The Influence of Social Media Expert
Endorsements on Cryptocurrency Market Price
and Trade Volume
A Time-Series Analysis
Author: Guy J. Schultz ID: 11697709
Supervisor: Dhr. Dr. G.T. Vinig
Thesis submitted in partial fulfilment of the requirements of a M.Sc. Degree in
Business Administration – Entrepreneurship & Innovation
University of Amsterdam
Graduate School of Business and Economics July 2018
i
Statement of Originality
This document is written by student Guy Schultz who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
ii
TABLE OF CONTENTS
ABSTRACT ……… iii
1. INTRODUCTION ………..… 1
2. LITERATURE REVIEW & HYPOTHESES DEVELOPMENT.……….. 7
2.1.Cryptocurrencies ……….…… 7
2.2.The Efficient Market Hypothesis ……….… 9
2.3.Endorsements……….. 12
2.3.1 Communication Medium.……… 13
2.3.2 Motive of the Endorser.………... 15
2.3.3 Source of Influence.……… 16
2.3.4 Market Complexity……….… 16
2.4. Effect of Endorsements on Trade Volume……….…… 18
2.5. Effect of Endorsements on Market Price………... 18
2.6. Moderation by Market Capitalisation.……… 20
3. RESEARCH DESIGN & ANALYSIS………..… 23
3.1. Data and Sample Description…...………..… 23
3.2. Empirical Methodology & Statistical Analysis……...………... 25
4. RESULTS………..… 29
5. DISCUSSION………...… 34
6. CONCLUSION……… 38.
6.1. Limitations & Future Research……….. 40
REFERENCES………..… 43
iii ABSTRACT
This study extends the growing research on the economic consequences of endorsements, measured by consequent trade volume and equity valuation. The results were inspected for signals of market efficiency, both across the whole sample and between assets within the sample using the ubiquitous efficient market hypothesis. The framework provides the strata to categorise the market’s maturity based upon the speed accuracy and size of market response. The empirical results were developed from data collected from two sources,
CoinMarketCap.com and YouTube between 08/2017 – 04/2018. The results reveal that the market does not immediately and accurately reflect new public information into asset value through purchasing behavior, instead demonstrating clear signs of market inefficiency consistent of a market lacking semi-strong form efficiency. Further weight is added to this assertion by the moderating effect of market capitalization, confirming the suspected presence of the ‘small-firm effect’. The finding suggests that smaller-capitalisation assets within the market are experiencing greater mispricing than large-capitalisation assets. As such, there are profound implications for the optimum investment strategy of information traders to maximise their risk-adjusted returns and outperform the market.
Keywords: Efficient Market Hypothesis; Cryptocurrencies; Endorsement; SMI; Social Media;
- 1 - 1. INTRODUCTION
“Price is what you pay – value is what you get” – Warren Buffet. (2008)
For investors seeking maximum return, the adage defines the true goal of investment -
to identify assets whose true value exceeds their current market price. Warren Buffet’s (2008)
statement insinuates that financial markets misprice assets and that the route to outperforming
the market is by evaluating the available forms of information to identify and acquire these
undervalued assets, subsequently capitalising upon the comparatively greater return achieved
when the market re-evaluates the asset and price reflects true value. The Efficient Market
Hypothesis (EMH) is the preeminent framework to categorise a market’s efficiency, the
levels defined by the forms of information that are effectively and immediately reflected by
market price. Malkiel (2003, p. 59) defines a fully efficient market as one where “prices fully
reflect all known information, and even uninformed investors … will obtain a rate of return as generous as that achieved by the experts.” By conducting event studies on a market’s movements ex-ante, insight can be gleaned into whether the market is fully, semi, weak-form
or lesser efficient. Consequently, this scholarship aims to conduct one of these examinations
on the cryptocurrency market by using endorsements as the subject study.
The term endorsement initially evokes contemplation of high-cost marketing
campaigns from industry giants contracting global icons to promote products and brands
through mass media advertisement. While this is one aspect of the subject, the breadth of the
application is far wider and more nuanced. The most commonly appropriated definition of
celebrity endorsements was presented by McCraken (1989, p. 310) who defines the act as
"any individual who enjoys public recognition and who uses this on behalf of a consumer
good by appearing with it in an advertisement.” The definition is amorphous by design and
- 2 -
intervening years. The preferred communication medium has continually evolved to meet the
consumers where they are most present, first through radio, television and, more recently, on
social media platforms. In the modern era, social media influencers (SMIs) have become a
new sub-category of celebrity, using their online platforms as tools to endorse the businesses
they support to wide audiences, often as unpaid unilateral earned media (Stephen & Galak,
2012; Colicev et al., 2018). It is vital that the efficacy of these new endorsements to attain
tangible economic results is understood comprehensively.
The positive effect of endorsements can be categorised as either first-order consumer
sentiment or second-order economic consequence when the former is converted to purchasing
behaviour. Quantifying the tangible effect of endorsements on a firm's bottom line has
become a priority within the recent study (Bergkvist & Zhou, 2016), measured by either the
additional demand generated or the increase in a business valuation attributed to the event
(Clark et al., 2002; Agrawal & Kamakura, 1995; Elberse & Verleun, 2012). Effort to identify
the second-order economic consequences of online celebrity endorsements across established
industries has ensued, largely confirming a positive relationship within a wide-variety of
settings (Clark et al., 2002; Agrawal & Kamakura; Clark et al., 2008); however, the efficacy
of expert SMI endorsements to achieve positive economic effect in the unique and
rapid-growing industry of cryptocurrencies has yet to be explored.
Cryptocurrency is the colloquial term assigned to a group of “digital asset that are
constructed to function as a medium of exchange, premised on the technology of
cryptography, to secure the transactional flow, as well as to control the creation of additional
units of the currency” (Chohan, 2017, p. 1). The first cryptocurrency, Bitcoin, was developed and released as a transactional currency in 2009, its price determined by the value attributed
by the free market. Since then, an innovative industry has formed around the proprietary
- 3 -
numerous sectors utilising the technology’s unique core attributes. The market value of the cryptocurrencies central to these projects has grown exponentially, reaching a peak of nearly
one trillion dollars in 2018 (Wei, 2018). As an ecosystem has grown around these projects,
social media influencers specialising in cryptocurrency have risen to prominence, regularly
using their platforms to endorse their favoured projects; this raises the question of whether
their support has significant impact on demand for the respective asset and subsequently
influences its associated value.
Academic research has understandably struggled to keep pace with the rapid growth
of the nascent cryptocurrency industry and rigorous academic work is currently limited.
Research attention has largely focused upon a few critical topics, most notably for this study,
market efficiency (Urqhart, 2016; Alvarez-Ramirez, Rodriguez & Ibarra-Valdez, 2018; Wei,
2018; Bartos, 2015) and token valuation. Studies assessing token valuation (Alabi, 2017;
Wheatley et al., 2018; Van Vliet, 2018) have analysed the accuracy of predictive models,
such as Metcalfe’s law, to forecast future price movement. This type of study delivers one perspective to the efficiency discussion focusing upon whether historical data can be
leveraged to accurately predict market collapse, although the research neglects to
acknowledge the market’s response to the two other forms of information - public and private. As such, these studies can only confirm if the market shows signs of weak-form
efficiency, but not whether the market meets the requirements of the EMH’s higher levels.
Within the stock market, significant research has been directed towards identifying
and explaining market anomalies such as the small-firm effect, January effect and
day-of-the-week effect and the implicit associations for investment behaviour (Fama & French, 1993).
As a new asset class, the same investigations have not been conducted, aside from the
January effect (Decourt, Chohan & Perugini, 2017), to confirm whether the anomalies
- 4 -
of these anomaly are hallmarks of inefficiency, confirming that a market has not reached
strong-form efficiency. The firm effect suggests greater mispricing of
small-capitalisation assets and would confirm that the assets within the cryptocurrency market have
unequal efficiency.
Alternatively, studies (Urqhart, 2016; Alvarez-Ramirez, Rodriguez & Ibarra-Valdez,
2018) focusing upon cryptocurrency market efficiency have assessed the signals of a given
cryptocurrency’s efficiency tested in isolation, extrapolating the results to the wider market. The studies make an assumption that the individual assets within the market have close
autocorrelation and equivalent information efficiency. However, a gap still remains to
analyse how a large body of cryptocurrencies responds to a specific form of disruptive
information by conducting event studies. The as yet unexplored methodological approach
incorporates the response of a collection of different cryptocurrencies to deliver a better
representation of the market’s response overall.
Lastly, testing the efficacy of endorsements to achieve economic consequence has
become more common within the mainstream endorsement literature, yet is still dwarfed by
the overwhelming majority of research focused upon consumer sentiment (Bergkvist & Zhou,
2016). Firms with stocks traded on the open market have been the exclusive subjects of
academic analysis. The relevant event studies have measured endorsements effect on both
consumer purchasing of product and investor purchasing of equities, returning consistent
positive correlation. However, cryptocurrencies provide an entirely new market to extend
these lines of enquiry through a robust quantitative analysis. The new market represents an
amalgamation of conditions outlined below that are traditionally expected to result in an
intensified market response, yet the effects on sales volume and asset price have yet to be
- 5 -
Taking the evident research gap and aim of this study into consideration, this
scholarship contributes to established theories in several ways. By examining the economic
consequences of SMI expert endorsement in the cryptocurrency market, this scholarship aims
to simultaneously shine a light on the overlooked aspects of extant finance and endorsement
literature and concurrently deliver valuable, actionable interpretation of the results to inform
the stakeholder of the wider implications. As an emerging market, the contributions of this
paper are increasing with adoption over time, particularly if the market experiences the
widely predicted ambitious growth projections in the coming years. The answers herewith
satiate one aspect of long-held cryptocurrency market curiosity currently dominated by
hearsay – do cryptocurrency YouTubers achieve greater returns by shilling their bags?
In addition, by interpreting the cryptocurrency market’s response to endorsements
through the lens of the efficient market hypothesis, this study’s primary underlying
framework is extended. As a form of public information, should the endorsement events be
effective and immediate reflected by the asset valuation, the EMH theory states that the
market has exceeded semi-strong form efficiency (Fama, 1970; Malkiel, 2003).Failure to do
so will confirm that the market remains short of the required criteria to be considered equally
as mature as the stock market. Regardless of the result, the EMH will have been applied to
classify a new financial market’s efficiency and suggest the consequent implications for investor decision-making behaviour.
While the EMH gives the forthcoming results context, this study also furthers the
existing cryptocurrency market efficiency debate in two further regards. Firstly, through
adopting event study methodology, the insight into the market’s response at the event level complements existing trend - and macro-level analyses. The event study approach is the first
of its kind in this field, certainly within the cryptocurrency market efficiency discussion, and
- 6 -
market’s efficiency. Considering the extensive literature devoted to this topic within the stock market and the ongoing contentiousness of the topic, there remains a pressing need to
continue the academic discussion of cryptocurrency market efficiency in order to build an
equally developed understanding of the market.
By analysing market response in regard to firm size, the scholarly discussion of
market efficiency is given one final contribution. The approach will clarify whether smaller
assets have been experienced greater undervaluation and thus achieve greater risk-adjusted
return when faced with the same positive endorsement event. If this is the case, the assets
within the market are unequally mature and thus offer disparate opportunities for information
traders. By seeking this answer, this paper aims to extend the ‘small-firm effect’ theory,
which states that the return from small-capitalisation assets is greater than those from
large-capitalisation assets (Fama & French, 1993). The effect was first recognised in the stock
market, but by measuring the magnitude of purchasing response to endorsement information,
the generalisability of the theory across markets will be elucidated.
The aspiration of this paper is to not only to acknowledge whether cryptocurrency
buyers value the input of SMI endorsers, but whether the introduction of new public
information to the market represents an opportunity to achieve above-market risk-adjusted
return from information trading. If this opportunity does exist, whether the greatest profits are
likely to be achieved by directing limited attentions resources towards identifying
undervalued large-capitalisation assets or small-capitalisation assets.
In order to most effectively approach the driving objectives of this thesis, the
following literature first present cryptocurrencies as an emerging digital asset with shared
attributes of both products in its role as a utility asset and with equities in its likeness to a
security asset. The efficient market hypothesis will then be introduced as the overarching
- 7 -
endorsements will be introduced and their appropriateness as the subject of analysis will be
justified, using the existing research on the economic consequences stemming for
endorsements to craft the first two hypotheses. Subsequently, market capitalisation is
considered as a potential source of moderating effect, informing the final hypothesis of this
study. The chosen approach to data collection and empirical methodology shall be elaborated
upon before presenting the consequent statistical results. Detailed interpretation of the findings
will then be presented within the discussion section before a concise concluding reflection on
the academic contributions; limitations and recommendations for the future directions for
successive study are offered as summation.
2. LITERATURE REVIEW & HYPOTHESES DEVELOPMENT
2.1. Cryptocurrencies.
The first decentralised, public blockchain-based cryptocurrency, Bitcoin, was released
as an open-source protocol in 2009 by a developer(s) under the pseudonym Satoshi
Nakamoto (Urqhart, 2016). Since then, the original code has been adapted and emulated by
over 1500 global projects seeking to appropriate the advantages of the underlying technology
to disrupt existing business models, primarily through greater security, transparency,
efficiency and decentralisation (Wei, 2018). As a means of project finance, the majority of
these businesses have distributed tokens to the public market and collectively, the tokens
have become an emerging financial asset class (Wheatley et al., 2018; Wei, 2018), attracting
substantial global attention for high volatility and repeated exponential price movements
(Adhami, Giudici, Stefano & Martinazzi, 2017). In 2018, the cryptocurrency market reached
a peak of over USD 800 Billion built upon investor speculation and exuberance over its
future potential (Wheatley et al., 2018), while the underlying projects largely remaining
- 8 -
As the asset class has gained recognition, governments globally are attempting to
assign the market appropriate regulation under the existing legal framework.
Cryptocurrencies were largely distributed through a new and unregulated process known as
an Initial Coin Offering (ICO) as self-proclaimed utility assets (Clayton, 2018; Adhami,
Giudici, Stefano & Martinazzi, 2017). The associated businesses have claimed that the tokens
hold value in use and are thus vouchers to be redeemed in return for access to the underlying
blockchain network. In this regard, the purchasers of cryptocurrencies can be perceived as
consumers and the asset class can be described as a utility with all associated legal
connotation.
Many regulators are promoting an alternate perspective, warning that the assets have
been distributed and accumulated more closely to traditional stock equities (Fagel, 2018),
where the token’s purpose is partially value in ownership (Clayton, 2018) due to the fact that at this early stage in the market development many cryptocurrencies cannot currently be
exchanged for utility. Instead, they are being used simply as a speculative vehicle and proxy
for a stake in the business (Adhami, Giudici, Stefano & Martinazzi, 2017). Through this lens,
the majority of cryptocurrencies arguably meet conditions of a security asset and closely
resemble stocks, since the cryptocurrency buyer can be considered primarily as an investor
purchasing an ownership right of part of the business’s digital assets (its tokens). In reality, cryptocurrencies likely fall somewhere between the two categories, where the archaic
existing legal framework cannot sufficiently classifythe complexity and novelty of the new
technology market under one encompassing label. Considering these two contrary
perspectives, some consumers purchase tokens with intent to use the blockchain in the future
and on the understanding that there is no ownership right, while others accumulate with intent
to speculate and consider their purchase as an investment stake in the business’s future success (Clayton, 2018).
- 9 -
Therefore, when developing hypotheses on the effect of SMI endorsements on
purchasing behaviour, endorsement literature relevant to persuasion of consumer purchasing
and investor purchasing is considered. While the two actions are closely related there is
subtle distinction in the success of endorsements to evoke each form of behaviour (Elberse &
Verleun, 2012). Considering the asset class as a form of equity and the buyers as investors,
the wider implications of purchasing behaviour can be analysed in respect to market
efficiency categorised through the ubiquitous Efficient Market Hypothesis framework.
2.2. The Efficient Market Hypothesis.
The seminal papers of Fama (1965) and Samuelson (1965) provide the first widely
accepted compelling evidence that the path of stock prices follow a so-called ‘random walk’.
Their works state that the stock price of tomorrow is independent of the direction and
magnitude of movement from the days preceding, establishing the necessary groundwork for
the subsequent development of the Efficient Market Hypothesis. The EMH remains a
cornerstone of finance literature to this day, and acts as the overarching theoretical
framework of this scholarship. Testing a market’s information efficiency at the event holds two primary purposes; firstly, to identify whether investors allocate value and respond to the
chosen information event - in this case endorsements (Mathur et al., 1997; Agrawal &
Kamakura, 1995; Clark et al., 2008), and secondly, to analyse the speed and accuracy of
market response to identify the categories of information the market is likely to misprice,
providing opportunities for investors to then capitalise and achieve consistent above-market
risk-adjusted return (Tellis & Johnson, 2007).
When a financial market meets the full requirements of the EMH, all relevant
information is freely available to the market, arrives without warning and is immediately
- 10 -
reflect all historic, public and private information (Fama, 1970) and the opportunity to return
consistent above-market risk adjusted returns (abnormal return) through information trading
would be eliminated entirely (Lo & MacKinley, 2011). If a market fails to achieve the full
EMH criteria, two further levels of lesser market efficiency are defined, weak and
semi-strong form, which act as additional benchmarks to categorise a market’s maturity. Weak form markets accurately reflect all historical data exclusively, and semi-strong markets
furthermore reflect all public information (Fama, 1998; Tellis & Johnson, 2007). In these two
latter forms of market efficiency, investors can hypothetically recognise the information not
accurately reflected by price in order to acquire the asset below true value.
While the EMH strata were originally developed to categorise the stock market’s response to dividend announcements and stock splits (McKinlay, 1997), they are now
liberally applied to assess a market’s response to alternate events to gain perspective on the efficiency of numerous broader asset markets (Degutis & Novickyte, 2014; Elberse &
Verleun, 2012). It is widely regarded that the maturity of a market is directly related to its
efficiency; young markets prior to regulation have overwhelmingly proven to lack objective
data and exhibit lower levels of efficiency (Healey & Palepu, 2001; Hirschlerfer & Teoh,
2009; Luo, Zhang & Duan, 2013). In the novel cryptocurrency market, much early academic
effort (Urqhart, 2016; Wheatley et al., 2018; Alvarez-Ramirez, Rodriguez & Ibarra-Valdez,
2018) has been directed towards establishing the market’s current level of efficiency. The question is complex and requires perspective from multiple complementary lenses to provide
a holistic and informed assessment, with no single determinative measure.
Studies analysing the cryptocurrency market’s overall efficiency across alternate timeframes have returned conflicting results on whether the market meets weak-form
efficiency; in the first comprehensive test of Bitcoin against the EMH over the period
- 11 -
historic information. However, in a follow-up study of more recent data, Alvarez-Ramirez,
Rodriguez and Ibarra-Valdez (2018) comment that the Bitcoin market has matured
substantially in a short period of time and is now approaching the requirements of
semi-strong form efficiency. Until now, prevailing consensus is that the market still falls short of
accurately and immediately reflecting public information (Wei, 2018; Urqhart, 2016;
Brauneis & Mestel, 2018), although Bartos (2015) contests the prevailing wind, providing
limited evidence to the contrary.
Furthermore, when price follows predictable patterns based on historical data the
market is less than weak-form efficient. In the cryptocurrency market, research (Alabi, 2017;
Van Vliet; 2018; Alabi, 2017; Wheatley et al., 2018) has demonstrated compelling evidence
that periods of overvaluation have been closely linked to deviation from Metcalfe’s law. The
theory has previously been applied as an effective means of estimating the value of alternate
digital networks (Alabi, 2017) and states that the true value of a network is proportional to
the square of the nodes of the network (Metcalfe, 2013). In the four periods of significant
market correction since 2009, cryptocurrencies’ market capitalisation strayed substantially from Metcalfe’s estimated valuation before promptly returning to the trend soon after (Van Vliet; 2018; Alabi, 2017; Wheatley et al., 2018). The implication is that price has followed a
somewhat predictable boom-bust cycle for those paying close attention to the historical data.
This has enabled them to capitalise from periods of excessive growth, recognise extended
market conditions and exit the market ex-ante, before re-joining at lower prices and
consequently outperform the market (Wheatley et al., 2018). If adherence to Metcalfe’s law is
more than coincidence, the common consensus among the aforementioned academic scholars
is that the cryptocurrency market has still to reach weak-form efficiency.
While both market-wide cryptocurrency EMH research and early-trend analyses
- 12 -
of inefficiency, further evaluation from alternate perspectives will add greater validity to the
argument or dispute the prevailing consensus. In the stock market, event studies have been
commonly applied to observe the market’s reaction to a specific stimulus, providing an as yet unconsidered micro-perspective to the efficiency discussion (Farrell et al., 2000; Ding et al.,
2010; Tellis & Johnson, 2007).
Expert celebrity endorsements are one form of information proven to be a catalyst for
market movement within equity markets (Farrell, 2000; Clark et al., 2002) and thus as an
appropriate subject to study market efficiency in the new financial market of
cryptocurrencies. The subject is particularly pertinent to this particular market due to the
prevalence and widespread perception of its influence on demand for the cryptocurrencies
consequently its likelihood of affecting the respective market value.
2.3. Endorsements.
Under the belief that a business’s economic performance is partly driven by inducing positive effects in the mind of consumers (Erdogan, 1999), businesses have embraced
endorsements as a pervasive means to obtain desired consumer sentiment and protracted
academic attention has meticulously studied its efficacy to do so (Erdogan, 1999; Agarawal
& Kamakura, 1995; Bergkvist & Zhou, 2016). Much early attention focused upon the
capability of positive endorsements to influence consumer brand image, measured in the form
of brand perception (Colicev et al., 2018; Hong, Pavlou & Dimoka, 2012), brand attitude
(John et al., 2017) and brand awareness (for comprehensive review, see Colicev et al., 2018;
Friedman & Friedman, 1979; Kamen et al., 1975; Ohanian, 1991; Erdogan, 1999; Bergkvist
& Zhou, 2016). However, while brand image can be described as the first-order positive
effect of endorsement activity, businesses only achieve greater economic performance if
consumer sentiment converts to second-order consequence – consumer purchasing behaviour
- 13 -
(2018) state, in reality the relationship between consumer sentiment and sales is not a directly
linear relationship; to understand the true efficacy of endorsements within a market of
interest, the explicit consequences of the technique on both categories of economic value
must be considered. Focus upon the efficacy of endorsements to evoke economic
performance has become a scholarly priority in recent years (Bergkvist & Zhou, 2016; Ding
et al., 2010) and can be categorised into the quantifiable metrics of trade volume and equity
valuation as considered below.
The term ‘endorsement’ is a polysemy that denotes many unique forms of relationship between endorser, endorsed business and consumer. The minutia of this bond affects the
likelihood of an endorsement achieving significant economic consequences. The relationship
can be effectively categorised via four major dimensions: communication medium, motive of
the endorser, source of influence and market complexity.
2.3.1. Communication Medium.
Just as McDonough (1995) observed that radio and television gave rise to a new wave
of celebrity endorsers, the Internet has given influential public platforms to a new wave of
Internet celebrity endorsers colloquially known as social media influencers (SMIs). SMIs are
defined as a “new type of independent third-party endorser who shape audience attitudes through blogs, tweets, and the use of other social media” (Freberg et al., 2011, p. 1), who meet the widespread notoriety and ‘recognisability’ required to be considered a celebrity under the prevailing definition (Bergkvist & Zhou, 2016, p. 643).
Much of extant academic study directed towards the efficacy of celebrity
endorsements (for review, see Erdogan, 1999; Friedman & Friedman, 1979; Kamen et al.,
1975) is universally pertinent across media platforms. Consumers follow the same cognitive
process preceding purchasing decision and endorsements shape consumer brand perception to
- 14 -
communication channel (Wood & Burkhalter, 2014; Colicev et al., 2018). However, while
the fundamental approach remains broadly equivalent, scholars (Stephen & Galak, 2012;
Wood & Burkhalter, 2014) propose that digital platform endorsements are more effective at
achieving positive economic performance per consumer impression than traditional media
channels; several justifications for this claim have been presented.
Social media enable more targeted communication directly to specific groups of
followers, who by design have a vested interest in the opinions of the influencers they follow.
The nature of communication between social media influencers and consumer is high
frequency and two-way, creating a deeper connection between the parties than traditional
uni-directional forms of communication (Stephen & Galak, 2012). Over time, social capital
unique to digital endorsers is built and can be leveraged effectively when later endorsing
products. The established parasocial relationship increases perceived endorser trustworthiness
(Wood & Burkhalter, 2014), reducing uncertainty and hence removing a critical barrier to
consumer purchasing behaviour.
Another cogent explanation for the observed dichotomy is that social media
endorsements act later in the purchasing decision process. When considering the decision
journey through the ubiquitous AIDA model (for review, see Smith & Swinyard, 1982),
where TV and print endorsements have proven successful at influencing consumers at the
attention and interest stage of the buying process, consumers seek out relevant social media
content as a means of gathering the necessary diagnostic information to confirm their
decision prior to the action stage (Luo, Zhang & Duan, 2013; Colicev et al., 2018). As such,
social media impression can rightfully be expected to return greater purchasing behaviour
(Stephen & Galak, 2012).
Regardless of whether digital platform endorsements influence purchasing behaviour
- 15 -
information, both lines of academic argument confirm that social media-based endorsements
have attributes that increase the scale of purchasing behaviour over traditional
communication channels.
2.3.2. Motive of the Endorser.
Traditional endorsements have overwhelmingly represented a financial transaction,
whereby the support of an endorser is exchanged for monetary recompense. In the vernacular
of endorsement literature, the relationship is paid media (Wood & Burkhalter, 2014).
However, social media platforms have enabled traditional and social media celebrities to
easily, effectively and cheaply endorse the businesses they truly support (regardless of
payment) direct to their followers as unilateral ‘earned media’, no longer requiring the communication platform formerly provided by third parties with biased interests (Stephen &
Galak, 2012). The primary distinction between the two forms of endorsement is the
endorser’s motive.
Whereas paid media are driven at least partially by payment, earned media are driven
exclusively by an endorser’s voluntary and impartial opinion (Wood & Burkhalter, 2014). Consumers have been shown to respond favourably to earned media, since the endorser is
independent (Colicev et al., 2018). Bergkvist and Zhou (2016) hereby remark that the
trustworthiness and credibility that are key elements of earned media make such
endorsements particularly powerful mechanisms to influence both consumer sentiment, but
critically also consumer action; these positive effects tend to be weaker in
financially-dependent situations since consumers are subconsciously aware that the celebrities have been
paid for their endorsement and their support is based on a self-serving motive.
- 16 -
2.3.3. Source of Influence.
Within the endorsement research, scholars have categorised endorsers as either
celebrities or experts (Ohanian, 1991; Atkin & Block, 1983; Biswas et al., 2006). The
delineation is an attempt to establish the personal attributes driving consumer response and to
isolate the effective forms of endorsement from the ineffective. However, the terms are not
mutually exclusive nor are they collectively exhaustive. A celebrity can possess expertise in
the endorsed industry and equally, experts may and often do choose to endorse products
beyond the scope of their expertise (Wood & Burkhalter, 2014). More germane are the
alternate interchangeable classification of fit, congruence or belongingness, that are used by
studies of the match-up hypothesis (Till & Busler, 2000; Bergkvist & Zhou, 2016), that
propose that an endorser’s effectiveness will be directly linked to the consumer’s perception of their appropriateness for the task, affecting the resulting efficacy of the endorsement.
When a celebrity endorser does not have recognisable relevant expertise, the body of
literature unanimously accepts that the endorser’s attractiveness and likability are the sources engendering consumer action, noting that consumer desire to emulate the celebrities drives
results (Atkin & Block, 1983; Till & Busler, 2000; Ohanian, 1991). More salient to this
study, when a celebrity has relevant expertise, the match-up hypothesis states that influence is
instead drawn almost exclusively from their superior knowledge (Ohanian, 1991; Malik &
Sudhakar, 2014). Academics have determined that when consumers and investors face
purchasing decisions and do not possess strong personal conviction, judgement is regularly
deferred to the recommendation of a relevant expert (Malik & Sudhakar, 2014; Till & Busler,
2000; Tetlock, 2007).
2.3.4. Market Complexity.
Under conditions of high complexity, both consumer and investor uncertainty follows.
- 17 -
expert advice holds greater weight to steer consumer purchasing behaviour. The same study
confirmed that the initial choice of consumers is only altered subsequent to expert investment
20-30% of the time unless uncertainty is high. In contrast, in markets where uncertainty is
high, decision-makers actively seek out and act upon secondary information from experts
(Sniezek, Schrah & Dalal, 2006). One context where expertise has demonstrated a
consistently large influence is when informing consumer technology-purchasing decisions
(Tellis & Johnson, 2007).
When responsibility for a decision is shared, the additional party can be held partly
accountable for decisions deemed sub-optimal in retrospect (Armstrong, 1980; Bonaccio &
Dalal, 2006). Whilst evaluating the returns from an investment expert’s endorsed portfolio of
high-tech equities against an exchange traded fund, Tellis & Johnson (2007) found that the
expert portfolios achieved 14% greater ROI. The evidence provides some justification to
validate the greater devolvement of responsibility of investment purchasing decisions in high
complexity markets.
When considering the cryptocurrency market, it is clear that the market represents a
high-complexity technology market. As the market has gained mainstream attention, expert
SMIs on the YouTube platform have emerged, gaining popularity and large audiences, and
effectively becoming some of the primary expert celebrities of the industry. They use digital
communication media to endorse cryptocurrencies as earned media, using the long-form
content as a means to provide diagnostic information to justify their endorsements. Based
upon the four key dimensions of endorsement efficacy outlined above, it can be expected that
the magnitude of effect on consumer decision-making is likely to be intensified in this market
over those previously researched. This should be born in mind when considering the
following case studies presented.
- 18 - 2.4. Effect of Endorsements on Trade Volume.
The evidence of endorsements to evoke consumer-purchasing behaviour is largely
anecdotal since isolating the effect of the endorsement event has proven exceedingly
challenging (Agrawal & Kamakura, 1995). However, several studies have quantitatively
measured the effect through time-series analyses and found strong correlation between
endorsement events and an immediate increase in trade volume. For instance, when studying
expert endorsers with product-endorser congruence on a sample of 51 unique events, Elberse
and Verleun (2012) discovered that sales increase by an average of 4% in response to positive
endorsements and decreased following negative endorsements. Similarly, Grange (2001)
found that celebrity expert endorsement led to a 2.9% increase in the share of market sales,
results then validated by Chung, Derdenger and Srinivasan (2013) whose follow-up study
documented the effect of the sustained endorsement over the subsequent ten-year period.
When recently inspecting the existing literature of expert celebrity endorsements to influence
sales volume, Bergkvist & Zhou (2016) conclude that a substantial relationship exists,
although the four outlined dimensions of an endorsement success exert directly influence on
the strength of that relationship.
Considering the prior efficacy of endorsements to produce positive consumer
purchasing behaviour and the optimal conditions to maximise the efficacy of that relationship
in the context of SMI cryptocurrency endorsements, this study posits that:
Hypothesis 1. – Following a positive expert celebrity (SMI) endorsement, the trade volume of the respective cryptocurrency will increase substantially.
2.5. Effect of Endorsements on Market Price.
Considering the total capitalisation of a business is said to reflect the discounted value
- 19 -
2010) it can be expected that the literature measuring the success of expert celebrity
endorsements to generate consumer purchasing response would be mirrored in the price
action of the respective business’s valuation. However, the correlation between the two variables has been relatively unpredictive (Bergkvist & Zhou, 2016) and the share price is
often affected to a lesser extent (Colicev et al., 2018). The phenomenon suggests that despite
the expectation that short-term sales will likely rise following the endorsement, investors
believe that long-term future cash flows will not be impacted enough to warrant significant
re-evaluation of the business’s valuation. Thus, to understand the resulting economic effect of
endorsements, both sales and equity value must be examined individually (Elberse &
Verleun, 2012).
A case study by Farrell et al. (2000) on celebrity-expert endorsements found that a
qualifying endorser successfully increased the market capitalisation of one sponsor while
simultaneously proving ineffective for two additional sponsors over the same period.
Similarly, in a quantitative examination of expert endorsement events, Agrawal and
Kamakura (1995, p.56) confirm the positive impact on stock price immediately following the
endorsement events, claiming that “on average, the impact of these announcements on stock
returns is positive.” Despite their initial statement, the fact remains that the strength of relationship found was extremely marginal (0.44%) (Agrawal & Kamakura, 1995), a result
later supported by the work of Elberse & Verleun (2012) who returned similarly weak results
(0.25%). On the other hand, some exemplar endorsement relationships have shown to have
significant influence on share value in the stock market. The rumour of Michael Jordan’s return to professional basketball immediately increasing the share value of his sporting
sponsors by 2% is one prominent example in the field of athleticism (Mathur et al., 1997).
Once again, the endorser and market context appear to influence the strength of the
- 20 -
suggests that there is a positive relationship between equity value and positive SMI expert
endorsements, although weaker than that the aforementioned relation between endorsement
and sales. I consequently propose that:
Hypothesis 2. - Expert celebrity (SMI) endorsements will moderately increase cryptocurrency valuation.
2.6. Moderation by Market Capitalisation.
When referring to the efficiency of a financial market, overall assessments are based
upon the sum of the findings across all firms observed. However, within the broad statement
of a market’s efficiency, each asset can demonstrate characteristics of higher or lower efficiency than the market norm based upon the businesses unique characteristics such as
market capitalisation (Fama & French, 1993). It is thus an incomplete statement to assert the
opportunity for abnormal return from information trading at the market level, yet not consider
the opportunity from less efficient market participants within the collective. One highly
documented example of an anomaly representing unequal efficiency within the stock market
is the ‘small-firm effect’ (Barry & Brown, 1984; Bondt & Thaler, 1985). The theory states that small-capitalisation stocks have achieved consistent superior risk-adjusted return than
larger-capitalisation stocks (Barry & Brown, 1984); the difference in firm size between
1963-1990 accounting for a 1% divergence (Malkiel, 2003).
The phenomenon suggests that small-capitalisation stocks have experienced periods
of inefficiency due to information mispricing. Information mispricing occurs when investors
are unaware of information that justifies price reflection or have not correctly recognised its
value (Healy & Palepu, 2001; Tirunillai & Tellis, 2012) allowing for shrewd investors
focusing on these under-observed stocks to capitalise on the consequential mispricing and
- 21 -
In order for a security asset to maintain a higher level of efficiency, new relevant and
valuable information must be quickly and accurately reflected by the price of that asset
(Fama, 1970). This is achieved in part by the participation of a large number of self-interested
active investors. However, investors have finite attention and can only commit their scrutiny
to new information regarding a limited number of securities at any one time (Tellis &
Johnson, 2007). When the number of active investors is low, an asset is prone to stray from
its true value (Healy & Palepu, 2001; Tirunillai & Tellis, 2012). Large capitalisation stocks
are known to attract a greater proportion of institutional investor attention due to the higher
liquidity and the relative ease of access to relevant information, while information regarding
small stocks can pass under the radar (Shefrin, 2007; Tellis & Johnson, 2007). As such, the
unrecognised information of small-capitalisation assets is not accurately reflected by price,
and due to the conservative tendencies of humans, the assets tend to be undervalued when the
judgement is made upon incomplete information (Nelson, 1970). When wider market
attention is directed towards the asset, such as following a major news event, the additional
investors attracted to evaluate the asset leads to re-adjustment of price to reflect the assets
true value. The investors that purchased the asset during the period of mispricing
subsequently receive above-normal return from the market correction.
The theory is contentious and significant scholarly attention (Malkiel, 2003; Fama &
French, 1993) has suggested that the higher return is simply a product of risk factors not
accounted for within the traditional finance models. Nevertheless, those academics have
failed to convincingly identify the source of this additional risk and hence the argument lacks
conviction. It is rational to expect that the same natural tendencies of investor attention that
lead to the small-firm effect in the stock market could plausibly recur within the alternate
financial market of cryptocurrencies. This logic is supported by the fact that other
- 22 -
have been identified (Decourt, Chohan & Perugini, 2017). In the first academically rigorous
analysis of efficiency of 456 cryptocurrencies, Wei (2018) describes signs of inconsistent
efficiency between tokens which would be expected if the small-firm effect prevails.
Diagnosing the source of inefficiency discrepancy fell beyond the bounds of Wei’s (2018)
study, although the small-firm effect may account for part or all of the unexplained finding.
By grouping the market’s response to positive SMI endorsements by firm-size, the strength of market response can be analysed for inconsistency between small and
large-capitalisation cryptocurrencies. Since smaller assets traditionally receive lesser investor
attention leading to undervaluation, the increased attention following a major endorsement
event can be expected to result in a greater response to the new information as the market
re-establishes the asset’s true value. Based on this logic, I hypothesise:
Hypothesis 2a. - Firm-size negatively moderates the strength of market response to celebrity expert endorsements. Hereby, larger capitalisation firms will experience a smaller effect on asset valuation following the endorsement event due to the small-firm effect.
Figure 1. Overall Theoretical Framework
- 23 -
3. RESEARCH DESIGN & ANALYSIS
3.1. Data and Sample Description.
IV: YouTube SMI endorsements. The explicit endorsement of a cryptocurrency by content
creators specialising in cryptocurrencies on the YouTube video streaming platform acted as
the independent variable of this study. YouTube represents the highest traffic website for
long-form amateur content creation and has been utilised as one medium for perceived
cryptocurrency experts to interact with the tech-friendly market participants. Retrieving raw
data from social media platforms has become a common source for academic analyses in
recent years under the wider topic of market sentiment analysis (Tetlock, 2007; Colicev et al.,
2018; Wood & Burkhalter, 2014).
Within the thriving YouTube cryptocurrency community, the nominal data represents
the 20 most recent explicit endorsement videos given by the four largest content creators who
actively endorse projects as ranked by total subscriber count (Appendix 1). The content
creator sample was non-probabilistic and selected upon the availability of the necessary data
and the size of audience in order to meet the requirement as both perceived experts and a
celebrity within the field, based upon the combined aforementioned definitions of celebrity
endorsement (McCracken, 1989) and social media influencer (Freberg et al., 2011). The final
sample represents a cross-section of content styles and alternative audiences within the
community and hence, any purchasing response should be representative of the reaction
across further digital platforms and alternate cryptocurrency endorsement events.
The time frame studied was the six months preceding this study and the additional
two months prior to data testing (08/2017 – 04/2018). The time frame is an extended but
recent duration, selected in order to present the most up to date results on the subject matter,
since the market is evolving extremely rapidly (Alvarez-Ramirez, Rodriguez &
- 24 -
(nominal data) and the full list of videos used can be found in Appendix 2. In total, the initial
sample included 80 relevant clips, whereby the necessary data to conduct an event study on
each unique video was subsequently collected. After adjusting for incomplete dependent
variable data, 63 videos remained.
DVs: Trade Volume and Market Price. Trade volume (DV1) and market price (DV2) are the
two most common forms of dependent variables measured by academic analysis of the
economic consequences of endorsements and were both adopted for this study. All dependent
variable data were retrieved from CoinMarketCap.com, the leading independent aggregator
of cryptocurrency trading data (Coinmarketcap.com, 2018). The website ranks 80th on the
Alexa 500 (Alexa.com, 2018) list of top traffic sites worldwide and is a highly legitimate
source within the cryptocurrency community. The two dependent variables were both
measured to within 15 minutes accuracy recorded at trading day close (UCT +2.00). The
trade volume is measured as a time-series of the 24-hour USD values cumulative from the
8900+ exchanges worldwide (Wei, 2018), and the cryptocurrency price (USD) was measured
as the average value between the same body of exchanges over the same 24-hour period.
Each trading exchange may present marginally different price due to unique demand and
supply at any moment in time; however, due to high-frequency traders consistently
identifying and capitalising upon the arbitrage between exchanges, the price remains within a
tight range. Therefore, using CoinMarketCap.com, which presents the average value across
trading exchanges is the most effective means of documenting the accurate market price of a
cryptocurrency.
CoinMarketCap.com provides public access to their API (CoinMarketCap.com,
2018), whereby a simple data crawler program was applied to extract market data of all
currently listed cryptocurrencies over their complete listing history; these ratio data were then
- 25 -
videos and to reduce the time-series data to a 40-day observation window surrounding each
of the YouTube endorsement events. From the 40-days observation window, a 28-day
estimation window, 7-day event window and 5-days post-event window were selected. All
cryptocurrencies to receive an endorsement without complete price and trade volume history
spanning the entire 40-day window were removed from testing, since the data was
insufficient to provide results representative of the population. Since some of the
endorsement videos contained multiple endorsements, a total 229 individual event studies
with complete information were conducted across the two dependent variables in order to
provide strong legitimacy to the findings.
The data used to test cryptocurrency price (DV2) was then re-organised into two
groups based upon market capitalisation of the endorsed cryptocurrency (Appendix 3) in
order to examine hypothesis 2a. The top 100 highest total value projects formed group one
and consisted of 71 endorsement events and the 44 remaining endorsed cryptocurrencies
formed the small-capitalisation group two.
3.2. Empirical Methodology & Statistical Analysis.
The quantitative event study methodology has been widely adopted, initially and most
frequently in the field of finance. However, the methodology has also been utilised widely by
papers attempting to isolate the economic effect of numerous information events and in a
broad range of asset markets (Clark et al., 2002). The underlying theory of event study
methodology is the Efficient Market Hypothesis, which states that the discrepancy between
the expected and achieved results reflects the market’s response to the event under observation (Agrawal & Kamakura, 1995).
Within the study of endorsements there is an established methodological precedent,
whereby examples of scholarships adopting the event study approach to quantify the efficacy
- 26 -
Kamakura, 1995; Elberse & Verleun, 2012; Clark et al., 2002; Farrel et al., 2000). This study
simply employs the respected event study approach to the novel cryptocurrency market as an
original means of analysis.
The event study methodology works upon the principle of generating a predictive
model from an estimation window, to forecast a firm’s expected performance barring changes
in market conditions and comparing the results to the true observations in the period
surrounding the event (event window). There are several alternative means of establishing the
expected values from an estimation window including the CAPM and constant-mean returns
model (McKinley, 1997).
In this instance, the market returns model was chosen, whereby the actual asset price
is ex-post compared with an index of the entire cryptocurrency market returns on the same
day. There is an implicit assumption that the returns of the asset will maintain a linear
relationship to the market index (McKinlay, 1997). By incorporating a market index,
influential events that affect all assets within the market, such as regulatory news, are
accounted for by the model and the influence of an event specific to an isolated asset can be
more effectively quantified. The expert SMI endorsements of this study were of individual
cryptocurrency assets, thus, keeping all shared variables equal through the market model was
an appropriate methodological choice to construct the estimation model.
Microsoft Excel software was used to conduct all statistical analysis. The data was
first examined for skewness using the = SKEW function, returning values of 0.444 (H1) and
0.137 (H2 & H2a). The results meant no further transformation of the data was necessary
before beginning the main body of analysis.
Firstly, all cryptocurrency and market index values were transformed into their rate of
- 27 -
Rate of change formula: (𝑋𝑖𝑡 − 𝑋𝑖(𝑡 − 1))⁄𝑋𝑖(𝑡 − 1)
The intercept (∝) was found by executing the = INTERCEPT (all known Xs, all known Ys) formula on the estimation window, where Xs are the actual rate of return achieved
by the observed cryptocurrency and Ys the market index’s rate of return. The same columns
of data (X, Y) were also appropriate for the slope function (= SLOPE) for (𝛽), the coefficient of determination (= RSQUARE) and standard error (= STEYX).
The Expected Return (ER) for all dates of the 40-day observation window were then
calculated through the standard formula (McWilliams & Siegel, 1997) shown below, where:
Rit = the rate of return of firm i on day t,
Rmt = the rate of return on the market index on day t, a = The intercept,
𝛽 = The slope
𝜀𝑖𝑡 = The error term where ∑ 𝑒𝑖𝑡 = 0
𝐸𝑅𝑖𝑡 = 𝛼 + (𝛽 ∗ 𝑅𝑚𝑡) + 𝜀𝑖𝑡
Once the expected returns were known, Abnormal Return (AR) was calculated as the
difference between expected and actual return.
𝐴𝑅𝑖𝑡 = 𝑅𝑖𝑡 − 𝐸𝑅𝑖𝑡
The abnormal return of each day of the 7-day event window, where the endorsement
- 28 -
the cumulative abnormal return (CAR) summarising any abnormal response over the entire
event window was recorded. The further analysis confirmed whether any market response
achieved on a single day (AR) was sustained, reversed or accelerated in the immediate
aftermath of each event.
CAR (-1,5) = ∑ 5−1𝐴𝑅𝑖𝑡
Both AR for each event day and total CAR were averaged across all events in the
sample: AARt = 1 𝑁 ∑ 𝐴𝑅𝑖𝑡 𝑁 𝑖=1 CAAR = ∑𝑡=5𝑡=−1𝐴𝐴𝑅𝑡
The merged values provide a suitable representation of an endorsements impact
across all event observations providing values appropriate for significance testing. The
t-statistics were compared against the two-tailed Z-table for significance at the α = .05.
Once this process was repeated using both sales volume data (H1) and token value
data (H2), the sample of cryptocurrencies were divided into two groups, of either large or
small(er) capitalisation assets based upon their rank within the total 1500+ token market as
described previously. An independent sample t-test was then applied to compare the efficacy
of endorsements to effect market price of small-capitalisation cryptocurrencies compared to
large-capitalisation cryptocurrencies. Once again, the CAAR values were used as the mean
from of each group, the = STDDEV function across the abnormal returns achieved and N the
number of cryptocurrencies within each group. The results were then compared against the
- 29 - 4. RESULTS
Hypothesis 1 suggests a positive relationship exists between positive SMI expert
endorsement events and the trade volume of the respective endorsed cryptocurrency. The
proposed relationship reflects the resulting change in actioned demand from cryptocurrency
consumers and investors stemming from the endorsement.
The t-statistic event days -1 through 5, was measured against the Z table at the 0.05
significance, which resulted in a two-tail significance cut-off point of +/- 1.96. Each value in
Table 1 represents the AAR result of the day averaged across the N of 114 unique
endorsements from the 63 YouTube videos studied.
AAR results for every day of the event window across all endorsements proved
insignificant (see Table 1). On the key event day (t=0), 12/114 events demonstrated
significant abnormal return, half of which were from endorser four (Suppoman). However,
when the results of all 114 events were aggregated, the AAR of event day 0 were extremely;
small 0.262, - 0.238, 0.272 and 0.640 for each endorser respectively and 0.07 overall.
Taking these results to a second level of analysis, the CAAR was calculated across
alternate time periods to present the significance of specific windows within the 7-day event
window. Once again, all results proved insignificant at the 0.05 level. Most notably, the
cumulative results between day -1 to 5 of endorser 2 was the closest value to significant when
accounting for standard error (-1.35) within a field of values elsewise all returning scores
- 30 -
Table 1. Effect of SMI Endorsements on Trade Volume.
- 31 -
Hypothesis 2 suggests a positive relationship between positive SMI expert
endorsements and the market price of the respective cryptocurrency. The proposed
relationship reflects the resulting imbalance of demand for the cryptocurrency against supply,
hence influencing price and demonstrating the influence of endorsers to evoke purchasing
behaviour.
As with hypothesis 1, the AAR of 115 cryptocurrency endorsements were tested for
significance on each of the separate 7 days of the event window at the 0.05 level. All event
days for all endorsers returned non-significant t-values (Table 3). In total 16 of the event
studies had a statistically significant result on the most likely day of significance (t= 0),
although across all 115 events the AAR’s were all extremely small (.003, -.024, 0.101, .062
for each of the four endorsers respectively) leading to reflectively small t-statistics.
When analysing the CAAR of the cryptocurrency valuation all observed timeframes
within the event window (7, 5, 3 and 2 days) return insignificant results (for full detail, see
- 32 -
Table 3. Effect of SMI Endorsements on Market Price.
- 33 -
Hypothesis 3 suggests that the magnitude of effect from positive SMI endorsement
will be greater for small-capitalisation cryptocurrencies than large-capitalisation
cryptocurrencies. The implication is that the increased market attention leads to recognition
of undervaluation stemming from too few market players, leading to a greater adjustment to
re-establish the equity’s true value. The presence of this relationship would suggest divergent
levels of efficiency between cryptocurrencies and evidence that the ‘small-firm effect’ of the stock market is also present in the cryptocurrency market.
Comparing the two sample means (CAAR (- 1, 5)) of 0.159 and - 0.012, the null
hypothesis of no moderating effect was strongly rejected, confirming that endorsement events
lead to a greater positive market response over the 7-day event window for
small-capitalisation cryptocurrencies than large-small-capitalisation cryptocurrencies. The two tailed
p-value is less than 0.0001, and the findings are confirmed at the 95% confidence interval; the
interval falling between 0.097 and 0.245 and does thus not include 0, indicating strong
significance.
- 34 - 5. DISCUSSION
Hypothesis 1. The empirical results strongly reject the original premise that the SMI
expert endorsements increase the sales of the respective cryptocurrency. This finding is
noteworthy and in contrast to prevailing academic theory on the efficacy of endorsements to
evoke consumer-purchasing behaviour. This study confirms that the theory crafted in regard
to the demand for products outside the cryptocurrency industry does not appear to transfer to
the nascent digital asset market of cryptocurrency. The four outlined contextual dimensions
that typically intensify the efficacy of endorsements to produce sales volume would be
expected to exacerbate the strength of resulting effect in the SMI cryptocurrency endorsement
setting. Instead, consumers appear unconvinced to act upon the endorsement advice.
Fundamentally, the SMI content must not contain the ‘diagnostic information’ necessary to reduce decision uncertainty and consequently increase consumer conviction to take
purchasing action. One potential argument is that the social media content is largely focused
upon the potential monetary gain of the asset as opposed to its value in use; therefore,
consumers whose primary aim is to use the asset’s utility are not convinced to take
purchasing action based upon the justification presented.
The finding is imperative as it highlights the inconsistency between the sharing of
positive sentiment towards the digital asset and converting that intangible brand value into
consumer purchasing action. The result validates the recent trend of endorsement scholarship
that do not assume a direct linear relationship between the first (sentiment) and second
(behavioural) order consequences of endorsements. By extending the bounds of SMI
endorsement research to a growing and unexplored context, the novelty of this study adds to
the holistic understanding of the economic consequences resulting from positive SMI
- 35 -
In addition to the primary contribution of identifying that SMI expert endorsements
do not increase cryptocurrency consumer purchasing behaviour, this study inadvertently
contributes further benefit by producing unexpected results and thus, generating an intriguing
and potentially valuable line of future academic inquiry to elucidate the findings. In a
burgeoning field of cryptocurrency academic research where innumerable questions and few
answers exist, providing the groundwork and direction for future academic attention holds
ancillary value to the cryptocurrency literature stream.
The result holds implications for several parties of the endorsement. For the endorser
(SMI), their endorsements do not result in significant purchasing-behaviour from their
followers. Therefore, there is not a heavy burden of accountability for the asset’s subsequent
performance. For the cryptocurrency businesses endorsed, receiving the earned endorsement
should not be seen as a catalyst for significant trade volume of their cryptocurrency and thus
seeking to attract such endorsements should not be at the expense of substantial financial
resources.
Hypothesis 2. To investigate the effect of expert SMI endorsements on
cryptocurrencies as an equity asset, whereby price is determined by investor demand and
supply, the resulting change in asset valuation was measured. Stocks have traditionally
increased in price following positive expert celebrity endorsement news, demonstrating the
USD value attributed by the financial market to the testimonial. When the public information
is received into a semi-strong form efficient market such as the stock market, the effect on
asset valuation is expected to be immediate and accurate.
The results of this study reject the replication of the traditional effect in the
cryptocurrency financial market; therefore, an SMI expert endorsement cannot be expected to
culminate in a consistent increase in the underlying asset value immediately following the
- 36 -
and the Cumulative Average Abnormal Return (CAAR) of the wider event window are both
insignificant, a trend within the data appears to emerge. In over one third of events a
significant price change does occur during the event window; however, the day of that price
action is inconsistent and thus as opposed to highlighting the effect, the results counteract
when measured across the whole sample, in effect causing a paralysis by over-analysis.
Similarly, the price movements observed subsequently reverse during the event window and
are therefore not highlighted by the CAAR statistic. The inherent limitations of the chosen
methodological approach when applied to a market without semi-strong form information
efficiency led to the oversight of this trend by the selected tests of significance.
This finding is arguably the key insight of the study, and although hidden amongst the
data, is not entirely unexpected. When a firm receives positive information in an inefficient
market, market response is commonly delayed, since the information is slow to spread
between market participants and its value is not immediately established, leading to
corrective price movements in the period of time following the initial announcement. The
early work on cryptocurrency market efficiency, as formerly discussed, has consistently
returned results that suggest the market does not meet semi-strong efficiency and thus could
potentially experience the signals observed. In addition, this interpretation of the results is
supported by hypothesis 2a which convincingly confirms that efficiency between
cryptocurrencies is currently unequal.
Consequently, the practical implications of these findings are inconclusive. The
cryptocurrency market likely remains short of semi-strong efficiency and thus investors are
able to use public or even historical data to return above-normal returns through information
trading. Future SMI endorsements are expected to produce a significant price movement