-related endogeneity based on generalized residuals (GR) and generalized residuals as an instrumental variable (IV (GR)), the significance disappears.

Table 6 The relations between token allocation options and the funding amount

4.1.2 Post-funding performance

Panel A of Table 7 presents estimates of the effect of token restriction and token retention on the post-funding performance of the tokens based on the 12-month buy-and-hold abnormal return. The 4 different models are defined as in the previous table, showing the OLS model in column (1) and the different two-stage models in columns (2), (3), and (4). We can observe that the share of retained tokens does not significantly impact the post-funding performance. Consequently, the findings do not support our hypothesis 1b, that investors considered token retention as a positive signal of firm quality even after the close of fundraising. Therefore, we also do not confirm the results shown in the context of the public equity market (e.g., Bhagat & Bolton, 2008; McConnell et al., 2008). A plausible explanation for the non-significant results is that the correlation between post-funding performance measured based on the price of the listed financial instrument and the operational performance of ventures using token-based crowdfunding might be less than for publicly listed companies in the traditional equity market. Our hypothesis was partially built on prior findings that token retention increases post-funding operational performance, which should translate to superior token performance. However, this might not be the case for several reasons. First, the investor has no audited report available to receive information about the operational performance of the venture. Second, it might be difficult for investors to assess the operational performance by themselves due to limited knowledge and resources. The previous studies created their own variables to measure operational success, such as the number of downloads (Davydiuk et al., 2023) or the number of wallets containing the issued token (Lyandres et al., 2022), which would be difficult to compute for an average investor. Third, token performance could be driven more by market speculation unrelated to the underlying operational performance than by the share performance in the traditional equity market. Additionally, the token retention variables are defined based on the published pre-ICO values in the whitepaper (ex-ante retention) and might differ from the actual retained amount of tokens (ex-post retention). The venture does not disclose updated information that could give the investors an indication of the ex-post retention amount. The validation of the exact ex-post retention amount on the blockchain might also be practically impossible since it would require knowledge about the relevant wallet addresses. This situation provides further reasoning as to why the ex-ante token retention amount might not be an important signal for post-funding performance, leading to the non-significant results in Table 7.

Table 7 The effects of token retention and restrictions on the post-funding performance

The relationship between token restrictions and post-funding performance is significantly negative, contradicting our initial hypothesis 2b. The marginal effect of the announcement of a token restriction in the whitepaper ranges from –0.376 to –0.385, statistically significant with p-values below 5%. The fact that the coefficient from the main OLS model and those of the two-stage models closely correspond implies that the treatment effect of token restrictions on ICO ventures’ funding is not impacted by any underlying selection effect. In economic terms, our findings indicate that the inclusion of token restrictions leads, on average, to a 37.6% decrease in the 12-month buy-and-hold abnormal return of the token. There are several possible reasons for this significant negative impact. First, the decision of the venture to restrict the retained tokens consequently restricts its flexibility to react to unexpected liquidity needs or changes in the market environment. It will not be possible to transfer or liquidate tokens, no matter which external factor arises. This consequence is accompanied by the related disadvantage of increased opportunity costs; the venture could not invest in any profitable project if an attractive opportunity arises. These disadvantages potentially negatively impact the operational performance of the venture and, ultimately, the token performance. Second, investors might have detected a breach of the disclosed restriction terms during the restriction period. These detections would decrease the overall trust of the investors in the project and lead to the divestment of their token holdings. Cohney et al. (2019) have found evidence that the majority of ventures that disclose a token restriction do not encode those rights into the smart contract, which, based on our explanation above, would lead to a negative impact on the post-funding performance of the token after the investor becomes aware of the breach of the communicated restrictions.

In Panel B of Table 7, we examine separately the impact of the two different types of restrictions, (i) initial restrictions (lock-up and cliff) and (ii) restrictions that are gradually released over time (reverse vesting and vesting). The results clearly show that the impact of the disclosure of initial restrictions is more negative than that of the other types. While a reverse vesting or vesting schedule does not have a significant impact on post-funding performance, the initial restrictions (lock-up and cliff) do, on average, lead to a decrease of 38.7% in the 12-month buy-and-hold abnormal return of the token, statistically significant with p-values below 5%. A plausible explanation for this finding might be that the market impact around the end date of the token restriction might be more negative for initial restrictions since the prior restricted token holders are now able to transfer the full amount. By contrast, a restriction gradually released over time would be expected to have a less significant impact around the end date of the token restrictions ($\sim$12 months after listing), given that tokens are gradually unlocked over the horizon of the restriction length (e.g., monthly). This is supported by studies conducted in the context of IPO lockups (Field & Hanka, 2001), showing that increased trading volume and a negative abnormal return are observable after the expiration of the restriction. Additionally, a breach of an initial restriction is easier for investors to detect than a breach of a gradually released restriction since no transaction from the restricted token pool should be conducted. In the case of a vesting schedule or a reverse vesting restriction, transactions are expected because, starting from the first day after the ICO close, tokens could get unlocked. Therefore, it might be harder to detect if a breach of the restriction happens and the transactions are larger than allowed under the disclosed restriction terms.

Taken together, our results confirm the token retention outperformance hypothesis (TRetO) related to the initial fundraising but not to post-funding performance. In particular, ventures that retain a higher degree of tokens during the ICO outperform during the funding period but do not have a significant difference in their post-funding performance. Overall, our results reject the token restriction outperformance hypothesis (TresO). While showing no significant effect during the funding period, ventures that implement a token restriction underperform their peers in the 12-month post-funding period.

Table 8 The moderating effect of the venture team size

4.1.3 Team size as a moderator

Table 8 highlights the moderating effect of venture team size on token retention and token restriction in their relationship with the token offering funding amount (Panel A) and post-funding performance (Panel B). Panel A shows that team size negatively moderates the impact of token retention on fundraising amount with a coefficient of –0.039. In economic terms, this means that the 0.757% increase in fundraising amount based on a 1% increase in retained tokens is reduced by, on average, 0.039 percentage points per additional FTE added to the venture team. The moderating effect and the variables for token retention and team size are all statistically significant at the 1% level in the main model and in the second-stage models, implying that the moderating effect is not impacted by any underlying selection effect. The results do not show any significant moderating effect of team size on the relationship between token restriction and the funding amount. The results in Panel A support hypothesis 3a, that team size negatively moderates the impact of our defined token governance variables. The findings might be explained by the possibility that investors consider the signals as belonging to the same category and ultimately as substitutes for each other. The effect of token governance signals would therefore decrease when the human capital quality increases. Additionally, the size of token retention is mainly driven by the number of tokens retained for insiders, among which is the venture team, which means that the increase in the team size could consequently lead to an increase in the share of retained tokens. This common underlying driver may further support the substitution effect.

The moderating effect of venture team size in the context of post-funding performance (Panel B) shows no significance related to the effect of token retention, which is itself not a significant factor in the regressions. However, despite team size not showing any significance in explaining the post-funding performance variation of listed tokens, the variable has a statistically significant moderating effect on the relationship between token restriction and the 12-month buy-and-hold abnormal return. Post-funding performance would be decreased by an average of 107.9% in case the venture published a token restriction in the offering whitepaper and in the hypothetical case that the team size is 0. The larger the number of FTEs within the venture team, the lower the reduction in post-funding performance caused by the token retention disclosure. For example, if the team consists of 10 FTEs, the average decrease of 12-month buy-and-hold abnormal return allocated to the token retention would be reduced to 48.9%. These findings provide evidence for our defined hypothesis 3b and might be explained by the fact that restricted tokens are seen as more valuable when the quality of the human capital of the venture is higher. The larger team size, representing increased human capital quality, in combination with restricted tokens, could therefore provide several positive signals to the investors. First, the future operations of the venture might have increased prospects of success because the persons responsible for the operations are of higher quality. Second, the personnel have an incentive to stay within the company due to its restricted token holdings.

4.2 Robustness tests and post-hoc analyses

Not all ventures publish token allocation information in their public whitepaper; hence, this information being required for inclusion in our baseline sample for the quantitative analyses might lead to a potential selectivity issue. We thus run a two-stage least squares test to address this concern and correct any potential selectivity bias. First, we predict the probability that the venture publishes the token allocation information based on an expanded sample of token offerings. Second, we compute the generalized residuals from our prediction results in the first step, following the approach we used for the main regressions and in line with Gourieroux et al. (1987). Third, we add the computed generalized residuals as a control variable to the different regression models. Table 9 reports all results for the regression based on the above-defined adjustments. We find that our main results do not qualitatively change after the inclusion of the adjusted generalized residuals as a control variable. These findings suggest that our results are robust, and the selectivity does not bias the results of our regressions.

Table 9 Robustness checks — controlling for token split-related selectivity

We conducted additional tests about the sensitivity of our results to the inclusion of further control variables and fixed effects. This might have the advantage that more variation could be absorbed by the additional variables, with the cost of reduced sample size, based on the limited availability of the variables. In Table Table 11 we additionally included control variables for the Shareholder Protection Index (SPI) and the Financial Development Index (FDI) to consider the institutional background. SPI is based on World Bank data and indicates the minority investor protection based on the ease of shareholder suits index, ranging from 0 to 10. FDI is published by the International Monetary Fund and ranges from 0 to 1, with 1 being the highest development score. We allocated each token offering the corresponding SPI and FDI score based on the published country of the project headquarters at the year of ICO start. While there are no significant effects of the FDI and SPI variables in the regressions related to the funding amount (log), we observe highly significant results related to the BHAR 12-month regressions. In column (4), (5), (6) we show that the Financial Development Index and Shareholder Protection Index have a positive significant effect on the post-funding performance at minimum the 5% level. A 0.1 point increase in the FDI increases the BHAR on average by 3.8%. A 1-point increase in the SPI increases the post-funding performance on average by 2.9%. However, we can conclude that the inclusion of FDI and SPI does not materially change the direction or significance of our identified main effects, suggesting that our results are robust.

Seperately, we included a platform fixed effect to control for the fact that the underlying project can be a platform project. Our sample includes 453 offerings that are indicated as platform projects, representing 59.7% of all considered ICOs. The rationale behind this robustness check is that we expect tokens that represent a platform project to share some structural similarities with each other, comparable to companies that operate within the same industry. Table 12 summarizes the regression results of all models in the manuscript, including the platform fixed effect. We observe that our results are robust and the additional fixed effect does not materially change the direction or significance of our identified main effects. While we tested for a variety of other control variables, we did not report additional results for the sake of brevity.

Further investigations regarding the relationship between token governance mechanisms and alternative dependent variables and human capital moderators were conducted during post-hoc analyses. We included the variable “crypto background” as alternative measure for human capital quality, which indicates the number of the total team members with prior crypto experience before ICO start. Crypto experience includes work experience directly related to previous initial coin offerings and other crypto business models. The underlying projects of the offerings considered in our sample have on average 4.6 team members with previous crypto experience. Table 13 shows that the number of team members with crypto experience has a significant negative moderating effect on the relationship between token retention and funding amount (log). In economic terms, this means for the main model (1) that the 0.569% increase in fundraising amount based on a 1% increase in retained tokens is reduced by, on average, 0.062 percentage points per additional FTE with prior crypto experience added to the venture team. We believe this negative moderating effect might be based on the same reasons we derived for the positive moderating effect of the total number of team members on token retention. Investors might consider the human capital signals and the token governance signals as belonging to the same category and ultimately interpret them as substitutes for each other. The effect of token governance signals therefore decrease when the human capital quality increases. We do not observe any significant moderating effect of the crypto experience on the relationship between token governance mechanisms and post-funding performance. Further post-hoc analyses included the regressions of our variables on a set of different depend variables, such as the fraction raised of the ICO soft-cap or hard-cap and the expert rating, could not show any significant results and are therefore not reported.

5 Discussion and concluding remarks

5.1 Summary of main results

This paper tests two overarching hypotheses related to the two main token governance mechanisms used in token-based crowdfunding and their impact on ICO ventures’ funding and post-funding performance. Our sample is based on initial coin offerings that issued utility tokens, which furnishes an optimal research environment due to (i) the lack of regulation that amplifies the importance of incentive alignment for investors, (ii) the absence of traditional contractual corporate governance tools for investors to rely on, and (iii) the listing of tokens on a liquid cryptocurrency exchange after ICO, which allows for the token holders to trade with other investors (Adhami et al., 2018) and for a transparent assessment of financial performance (Fisch & Momtaz, 2020).

The first hypothesis, the token retention outperformance hypothesis (TRetO), posits that token allocation resulting in a higher degree of retained tokens in the primary market positively impacts ventures’ fundraising amount during the ICO and post-funding performance. Our results confirm the TRetO relating to the initial fundraising but not to the post-funding performance. We find significant results that the amount raised by the venture is increased by 0.318% when 1% more tokens are retained. Additionally, we provide evidence that the effect is mainly driven by the signal generated by retention allocated to stakeholder groups rather than venture development or reserve pool allocations. We utilize various approaches to tackle potential endogeneity issues, including two-stage and instrumental variable methods. Our results affirm that the superior fundraising amount of the startup arises from the treatment effect of a higher share of retained tokens rather than being confounded by a selection bias.

The second hypothesis, the token restriction outperformance hypothesis (TresO), posits that token restrictions that limit the reselling of tokens in the secondary market positively impact the venture’s fundraising amount during the ICO and the post-funding performance. Our results do not confirm the TresO. Rather, we find that the disclosure of a token restriction has a significantly negative effect on the 12-month buy-and-hold abnormal return. The presence of the restriction reduces the return on average by 37.6%. Additionally, we divided the token restriction variable into its different types and found evidence that the effect is mainly driven by the impact of the disclosure of an initial restriction. While (reverse) vesting does not show any significant effect, the relationship between the inclusion of initial token restrictions (i.e., lock-up or cliff) and post-funding performance is negative. The disclosure of an initial restriction leads, on average, to a reduction of 38.7% in the 12-month buy-and-hold abnormal return. In addition, we do not find a significant relationship between the disclosure of token restrictions and the fundraising amount in initial coin offerings.

We also find evidence that the presence of human capital signals, such as the team size of the venture, moderates the effect of token governance. Our hypothesis 3a, that the relationships above are negatively moderated by team size, is partially confirmed with regard to the effect of token retention on the fundraising amount. We observe that the 0.757% increase in fundraising amount based on a 1% increase in retained tokens is reduced by, on average, 0.039 percentage points per additional FTE added to the venture team. Our hypothesis 3b, that the relationships are positively moderated by team size, is rejected with regard to fundraising amount. However, in the case of post-funding performance, we find supporting evidence. The negative impact of token restriction on post-funding performance is decreased by, on average, 5.9 percentage points for each additional full-time employee added to the venture team. Our robustness test based on a larger sample of token offerings suggests that our results are robust, and the selectivity related to the sole consideration of ICOs that published the token allocation split in the whitepaper does not bias the results of our regressions.

5.2 Theoretical contributions and practical implications

Our study makes several important theoretical contributions to the early-stage crowdfunding literature. First, we provide new findings to the general literature on ICO success relating to both funding and post-funding performance. Prior research has already identified numerous important factors influencing ICO ventures’ funding amount that are related to, for example, offering and venture characteristics (e.g., Belitski & Boreiko, 2022; Howell et al., 2020; Momtaz, 2020), and founder CEO characteristics (Colombo et al., 2022; Xia et al., 2023). We expand this literature with new findings regarding the token governance features of the tokens. The literature on post-funding performance in early-stage fundraising mainly focuses on operational performance (Vanacker et al., 2019; Davydiuk et al., 2023; Lyandres et al., 2022), while research about the characteristics that drive the post-ICO financial performance of the venture is still less conclusive (Benedetti & Nikbakht, 2021). Our research into token governance mechanisms identifies token restrictions as a salient token feature that negatively influences the financial return of the listed token. Our research thus helps to close the research gap regarding factors that influence the abnormal returns of token offerings and advances our general understanding of the cryptocurrency market. The additional findings related to funding amount and post-funding performance can also be supportive of the investors’ due diligence prior to the submission of an investment.

Second, we present new knowledge about token governance and the design of digital assets. We analyze token governance in a novel and more granular way than existing empirical studies, using the implied purpose of the token retention and token restriction disclosures based on the information provided in the individual whitepapers. Prior work has measured the share of token retention as the percentage not distributed during the ICO (Davydiuk et al., 2023; Giudici & Adhami, 2019; Lyandres et al., 2022) or solely the portion allocated to insiders (Bourveau et al., 2022). Our process allows us to define subgroups, categorize the provided information, and potentially help investors assess the unstructured information in the various token offering documents, given the lack of standardization in the ICO market. The results of our study based on the more granular analysis of retained tokens confirm prior findings that token retention has a positive impact on ICO ventures’ funding (Davydiuk et al., 2023; Giudici & Adhami, 2019; Lyandres et al., 2022). Our novel research setting also allows us to expand the existing literature (Cummings et al., 2020), allowing us to estimate the effects of the retentions allocated to the different groups. We find empirical evidence that the positive effect of token retention is mainly driven by the tokens allocated to stakeholders while showing no constant significance of reserve pool or venture development allocations among our regression models. Additionally, prior research mainly focuses on the impact of token retention on the operational post-funding performance of the ventures that have raised funds via token offering (Davydiuk et al., 2023; Lyandres et al., 2022). We expand the literature in two directions. First, we analyze the impact on the post-funding financial performance of the listed tokens, showing that the increased operational performance does not translate to a significant increase in financial post-funding performance. Second, we provide evidence that token restrictions negatively impact the buy-and-hold abnormal return of the tokens. These findings have several practical implications for entrepreneurs and investors. Entrepreneurs should be aware that not all token governance tools are associated with positive financial returns or increased fundraising. The direction and strength of the signal depend not only on the chosen incentive alignment tool but also its defined characteristics in the offering document.

Third, we contribute to the literature on the role of human capital in new venture success. More specifically, we provide evidence of how token governance signals are moderated by human capital quality. Our results extend the findings of prior studies which unveiled the moderating effect of different variables on the effect of token governance mechanisms in ICOs, such as the scope of disclosures (Bourveau et al., 2022; Davydiuk et al., 2023), the number of active ICOs (Davydiuk et al., 2023), and investor rights embedded in the token (Giudici & Adhami, 2019). Our findings help to better understand the impact of governance signals in a market with a high level of information asymmetry and also have practical implications. Ventures that are conducting token offerings should keep in mind the potential substitution effect for token governance tools and other types of signals when designing the structure of the offering.

5.3 Limitations and avenues for future research

Our study represents an important step towards understanding the token governance tools in ICOs and their impact on ventures’ funding and token price performance. We thus propose potential directions for further research in this area.

Differences among the underlying offering types Our study focuses on ICOs, in which utility tokens are issued during the token offering. While using this empirical setting for a study addressing the impact of token governance provides important insights, the transferability of the results to other offerings and token types might be limited. For example, ventures issuing security tokens (e.g., Lambert et al., 2022; Kreppmeier et al., 2023), which are usually considered securities and therefore must follow the security laws of the underlying jurisdiction, might be forced to implement other incentive alignment tools based on the applicable regulatory requirements. In addition, the required level of trust from investors in the accuracy of the disclosures of the entrepreneurs would decrease, given the possible legal enforcement in case of a breach of any clause by the venture. These variations in the offering characteristics provide potential research avenues regarding the effectiveness and investor perception of governance mechanisms for different token offering types. Avenues that will further develop as the research around policy and regulation of ICOs is expected to be increasingly urgent in the coming years (Brochado & Troilo, 2021).

Increase in investor sophistication Our research is based on a hand-collected sample of token offerings before 2021, which allows us to examine the impact of token governance on the tokens’ post-funding financial performance. However, the sophistication of investors, as measured by their understanding of the underlying technology, may have increased over the past years. An elevated knowledge among investors would lead to an enhanced ability to assess the underlying source code and hence the accuracy of ICO disclosures published by the venture. Given the blockchain’s public nature, investors might conduct their own expanded due diligence on the projects before investing to decrease the required level of trust, which could ultimately impact the signaling effect of the different disclosures. Further research might focus on the relationship between the level of investor knowledge and the signaling effect of the token governance tools.

Variety of available signals in token offerings Our study finds empirical evidence that token governance and human capital signals might be perceived by the investor as substitutes or complementary signals. Future research could expand this study in two directions. First, we sought to capture human capital quality through the team size of the ventures and included a post-hoc analysis where we considered the crypto experience of the team members. It would be possible to include further human capital quality indicators, such as the educational and professional background of the funding team or the gender diversity of the management. Second, we solely analyzed the interaction of token governance signals with those of human capital quality. Examining the relationship with other signals, including social or intellectual capital, might yield equally compelling results.

5.4 Concluding remarks

This paper has sought to shed light on the token governance mechanisms used in ICOs and their role in venture valuation and post-funding token performance. The main governance tools used in token offerings are token retention and token restriction. Our results suggest that token retention positively impacts the fundraising amount of the venture, mainly driven by allocation to stakeholders, but does not significantly impact the post-funding performance. Token restrictions include both initial restrictions and restrictions that are gradually released over time. We provide empirical evidence that initial restrictions negatively impact the 12-month buy-and-hold abnormal return. The signaling effects of the token governance disclosures are moderated by human capital signals, whereas the direction differs among the types of signals. This study contributes to understanding, developing, and implementing effective token governance to mitigate risks in early-stage venture financing.

Appendix

Table 10 Variable Definitions

Table 11 Robustness checks – controlling for alternative control variables

Table 12 Robustness checks – controlling for platform fixed effect

Table 13 The moderating effect of the venture team crypto experience