Management Decision

When the investors speak: intellectual capital disclosure and the Web 2.0
Maurizio Massaro, John Dumay, Carlo Bagnoli,
Article information:
To cite this document:
Maurizio Massaro, John Dumay, Carlo Bagnoli, (2017) "When the investors speak: intellectual
capital disclosure and the Web 2.0", Management Decision, Vol. 55 Issue: 9, pp.1888-1904, https://
doi.org/10.1108/MD-10-2016-0699
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

Permanent link to this document:

https://doi.org/10.1108/MD-10-2016-0699
Downloaded on: 27 March 2018, At: 02:15 (PT)
References: this document contains references to 53 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 205 times since 2017*
Users who downloaded this article also downloaded:
(2017),"Relationships between religious work values, sustainable work behaviours and sustainable
energy consumptions: An empirical analysis using Muslim employees", Management Decision, Vol.
55 Iss 9 pp. 1854-1867 <a href="https://doi.org/10.1108/MD-01-2017-0039">https://doi.org/10.1108/
MD-01-2017-0039</a>
(2017),"The consequences of outness: gay men’s workplace experiences", Management
Decision, Vol. 55 Iss 9 pp. 1868-1887 <a href="https://doi.org/10.1108/MD-12-2016-0900">https://
doi.org/10.1108/MD-12-2016-0900</a>

Access to this document was granted through an Emerald subscription provided by emerald-
srm:551360 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald
for Authors service information about how to choose which publication to write for and submission
guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as
well as providing an extensive range of online products and additional customer resources and
services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the
Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for
digital archive preservation.

*Related content and download information correct at time of download.


MD
55,9
1888
Received 5 November 2016
Revised 6 June 2017
Accepted 13 July 2017
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)
Management Decision
Vol. 55 No. 9, 2017
pp. 1888-1904
© Emerald Publishing Limited
0025-1747
DOI 10.1108/MD-10-2016-0699
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/0025-1747.htm
When the investors speak:
intellectual capital disclosure
and the Web 2.0
Maurizio Massaro
Department of Economics and Statistics Sciences,
University of Udine, Udine, Italy
John Dumay
Department of Accounting and Corporate Governance, Macquarie University,
Sydney, Australia, and
Carlo Bagnoli
Department of Management, Ca’ Foscari University of Venice, Venice, Italy
Abstract
Purpose – The purpose of this paper is to investigate intellectual capital (IC) discussions held between
investors using Web 2.0 tools. More precisely, this paper investigates the determinants of IC disclosures
(ICDs) on internet stock message boards (IMBs).
Design/methodology/approach – Four hypotheses were developed and tested through content analysis of
60,996 messages posted on two main IMBs, Yahoo!Finance and TheLion.com, followed by descriptive
statistics and logistic regression testing.
Findings – The findings show that Web 2.0 is bringing new opportunities to disclose IC. Traditional theories,
such as agency, stakeholder, signalling, and legitimacy theory, cannot be applied to the Web 2.0 context.
Therefore, a new approach that focusses more on the personal motivations for disclosing IC is called for. At a
glance, the results show that IC is disclosed on IMBs, and several elements influence both the quantity and
quality of those disclosures. Sometimes “trolls” disturb the dialogue and discourage participation by other
investors. Conversely, online influencers facilitate ICD. To filter messages, the time of posting, the length of the
messages, and the sentiment the messages contain should be considered along with the author of the message.
Originality/value – This paper contributes to the existing literature by investigating the IC disclosed on IMBs.
The findings provide insights about how ICDs are developed using Web 2.0 tools.
Keywords Web 2.0, Disclosure, Intellectual capital
Paper type Research paper
1. Introduction
In intellectual capital (IC) research, most studies use a top-down approach to IC disclosure
(ICD), and scholars employ theories that mainly discuss a company’s role in providing
information to the market (Yi et al., 2011). In these types of studies, annual reports are the
main source of data (Dumay and Cai, 2014). However, contemporary communication
channels are more complex and unstructured than formal corporate reports. Zhang (2016)
argues that the internet and Web 2.0 technologies are redefining how people use media to
disseminate information[1]. To obtain company information, brokers, traders, and investors
often seek web-based information from both paid and free sources (Sabherwal et al., 2011).
Internet stock message boards (IMBs) are a free resource that analysts, traders, and
investors can use to acquire, share, and spread information (Sabherwal et al., 2011),
and IMBs users can be employees, customers, or business experts who contribute by
providing information about companies. IMBs represent a new channel of unstructured and
uncontrolled company information. Therefore, new theoretical approaches are required to
investigate how and why investors can contribute to the ICD process.
This paper advances the IC literature by investigating the impact of unstructured
and interactive IC information using a new theoretical approach based on personal
 behaviour theory (Ngai et al., 2015). The approach examines user-generated messages on ICD and the
IMBs to better understand the role that Web 2.0 tools are playing in ICD. Taking two IMBs, Web 2.0
Yahoo!Finance and TheLion.com, we used content analysis to examine 60,996 messages in
threads relating to the world’s ten most valuable companies. Additional analyses in the form
of descriptive statistics and logistics regression offered further insights into the
characteristics of the users who posted the messages, the posting time, the length of the
messages, and the sentiments the messages expressed. 1889
2. Literature review and research hypotheses
This section illustrates the main debates linking IC and IMBs to outline the research hypotheses.
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

2.1 The theoretical foundations of ICD in Web 2.0

Typically, IC represents human, relational, and structural capital. Human capital is the sum of
the competencies, experience, and skills of an organisation’s people. Relational capital
comprises an organisation’s formal and informal relationships with external stakeholders.
Structural capital is the explicit knowledge embedded within an organisation (Albertini, 2016).
The term IC was first used by Kenneth Galbraith in 1969, but its major developments occurred
in the mid-1990s (Kristandl and Bontis, 2007) with a growing number of studies on the
voluntary ICDs of the new millennium (Guthrie et al., 2012).
The regulated financial and voluntary information relating to environmental, social, and
governance issues companies provide in their annual reports has made them a prime source
for academic research on company disclosures (Dumay and Cai, 2014). However, the way
companies and investors communicate is evolving. The internet is continually providing
new ways to share, discuss, and disclose information (Dumay and Guthrie, 2017), and a
significant amount of that information is now being produced by people external to the
organisation using Web 2.0 applications (Alexander and Gentry, 2014). This information
pertains to the organisation but is not produced under the organisation’s direct control.
As a result, there is a need to update ICD research with approaches that consider the
development of different communication channels beyond company controlled corporate
reporting, including involuntary forms of disclosure (Dumay and Guthrie, 2017).
The need to investigate new communication channels raises questions about the theories
that have previously been used to explain ICD. According to Yi et al. (2011), ICD research
employs four main theories to justify why companies should disclose information, namely,
agency, stakeholder, signalling, and legitimacy theory. According to agency theory, ICD can
reduce the information asymmetry between the principal and the agent, decreasing
opportunistic behaviour and the cost of capital (Singh and Van der Zahn, 2008).
Using stakeholder theory, Deegan and Samkin (2009) state that ICD can foster good
relationships with stakeholders and engender company support and approval. In signal
theory, ICD might be used to send market signals to improve one’s corporate image, attract
potential investors, or decrease the volatility of stocks (Guthrie and Petty, 2000). Finally,
in legitimacy theory, voluntarily reporting the information expected by society may help
managers gain legitimacy (Yi et al., 2011).
Agency, stakeholder, signalling, and legitimacy theory all help explain why companies
should disclose IC information by analysing IC’s costs and benefits from the company’s
perspective. However, these theories are not applicable in the context of Web 2.0. Here, the
information producer is also its user and, hence, the author does not gain direct benefit from
disclosing new information. Thus, there is a need to understand what motivates a user’s
behaviour to produce and disseminate information.
A user’s reasons for participating in Web 2.0 can be understood using theories that also
help us to understand ICD. Among the theories used in social media research, Ngai et al. (2015)
define personal behaviour theory as a group of theories adopted to explain the behaviour
 MD of human participation in social media at an individual level. For example, Hsu and Lin (2008)
55,9 show that the enjoyment that stems from being part of an online community, the ease of a
technology’s use, and personal characteristics, such as altruism, all affect user participation in
blog communities. Similarly, the perceived advantages to users and their trust in the
community affect their willingness to participate in web communities (Chao-Min et al., 2011).
Therefore, we argue that moving from agency, stakeholder, signalling, and legitimacy theory,
1890 to personal behaviour theory may be a better way to understand ICD in the Web 2.0 era.

2.2 ICD in IMBs

Among the different Web 2.0 applications, IMBs represent online communities where investors
read, analyse, and discuss topics related to the stock market (Zhang, 2016). These IMBs provide
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

new channels for ICD where IC information can be reported, disclosed, and discussed.
Understanding how users contribute to ICD through Web 2.0 tools is important because
Yu et al. (2013) show that social media content has a strong relationship to a firm’s stock
performance. The impact of IMBs is also of interest to the Securities Exchange Commission
(2009), which, in 2009, charged a registered representative of the New York stock brokerage
firm Global Arena Capital Corporation for posting fake press releases on IMBs to manipulate
the stock prices of publicly traded companies. However, even without reporting false
information, specific users, usually called influencers, are able to shape investor attitudes
(Freberg et al., 2011). Some examples of information disclosed by investors on IMBs are
provided in the following quotes. In the first post, an investor discloses information through a
direct observation of Apple’s employee skills and their market share. In the second post, a user
discloses information about how an IBM employee feels about his/her company.
Examples of information disclosed by investors:
So my wife buys me a beautiful MacBook as a present, but I’m like, why did you spend all of that
money when my laptop still works well? So we decide to return it (and I don’t think I will end up
with any present now, but that’s a different story). So I walk into the mall and it’s pretty much dead
[…]. It’s Wednesday night and there’s not many people walking around. Except for the Apple store,
which is packed. It’s like a party in there. I walk in, there’s no counter so I kind of meander for a bit
then ask for help. The return was super easy with my on line receipt/bar code, and everyone
working there was super friendly. I ask the guy […] Hey, what’s that long line for? He says it’s for
those who are trying to buy an iPhone. I’m like WHAT?!?!?!?! How long has it been like this? He says
every day since the iPhone 6/6plus launch. Side note: The Microsoft store was completely empty,
except for one chubby teenager, who was playing with something […] Who knows? All the
employees were standing around and talking to each other in this pathetic looking store […]
I THINK I’LL HOLD ON TO MY SHARES A BIT LONGER!!!!! (Message posted by
apophis_2036_end on Yahoo Finance Message Board – APPL, on 5 November 2014, 8:07:00 p.m.).
If you worked at IBM anytime in the last 15 years or so you would have more than a clue as to why
one would want to leave IBM. On the inside, for those who actually do the work, it feels like a
sinking ship. Its people (“resources”) have exactly no job security at all. Makes no difference what
you work on or how good your work is […]. If your exec wants to show leadership (“suck up to his
or her boss”) you could be RAed (“randomly fired”) to save money. I have come to believe that while
it probably won’t totally go under, in 5 years it will a tenth of its size (Message posted by e_is_ir on
Yahoo Finance Message Board – IBM, on 15 September 2015, 4:17:00 p.m.).

2.3 Research hypotheses

To develop our research hypotheses, we applied personal behaviour theory (Ngai et al., 2015)
to IMBs as described in the following subsections.
H1: characteristics of the information producer. Correa et al. (2010) used personal
behaviour theory to show that user’s personality traits may be a crucial factor to leading
people to engage in social media. Previous ICD studies focussing on traditional
 communication channels reveal that the characteristics of the information producer, such as ICD and the
their cultural values, can impact financial disclosures ( Jaggi and Low, 2000). Similarly, Web 2.0
Archambault and Archambault (2003) outline that culture influences how managers
organise companies and, therefore, affects corporate disclosure. Catalfo and Wulf (2016)
reinforce the role of culture on ICD and state that for a long time there has been a lack of
managerial awareness in the field of ICD. The characteristics of the information producer,
such as nationality, corporate culture, managerial awareness, and culture are already 1891
established determinants of ICD in traditional communication channels and can, therefore,
be extended to the Web 2.0.
Focussing on Web 2.0 platforms, Sprenger et al. (2014) find that specific information
producers, market mavens, and investment gurus have a higher impact on the financial
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

information disclosure process with higher numbers of retweets. Zhang (2016), in analysing
IMBs, identifies two subgroups of information producers. The first group represents
masters with high credibility in the community; however, their expertise is neither checked
nor proven. They are deemed masters based on the number of posts they author, the
number of likes their posts receive, and, potentially, on their ability to influence others’
opinions. These elements meet the definition of online influencers (OIs) (Lahuerta-Otero and
Cordero-Gutierrez, 2016). Topics discussed by OIs may be more financial based with limited
IC insights or may be more IC based (Dumay, 2016), but they tend to contribute to
discussions by writing posts about the companies. The second group are called “trolls”.
Trolls are users who repeatedly and deliberately break the etiquette of an established online
community by posting inflammatory, extraneous, off-topic messages. Thus, two different
personal behaviours are found in IMBs: OIs support the development of relevant
discussions with the aim of contributing to the discussion on companies, while trolls move
the discussion away to topics not relevant to the IMB. From this premise, we develop the
first hypothesis based on OIs (a) and trolls (b):
H1a. Messages posted by OIs are more likely to contain ICD.
H1b. Messages posted by trolls are more likely to not contain ICD.
H2: posting time. Situational conditions, such as the time available to post on social media, can
affect the participation process for more cognitively demanding activities (Zhong et al., 2011).
Additionally, time is an important variable when analysing the information diffusion process
in social media studies since important information can be lost when user pages are
overloaded with content from multiple sources (Sabate et al., 2014). Toubia and Stephen (2013,
p. 372) state “posting content on a given day influences […] the expected number of followers
the user will have in the future”. Thus, posting on different days can have a variable impact,
and may result in an overload of information or only a few active users. Further, the timing
of an ICD impacts stock prices because analysts need time to process the information
(CRA Insights, 2014). Abeysekera (2006, p. 62) argues that “benefits from IC are not
necessarily immediately identifiable, but rather are accrued over a long-term period”, and thus
more time might be needed to accumulate and analyse posts containing IC.
Reading and posting on IMBs requires a considerable amount of time and cannot be read
in real time (Antweiler and Murray, 2004, p. 1259), yet IMBs do allow users to post whether
markets are open or closed. Over the weekend, investors are more likely to have time to
study a company’s situation and learn other investors’ opinions (Zhang, 2016). Therefore,
the timing of a post can affect the impact of an ICD. Similarly, by focussing on the specific
day or hour a message was posted, users might more easily find interesting information
on IC. Thus, the second hypothesis is:
H2. Messages posted during weekends are more likely to contain ICD.
 MD H3: buy, sell, and hold sentiments. Sentiments and emotions can affect social media
55,9 participation (Zhong et al., 2011). The sentiments expressed in internet messages are an
important part of Web 2.0 communications because they are public displays of the emotions
and opinions of the authors (Zhong et al., 2011). The emotional aspects of a message, such as
likes and emoticons, may affect whether specific messages are shared by users because
“people may share emotionally charged content to make sense of their experiences, reduce
1892 dissonance, or deepen social connections” (Berger and Milkman, 2012, p. 192). Similarly,
Stieglitz and Dang-Xuan (2013, p. 218) “hypothesize that sentiment of social media content
might be positively associated with information diffusion”. Focussing on the impact of
sentiments expressed on Twitter about stock prices, Sul et al. (2014, p. 1) state
“The emotional valence of tweets from users with many followers (more than the median)
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

had a stronger impact on same day returns, as emotion was quickly disseminated and
incorporated into stock prices”. This evidence suggests that emotional content can affect the
information diffusion process and, thus, impact stock prices.
IMBs allow users to enrich messages with personal sentiments to buy, strong buy, sell,
strong sell, or hold shares. Kim and Kim (2014, p. 708) define investor sentiment
as their “attitude or feeling toward a particular security or financial market, which
tends to be revealed through an event (such as an earnings announcement) or price
movement of the security traded in the market”. Investor sentiment has a predictive
power on stock returns, volatility, and trading volume (Kim and Kim, 2014). Interestingly,
Haji and Ghazali (2012) argue that communicating IC information has strategic
importance to shareholders and their investment decisions. From this premise, the third
hypothesis is:
H3. 1CD is more likely to be fostered by messages that contain expressions of sentiment
(such as buy, sell, or hold).
H4: message length. Cognitive effort and the perceived usefulness of the acquired
information may affect social media participation (Zhong et al., 2011). The length of a
message is an important variable that influences cognitive effort on one side, and its
perceived usefulness on the other, because it represents how the richness of messages in
social media affects information diffusion (Fortin and Dholakia, 2005). The richness of a
social media message can be measured by the presence of videos, images, hyperlinks, or
plain text (Sabate et al., 2014, p. 1003). Focussing on the role of plain text, De Vries et al.
(2012, p. 86) find “message length may affect outcome measures such as click-through
rates”. Therefore, the richness of messages, as measured by their length, can affect
information disclosure. Typically, IMBs allow users to post short messages with few words,
or longer messages containing external links, but they rarely allow users to include pictures
or videos. According to Zhang (2016, p. 114), “message length is a proxy for the amount of
information in messages”. Therefore, longer messages could provide more detail about the
company’s IC, and the following hypothesis can be derived:
H4. ICD is more likely to be fostered by longer messages.

3. Data collection and data analysis

The ten most capitalised companies listed on the New York Stock Exchange were
analysed. The selected companies were: Apple, Microsoft, Walmart, IBM, JP Morgan,
Oracle, Alphabet, Pfizer, Citigroup, and BP. The messages posted about each company in
a one-year period, from 1 October 2014 to 30 September 2015, on Yahoo!Finance and
TheLion.com were downloaded, resulting in 60,996 messages. These two IMBs were
selected because they are considered to be the most important (Das and Chen, 2007;
Sabherwal et al., 2011). The downloaded messages were first imported into NVivo for
 content analysis (Krippendorff, 2013) and were then statistically analysed to test the ICD and the
hypotheses following a three-step procedure. Web 2.0
First, we develop a framework that represents the analytical construct (Krippendorff,
2013) used to analyse the downloaded messages. To develop the framework, we performed a
keyword search using the terms “IC” and “disclos*” or “reporting” on Scopus. Scopus was
selected because it is one of the largest abstract and citation databases of peer-reviewed
literature (Massaro et al., 2016). A total of 57 papers were identified, but only 37 studies 1893
displayed the framework used. These studies were analysed to identify the IC items used in
each study. Table I lists the main terms and their frequency among the studies analysed and
represents the framework used in this study.
Second, due to the number of messages to code, a group of external researchers was
trained to manually code all the messages (Krippendorff, 2013) according to the variables
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

listed in Table I. Additionally, we asked them to code non-company posts. Each message

Human capital Relational capital Structural capital

No. of No. of No. of
Variables studies Variables studies Variables studies

Education 24 Brand 20 Patents and intellectual property 21

Know-how 15 Customers 27 Corporate culture 20
Training 14 Business collaborations 17 Information systems 20
Work-related knowledge 14 Distribution channels 17 Management processes 18
Skills 13 Company name/reputation 13 Networking systems 16
Work-related competencies 13 Customer loyalty 13 Financial relations 14
Experiences 12 Licenses agreement 13 Research and development 14
Entrepreneurial spirit 11 suppliers 13 Management philosophy 13
Vocational qualification 9 Franchising agreements 11 Trademarks 13
Employee 8 Customer satisfaction 7 Copyrights 12
Knowledge 8 Favourable contracts 7 Operating system 7
Innovativeness 7 Partnership 7 Organisational culture 7
Employee benefits/compensation 5 Customer relation 6 Infrastructure assets 6
Employee retention 5 Investors 5 Procedures 5
Abilities 4 Reputation 5 Routines 5
Changeability 4 Market share 4 Innovation 4
Competences 4 Stakeholder relationships 3 Knowledge sharing 4
Employee satisfaction 4 Trademarks 3 Concepts 3
Proactive and reactive abilities 4 Customer retention 2 Databases 3
Cultural diversity 3 Joint venture 2 New products 3
Employee career development 3 Research collaborations 2 Corporate learning 2
Employee motivation 3 Customer penetration 1 Corporate university 2
Employee safety 3 Goodwill 1 Design 2
Values 3 Government 1 Leadership 2
Employee productivity 2 Public relations 1 Organisational learning 2
Empowerment 2 Quality standards 1 Alliances 1
Management quality 2 Archives 1
Qualifications 2 Communication 1
Career development 1 Computer and administrative 1
systems
Confidential information 1 Co-operation 1
Employment policy, security, 1 Governance principles 1
and opportunities
Executive compensation plan 1 Internet domain name 1
Genetic inheritance 1 Job descriptions and manuals 1
Mentoring 1 Knowledge transfer 1
Productivity 1 Management quality 1
Social competency 1 Market value 1
Staff capacity 1 Product description 1
Union activity 1 Social responsibility 1 Table I.
Work-life balance 1 Software 1 ICD frameworks used
Subsidiaries 1 in previous studies
 MD was used as a unit of analysis (Krippendorff, 2013). To ensure reliability, one author
55,9 independently performed NVivo-aided coding based on a word-search of the terms in Table I,
which included synonyms (e.g. talk and speak), generalisations (e.g. whisper), and
specialisations (e.g. communicate). The results of the manual coding and the NVivo coding
were compared to resolve any discrepancies and calculate Krippendorff’s α (Krippendorff, 2013).
The Krippendorff’s α obtained is 0.930, demonstrating the reliability of the content analysis.
1894 Third, descriptive statistics and a logistic regression using the software R were used to
test the developed hypotheses and provide insights into the data.

4. Variables used
One dependent variable and four independent variables were used to test our hypotheses.
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

The following subsections describe the dependent and independent variables and how they
were calculated.

4.1 Dependent variable

An ICD index was used as the dependent variable. An ICD index is “a semi-objective
approach as the researchers have specified ex-ante a list of items and scrutinises the text for
their presence” (Husin et al., 2012, p. 198). Previous ICD studies have used a progressive
measure based on the number of codes found in the analyses (e.g. Joshi et al., 2012).
However, this study used one message as the unit of analysis; therefore, a dummy variable
was used for each message. If one or more of the IC elements was present in the discussion,
the message was coded as 1; otherwise, the message was coded as 0. The underlying proxy
is that: the more messages that were coded as 1, the more an IMB discloses IC.

4.2 Independent variables

H1: characteristics of the information producer. Methods for analysing OIs in social media
have been widely discussed in the literature, and several measures have been proposed to
identify them: views per month, links, the frequency of posts, media citations, core, and
engagement indexes, just to cite a few metrics (Booth and Matic, 2011). Most of these
approaches are not applicable to IMBs since they do not allow users to have followers and
do not provide statistics such as the views per month. Therefore, starting from approaches
used in previous studies (Booth and Matic, 2011), we focussed on some statistics provided
by IMBS.
To measure and detect OIs and trolls, we counted the number of messages posted by
each author and the number of likes each author had received. Focussing on stock micro-
blogging, Sprenger et al. (2014) find that almost 8 per cent of users post more than ten tweets
in a six-month period. Additionally, in analysing the credit scores of authors on the message
board WallStreetPit, Zhang (2016) finds that 7 per cent of users have a credit score of over
3,000. Using these results, we partitioned the number of posts and likes, defining OIs as
users with the top 10 per cent of total messages and total likes. Similarly, trolls were defined
as users with more than 90 per cent of their posts off-topic. Messages posted by OIs and
trolls were identified using a dummy variable.
H2: posting time. To measure the days of the week, we downloaded the date of the posts
and calculated the day of the week from Monday to Sunday. Weekends were measured
using a dummy variable – 0 for messages posted on a weekday, and 1 for messages posted
on the weekend.
H3: buy, sell, and hold sentiments. Sentiment was measured by analysing the specific
sentiment expressed by the author (i.e. sell, buy, and hold). While Yahoo!Finance provides a
rigid choice between strong sell, sell, hold, buy, and strong buy, TheLion.com provides an open
field to be completed. Therefore, sentiments in TheLion.com messages were manually coded.
 For example, hold and keep were both coded as hold, and messages with missing data were ICD and the
coded as “No sentiment provided”. The coding for each message was discussed among the Web 2.0
authors to ensure the validity of the results.
H4: message length. Message length was calculated as the number of characters in each
message. Yahoo!Finance allows the user to include a title with each message, TheLion.com
does not. In this study, message length only considered the number of characters in the body
of the message and not the title. 1895
5. Results
The main results of the analysis are presented first, followed by the results of the hypothesis
tests. Table II presents the descriptive statistics of the analysis and Table III provides the
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

correlation matrix.
At a glance, the results shown in Table III show that few messages focus on IC – only
4 per cent of the total messages posted. Therefore, to further answer the research hypotheses a
logistic regression analysis was used to test the ICD determinants, applying the following model:
Logit ðIC disclosureindexÞ ¼ a þb1 OIsþb2 trollsþb3 posting time
þb4 sentimentþb5 textlength
To develop the analysis, we first checked each of the relevant predictors for the dependent
variable to determine whether they should be included in the analysis (Gelman and Hill, 2007).

(A) (B) (B/A)

Company Code Market Capitalisation No. of messages IC % of total messages

Apple Inc. AAPL NASDAQ 637.91 bn 9,716 737 8

Microsoft Corp. MSFT NASDAQ 375.99 bn 7,999 377 5
Walmart WMT NYSE 190.21 bn 8,008 567 7
IBM IBM NYSE 147.02 bn 4,605 415 9
JP Morgan JPM NYSE 228.84 bn 4,250 39 1
Oracle Corp ORCL NASDAQ 160.14 bn 7,981 9 0
Google Inc./Alphabet GOOG NASDAQ 455.07 bn 4,717 135 3
Pfizer Inc. PFE NYSE 210.18 bn 1,971 48 2
Citigroup Inc. C NYSE 159.43 bn 5,715 77 1
BP PLC BP NYSE 108.94 bn 6,034 177 3 Table II.
Total 60,996 2,581 4 Descriptive statistics
Krippendorff’s α 0.930 for ICDs on IMBs

Variable ICD OIs Trolls Weekend Sentiment Text length

ICD 1.00
OIs 0.10*** 1.00
Trolls −0.07*** −0.07*** 1.00
Weekend 0.02*** 0.03*** 0.10*** 1.00
Sentiment 0.04*** 0.01 −0.18*** −0.02*** 1.00
Text length 0.17*** 0.05*** 0.01*** 0.05*** −0.06*** 1.00
Min 0 0 0 0 0 1
Max 1 1 1 1 1 13.280
Mean 0.04 0.01 0.22 0.15 0.15 257
SD 0.20 0.12 0.41 0.32 0.36 392
Number of messages 60,996 60,996 60,996 60,996 60,996 60,996 Table III.
Notes: Significance codes: * o 0.05; ** o0.01; ***o0.001; ****o0.1 Correlation matrix
 MD To perform this test, we developed a maximum likelihood ANOVA table to compare
55,9 alternative models, adding one variable at a time. The results confirmed that all the
variables of the model were relevant. Additionally, we checked multicollinearity using
a variance inflation factor index, which resulted in the highest possible value of 1.01.
Following this, we performed a logistic regression. The results are shown in Table IV.

1896 5.1 H1: characteristics of the information producer

The results in Table IV show that when a message is posted by an OI the chance that the
message contributes to ICD increases to 341 per cent (p o0.001). We, therefore, reject
the null hypothesis and confirm H1a. OIs are well-known posters due to their number of
posts and can capture high levels of likes. The credibility that OIs attract derives from their
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

well-focussed messages. Traditionally, research has focussed on sell-side analysts, fund

managers, credit raters, auditors, investment bankers, and information specialists such as
Bloomberg and Reuters (Holland, 2003). However, the existence of OIs introduces new actors
to this process, and the role of OIs in attracting likes highlights the need for a better
understanding of how investors react as an audience in terms of idea consumption
orientations (Groß et al., 2015).
Table IV also shows that when a troll posts a message, the chance that the message
contributes to ICD decreases to 63 per cent (p o0.001), confirming H1b. Surprisingly,
the results show that 22 per cent of the messages were posted by trolls. Delving deeper,
we analysed the distribution of trolls among the companies. Interestingly, most of the trolls
were active on just one IMB. On Oracle’s message board, 77 per cent of the messages were
posted by four authors, with one accounting for more than 4,020 messages. Table V lists
results of this analysis.
Prominent authors, such as delacruze171, do not relate to the market but provide politically
connected messages. For example, the following is an example of a message posted by
delacruze171 on 8 March 2015, 4:52:00 p.m.:
Obama is the best man alive! He is AMAZING beyond belief […] PROOF is, look at all the
garbage spewed about him by ugly repubs/If Obama wasn’t soooooooooo fantastic they wouldn’t
spew as much.
These results show that some discussions are authored by trolls (Zhang, 2016), who
repeatedly and deliberately break the etiquette of an established online community by
posting inflammatory, extraneous, off-topic messages. Interestingly, message boards that
have active trolls show a lower number of other authors. For example, on Oracle’s message
boards, even though there are 7,981 messages, there are only 311 active users, while at IBM
there are 1,051 active users with 4,605 messages. These results support Zhang (2016) who
argues that online discussions are not purely useless noise. However, when the noise is too
high due to “trolling”, investors are discouraged and leave the message board.

Variable β SE Z-value Pr Exp(β)

Intercept −3.516 0.031 −112.761 o0.001*** 0.029

OIs 1.485 0.086 17.236 o0.001*** 4.416
Trolls −0.972 0.071 −13.760 o0.001*** 0.378
Table IV. Posting time 0.181 0.054 3.350 o0.001*** 1.199
Logistic regression: Sentiment 0.497 0.050 9.786 o0.001*** 1.644
dependent variable Text length 0.001 0.001 31.999 o0.001*** 1.001
ICD index Notes: Nagelkerke’s R2 ¼ 0.08. Significance codes: * o0.05; **o 0.01; *** o0.001; ****o 0.1
 Authors APPL MSFT WMT IBM JPM ORCL GOOG PFE C BP Total
ICD and the
Web 2.0
cloudyplayer438 795 795
dbs_l952 638 638
delacruze171 4,020 4,020
edbanks999 1,675 2 1,677
fernando_rua 478 478
hawcreek_debunker 516 516 1897
hawcreekl2 856 856
messageboardpreacher 655 655
phil_mcrakkin 668 668
predisposable_nym_02 590 590
seeburgc 668 668
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

shep2l998 413 413

tigersnumberone962 1,013 1,013
wottowwottow 438 438
y0000000000h0000000000 483 483
Total 483 2,400 2,732 2,153 6,138 2 13,908 Table V.
Total messages 9,716 7,999 8,008 4,605 4,250 7,981 4,717 1,971 5,715 6,034 60,996 Trolls and most
% 5 30 34 0 51 77 0 0 0 0 23 active users

5.2 H2: posting time

Table IV shows that, when a message is posted on the weekend, the chance of a message
contributing to ICD is higher by 20 per cent ( p o0.001). Therefore, the results confirm H2.
Investors can suffer from sensitivity to the complexity of ICD (Farooq and Nielsen, 2014).
Weekends provide more time for investors to analyse specific situations and focus on the
relevant but more complex elements of a message, such as IC. Hence, these results call for a
more complex approach to ICD. The benefits of IC are not necessarily immediately
identifiable, they are complex and require time to be understood.

5.3 H3: buy, sell, and hold sentiments

Table IV shows that when a message contains investor sentiment, the odds that the
message contributes to ICD is higher by 64 per cent when compared to messages with no
sentiment. Statistical tests show that this result is significant ( p o0.001) and confirms H3.
These results build on Nimtrakoon’s (2015) findings and confirm the importance of IC for
investors in making decisions about the stock market. Additionally, deepening the analysis
by splitting the sentiments expressed, our findings show that the sentiments “strong sell”
and “strong buy” combined account for more than 82 per cent of the total messages with a
sentiment. These results confirm Kim and Kim’s (2014) findings that investors tend to reveal
extreme sentiments (i.e. strong buy and strong sell) rather than moderate sentiments
(i.e. buy and sell) and confirm the importance of IC to investors’ decision-making processes.

5.4 H4: message length

Table IV shows that message length has a statistically significant effect ( p o0.001) on the
odds that the message contributes to the ICD and therefore confirms H4. Campbell and
Rahman (2010, p. 66) state that IC is a more complex topic to describe than a physical asset,
which can be simply be explained using terms such as the age of the asset or its
amortisation percentage. Consequently, posters need more words to describe IC. Analysing
the text length, the results in Table IV show that messages have an average length of 257.3
characters with a minimum of one, a maximum of 13,280, and a standard deviation of
392.63. This shows that investors can be very engaged in the discussion, and very concise in
their comments, but longer messages are more likely to disclose information about IC.
 MD 6. Discussion and implications
55,9 This section presents the four main implications that can be drawn from our results.

6.1 Implication no. 1 – OI and trolls: validity, trust, and discussion moderation
The first implication emanates from our initial search and analysis of the posts because we
discovered a significant amount of noise in this communication channel. Contemporary
1898 communication channels are more complex and unstructured, and the internet and social
media are redefining the way we now gather and disseminate information (Zhang, 2016).
However, from our results, we conclude that the messages that relate directly to companies do
contain IC information of interest to investors and, therefore, IMBs offer a valid source of ICD.
The disclosures that emanate from bonafide investor participants in the discussions are
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

interesting and have the potential to be used in investors’ decision-making processes. But, our
results also show that not all messages are valid, and not all can be trusted due to noise.
The validity of many disclosures is questionable because investors can never be sure of
the source of the disclosure, or the intention of the person participating in the discussion. For
example, many posts are authored by trolls with political agendas and do not relate to the
company. Similarly, the data sources used by participants to develop their posts are not
verifiable and can include rumours or innuendo. Moreover, it is impossible to tell whether an
OI is an independent author or an employee of a particular company. Thus, trust in the
disclosure is questionable.
While researchers are able to filter out invalid or untrustworthy posts, it is arguably
beyond the ability of many smaller investors to do this effectively. However, the evidence
shows that when there are too many posts from trolls spruiking political or other
agendas, this noise is enough to turn participants away from engaging in the discussion.
These results build on personal behaviour theory and show that perceived utility
(Chao-Min et al., 2011) is an important element for fostering user participation in the ICD
process. Therefore, to improve the quality of the information, there is a need to reduce
noise, especially from posts unrelated to the company and investment decision making.
We recommend that IMB administrators include a system to moderate posts against trolls
who continually post off-topic messages.

6.2 Implication no. 2 – timing of disclosures

The second implication emanates from the number of messages analysed and the effect of
the posting time. According to Haji and Ghazali (2012, p. 377), “corporate communication
has undergone a notable number of paradigm shifts, from a mere reporting of financial
statements to the inclusion of a more strategic and forward-looking information such as
intellectual capital”. Additionally, communication is evolving due to the role of social media.
As Scarfi (2012, p. 2) states, “we have millions of tweets about our brands, thousands of
Facebook likes, hundreds of thousands of check-ins on Foursquare. Pinterest and Instagram
are adding even more to [the] social-media data deluge”. IMBs are part of this evolution, with
thousands of messages posted every day, and are providing new ways for investors and
companies to communicate about IC.
It is difficult for companies and investors to cope with the large amounts of data
available, especially considering that using a personal behaviour theory we can claim that
ICD requires more cognitively demanding activities (Zhong et al., 2011). Thus, our results
are useful because when investors are searching for information about a company or
looking for other opinions and independent evaluations, it is evident that investors can
better focus their attention on specific days of the week. We recommend that long-term
investors looking for more in-depth evaluations of IC should focus their efforts on
discussions developed when the markets are closed, especially during the weekends, when
messages are longer, richer, and more likely to focus on IC.
 6.3 Implication no. 3 – using sentiment to assess the value of the disclosure ICD and the
The third implication emanates from the use of sentiment. Our results confirm that Web 2.0
messages expressing sentiment talk more about IC, although the statistical effect is lower
than the other variables. Considering the high level of noise, sentiment is useful for filtering
search results to solely company-related messages. Neither Yahoo!Finance nor TheLion.com
allows users to filter data by sentiment. We believe that implementing a sentiment filter
could help investors and managers to focus on relevant messages. Additionally, both 1899
investors and managers who are interested in understanding the perception of investors
should focus on the messages that express sentiments to both reduce the number of
messages that need to be read and facilitate user participation.
Further, the concept of sentiment in social media is evolving. Facebook has recently
modified its “Like” button and introduced new emoji reactions to give “users more authentic
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

ways to quickly and easily respond to posts, whether they are sad, serious, funny or happy”
(Chaykowski, 2016, p. 2). According to Chaykowski (2016, p. 3), “users who have Reactions
have already been responding more frequently to posts than users without them”.
Considering that sentiment expressions help the development of discussions in social media,
IMBs like Yahoo!Finance and TheLion.com should give thought to implementing features
similar to Facebook, as they could be beneficial for both managers and investors.

6.4 Implication no. 4 – volume and length of IMB disclosures

The fourth implication relates to message length and the volume of ICDs. Our results
confirm that IC is a complex topic and requires longer messages to explain. These
results confirm Campbell and Rahman’s (2010, p. 56) findings, who conducted a
longitudinal analysis of IC reporting. They state, “narrative (as opposed to quantitative)
reporting has increased and “factual” (as opposed to opinion and judgement) reporting has
decreased”. Narratives help to explain IC and, due to its complex nature, requires longer
disclosures, which reflect the characteristics of a highly demanding cognitive process
(Zhong et al., 2011).
Additionally, social media disclosures, other than narratives, allow a richer way of
communicating. Most social media platforms provide users with a way of posting content
other than text, such as tables, charts, movies, pictures, hyperlinks, and emoticons.
Studies on social media show richer messages attract more attention (Sabate et al., 2014) and
open new ways of communicating. Interestingly, Yahoo!Finance and TheLion.com offer
limited options to enrich messages. We argue that introducing more functions to enrich the
message formats on IMBs, by allowing charts and tables or even allowing users to highlight
concepts in bold or italic, for example, would improve communication.

7. Conclusion
This section presents our final remarks, opportunities for further research, and the
limitations of this study.

7.1 Final remarks

This paper contributes to the field of IC by analysing ICD in the Web 2.0 era. Unlike
traditional ICD research, which examines company-created and voluntary ICDs, we analyse
how investors contribute to ICD through IMBs. Analysing messages by market participant
is important because ICD research has always espoused the benefits of ICD from a company
perspective. However, in this study, we analyse how investors contribute to ICD through
involuntary disclosures and propose a new theoretical approach to ICD research using
personal behaviour theory (Ngai et al., 2015).
 MD Social media presents a new and different context where information is not produced and
55,9 controlled by the company, but rather by many users who contribute to the creation and
dissemination of information using Web 2.0 applications. Within this context, the theories
normally used to examine ICD, such as agency, stakeholder, signalling, and legitimacy
theory, do not apply. We propose that personal behaviour theory (Ngai et al., 2015), as a
group of theories, better addresses the matter of understanding what fosters user
1900 participation in IMBs.
From an analysis of the characteristics of the information producers, our results show
very little discussion about IC on IMBs. However, further analysis reveals a more complex
picture. The discussions on IMBs cover many topics, but trolls can misdirect these
conversations and estrange participants. When posts by trolls are excluded, IC assumes a
more important role. These results build on ICD research that shows perceived utility is an
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

important element in fostering involuntary disclosures (Chao-Min et al., 2011). When too
many messages are posted by trolls, perceived utility is reduced, and investors leave the
discussion. In the Web 2.0 era, there are many ways and many users that can contribute to
the information diffusion process – positively and negatively. Perhaps, more importantly,
there are many ways other than simply posting “fake news” to mislead markets. Therefore,
managers, investors, and authorities with the aim of assuring market functionality should
consider the implications of these results.
Second, several situational characteristics interact with personal behaviour, such as the
time of posting. ICD is a complex topic that requires time to be understood and discussed.
These results build on Zhong et al. (2011) and show how cognitively demanding activities,
such as IC discussions, influence investor participation in ICD. The time of posting may be
an interesting variable for filtering thousands of messages when searching for posts that
contain useful information.
Third, the sentiments expressed by investors (such as hold, buy, or sell) affect ICD
since investors need to justify their opinion. Considering that sentiment might be positively
associated with information diffusion (Stieglitz and Dang-Xuan, 2013), ICD might
be influenced by the intention of investors to disclose their investment opinion on a
specific asset.
Fourth, message length is important because it represents how the richness of messages
in social media affects information diffusion (Fortin and Dholakia, 2005). The richness of a
social media message can be measured by the presence of videos, images, hyperlinks, or
plain text (Sabate et al., 2014). Considering that information usefulness affects investor
participation in social media (Zhong et al., 2011), our results show that involuntary
disclosures can be affected by the richness of the message.
Overall, we believe that adding some of the variables derived from personal behaviour
theory to the analysis of IMB content (Ngai et al., 2015), such as user characteristics, the
time of posting, the richness of the messages, and the sentiments they express could
help explain the attitudes towards involuntary disclosures by investors and provide new
insights into ICD.

7.2 Opportunities for further research and the limitations of this study
This study opens up new research opportunities. First, this study only focussed on variables
thought to affect ICD. New research lines could focus on the impact that IMBs have on the
market, such as the relationship between ICD and market volatility or stock prices. Second,
IMBs are only one of many new Web 2.0 tools. Further research could enlarge this analysis
to other social media platforms used to disclose and spread company-related information,
such as Twitter, Facebook, and LinkedIn. Specific research could narrow its focus to
particular aspects of ICD communication. For example, LinkedIn appears to have an
important role in attracting and fostering human capital. Other social media, such as
 Facebook, might be used more to promote company brands and, therefore, a company’s ICD and the
relational capital. Third, this analysis focussed on ICD in listed companies, but social media Web 2.0
is also used by non-listed companies, and further studies could extend the scope of their
analyses to include a wider range of company structures.
As with any research, this study has some limitations. One important limitation is the
subjectivity of applying manual content analysis and the focus on only ten of the most
capitalised companies. Our results may, therefore, be biased by the limited number of 1901
companies analysed, the level of assets held by those companies, and/or their specific
business sectors.

Note
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

1. According to Kaplan and Haenlein (2010, pp. 60-61), “Web 2.0 is a term that was first used in 2004
to describe a new way in which software developers and end-users started to utilize the World
Wide Web; that is, as a platform whereby content and applications are no longer created and
published by individuals, but instead are continuously modified by all users in a participatory and
collaborative fashion”.

References
Abeysekera, I. (2006), “The project of intellectual capital disclosure: researching the research”, Journal
of Intellectual Capital, Vol. 7 No. 1, pp. 61-75.
Albertini, E. (2016), “An inductive typology of the interrelations between different components of
intellectual capital”, Management Decision, Vol. 54 No. 4, pp. 887-901.
Alexander, R.M. and Gentry, J.K. (2014), “Using social media to report financial results”, Business
Horizons, Vol. 57 No. 2, pp. 161-167.
Antweiler, W. and Murray, F.Z. (2004), “Is all that talk just noise? The information content of internet
stock message boards”, Journal of Finance, Vol. 59 No. 3, pp. 1259-1294.
Archambault, J.J. and Archambault, M.E. (2003), “A multinational test of determinants of corporate
disclosure”, International Journal of Accounting, Vol. 38 No. 2, pp. 173-194.
Berger, J. and Milkman, K.L. (2012), “What makes online content viral?”, Journal of Marketing Research,
Vol. 49 No. 2, pp. 192-205.
Booth, N. and Matic, J.A. (2011), “Mapping and leveraging influencers in social media to shape
corporate brand perceptions”, Corporate Communications, Vol. 16 No. 3, pp. 184-191.
Campbell, D. and Rahman, A.M.R. (2010), “A longitudinal examination of intellectual capital reporting
in Marks & Spencer annual reports, 1978-2008”, British Accounting Review, Vol. 42 No. 1,
pp. 56-70.
Catalfo, P. and Wulf, I. (2016), “Intangibles disclosure in management commentary regulation in
Germany and Italy: a semantic approach”, Journal of Intellectual Capital, Vol. 17 No. 1,
pp. 103-119.
Chao-Min, C., Wang, E.T.G., Shih, F.-J. and Yi-Wen, F. (2011), “Understanding knowledge sharing in
virtual communities”, Online Information Review, Vol. 35 No. 1, pp. 134-153.
Chaykowski, K. (2016), “Facebook no longer just has a ‘like’ button, thanks to global launch of emoji
‘reactions’ ”, Forbes, 24 February, pp. 1-8.
Correa, T., Hinsley, A.W. and de Zúñiga, H.G. (2010), “Who interacts on the Web? The intersection of
users’ personality and social media use”, Computers in Human Behavior, Vol. 26 No. 2,
pp. 247-253.
CRA Insights (2014), “Price impact of disclosures before, during, or after a trading day: implications for
event studies”, August, pp. 1-3, available at: www.crai.com/sites/default/files/publications/FM-
Insights-Intraday-Trading-August-2014.pdf (accessed 14 September 2017).
 MD Das, S.R. and Chen, M.Y. (2007), “Yahoo! for Amazon: sentiment extraction from small talk on the web”,
55,9 Management Science, Vol. 53 No. 9, pp. 1375-1388.
Deegan, C. and Samkin, G. (2009), New Zealand Financial Accounting, McGraw-Hill, Sydney.
De Vries, L., Gensler, S. and Leeflang, P.S.H. (2012), “Popularity of brand posts on brand fan pages: an
investigation of the effects of social media marketing”, Journal of Interactive Marketing, Vol. 26
No. 2, pp. 83-91.
1902 Dumay, J. (2016), “A critical reflection on the future of intellectual capital: from reporting to disclosure”,
Journal of Intellectual Capital, Vol. 17 No. 1, pp. 168-184.
Dumay, J. and Cai, L. (2014), “A review and critique of content analysis as a methodology for inquiring
into IC disclosure”, Journal of Intellectual Capital, Vol. 15 No. 2, pp. 264-290.
Dumay, J. and Guthrie, J. (2017), “Involuntary disclosure of intellectual capital: is it relevant?”, Journal
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

of Intellectual Capital, Vol. 18 No. 1, pp. 29-44.

Farooq, O. and Nielsen, C. (2014), “Improving the information environment for analysts: which
intellectual capital disclosures matter the most?”, Journal of Intellectual Capital, Vol. 15 No. 1,
pp. 142-156.
Fortin, D.R. and Dholakia, R.R. (2005), “Interactivity and vividness effects on social presence and
involvement with a web-based advertisement”, Journal of Business Research, Vol. 58 No. 3,
pp. 387-396.
Freberg, K., Graham, K., McGaughey, K. and Freberg, L.A. (2011), “Who are the social media
influencers? A study of public perceptions of personality”, Public Relations Review, Vol. 37 No. 1,
pp. 90-92.
Gelman, A. and Hill, J. (2007), Data Analysis Using Regression and Multilevel/Hierarchical Models,
Cambridge University Press, Cambridge.
Groß, C., Heusinkveld, S. and Clark, T. (2015), “The active audience? Gurus, management ideas and
consumer variability”, British Journal of Management, Vol. 26 No. 2, pp. 273-291.
Guthrie, J. and Petty, R. (2000), “Intellectual capital: Australian annual reporting practices”, Journal of
Intellectual Capital, Vol. 1 No. 3, pp. 241-254.
Guthrie, J., Ricceri, F. and Dumay, J. (2012), “Reflections and projections: a decade of intellectual capital
accounting research”, British Accounting Review, Vol. 44 No. 2, pp. 68-82.
Haji, A.A. and Ghazali, M.N.A. (2012), “Intellectual capital disclosure trends: some Malaysian evidence”,
Journal of Intellectual Capital, Vol. 13 No. 3, pp. 377-397.
Holland, J. (2003), “Intellectual capital and the capital market: organisation and competence”,
Accounting, Auditing & Accountability Journal, Vol. 16 No. 1, pp. 39-48.
Hsu, C.L. and Lin, J.C.C. (2008), “Acceptance of blog usage: the roles of technology acceptance, social
influence and knowledge sharing motivation”, Information and Management, Vol. 45 No. 1,
pp. 65-74.
Husin, N.M., Hooper, K. and Olesen, K. (2012), “Analysis of intellectual capital disclosure – an
illustrative example”, Journal of Intellectual Capital, Vol. 13 No. 2, pp. 196-220.
Jaggi, B. and Low, P.Y. (2000), “Impact of culture, market forces, and legal system on financial
disclosures”, The International Journal of Accounting, Vol. 35 No. 4, pp. 495-519.
Joshi, M., Ubha, D.S. and Sidhu, J. (2012), “Intellectual capital disclosures by Indian and Australian
information technology companies: a comparative analysis”, Journal of Intellectual Capital,
Vol. 13 No. 4, pp. 582-598.
Kaplan, A.M. and Haenlein, M. (2010), “Users of the world, unite! The challenges and opportunities of
social media”, Business Horizons, Vol. 53 No. 1, pp. 59-68.
Kim, S.-H. and Kim, D. (2014), “Investor sentiment from internet message postings and
the predictability of stock returns”, Journal of Economic Behavior & Organization, Vol. 107,
Part B, pp. 708-729.
 Krippendorff, K. (2013), Content Analysis: An Introduction to its Methodology, Sage Publications, ICD and the
Thousand Oaks, CA. Web 2.0
Kristandl, G. and Bontis, N. (2007), “Constructing a definition for intangibles using the resource based
view of the firm”, Management Decision, Vol. 45 No. 9, pp. 1510-1524.
Lahuerta-Otero, E. and Cordero-Gutierrez, R. (2016), “Looking for the perfect tweet: the use of data mining
techniques to find influencers on twitter”, Computers in Human Behavior, Vol. 64, pp. 575-583,
available at: https://doi.org/10.1016/j.chb.2016.07.035 1903
Massaro, M., Dumay, J.C. and Guthrie, J. (2016), “On the shoulders of giants: undertaking a structured
literature review in accounting”, Accounting, Auditing and Accountability Journal, Vol. 29 No. 5,
pp. 767-901.
Ngai, E.W.T., Tao, S.S.C. and Moon, K.K.L. (2015), “Social media research: theories, constructs, and
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

conceptual frameworks”, International Journal of Information Management, Vol. 35 No. 1, pp. 33-44.
Nimtrakoon, S. (2015), “The relationship between intellectual capital, firms’ market value and financial
performance”, Journal of Intellectual Capital, Vol. 16 No. 3, pp. 587-618.
Sabate, F., Berbegal-Mirabent, J., Canabate, A. and Lebherz, P.R. (2014), “Factors influencing popularity
of branded content in Facebook fan pages”, European Management Journal, Vol. 32 No. 6,
pp. 1001-1011.
Sabherwal, S., Sarkar, S.K. and Zhang, Y. (2011), “Do internet stock message boards influence trading?
Evidence from heavily discussed stocks with no fundamental news”, Journal of Business
Finance and Accounting, Vol. 38 Nos 9/10, pp. 1209-1237.
Scarfi, M. (2012), “Social media and the Big Data explosion”, Forbes, 28 June, pp. 1-8.
Securities Exchange Commission (2009), “SEC charges New York broker for manipulating stock prices
through fake press releases and internet postings”, Press Release, New York, NY.
Singh, I. and Van der Zahn, J. (2008), “Determinants of intellectual capital disclosure in prospectuses of
initial public offerings”, Accounting and Business Research, Vol. 38 No. 5, pp. 409-431.
Sprenger, T.O., Tumasjan, A., Sandner, P.G. and Welpe, I.M. (2014), “Tweets and trades: the
information content of stock microblogs”, European Financial Management, Vol. 20 No. 5,
pp. 926-957.
Stieglitz, S. and Dang-Xuan, L. (2013), “Emotions and information diffusion in social media – sentiment
of microblogs and sharing behavior”, Journal of Management Information Systems, Vol. 29
No. 4, pp. 217-248.
Sul, H.K., Dennis, A.R. and Yuan, L.I. (2014), “Trading on Twitter: the financial information content of
emotion in social media”, Hawaii International Conference on System Sciences, January, pp. 1-11.
Toubia, O. and Stephen, A.T. (2013), “Intrinsic vs image-related utility in social media: why do people
contribute content to twitter?”, Marketing Science, Vol. 32 No. 3, pp. 368-392.
Yi, A., Davey, H. and Eggleton, I.R.C. (2011), “Towards a comprehensive theoretical framework for
voluntary IC disclosure”, Journal of Intellectual Capital, Vol. 12 No. 2, pp. 571-585.
Yu, Y., Duan, W. and Cao, Q. (2013), “The impact of social and conventional media on firm equity value:
a sentiment analysis approach”, Decision Support Systems, Vol. 55 No. 4, pp. 919-926.
Zhang, Y. (2016), Stock Message Boards: A Quantitative Approach to Measuring Investor Sentiment,
Palgrave Macmillan, New York, NY.
Zhong, B., Hardin, M. and Sun, T. (2011), “Less effortful thinking leads to more social networking? The
associations between the use of social network sites and personality traits”, Computers in
Human Behavior, Vol. 27 No. 3, pp. 1265-1271.

About the authors

Maurizio Massaro, PhD, has been an Assistant Professor at Udine University since 2008. Before joining
academia, he was the Founder and the CEO of multiple consultancy firms. He has also served as a
Research Centre Vice President in the field of metal analysis. He has been a Visiting Professor at
Florida Gulf Coast University and Leicester University. He enjoys several contacts and research
 MD partnerships with universities in the USA, continental Europe, and the UK. His research interests
55,9 include knowledge management, intellectual capital, and sustainability. Maurizio Massaro is the
corresponding author and can be contacted at: maurizio.massaro@uniud.it
John Dumay is an Associate Professor in Accounting at Macquarie University, Sydney. He worked
for over 15 years as an Independent Business Consultant across a wide variety of industries before
joining academia after completing PhD in 2008. The PhD entitled “Intellectual capital in action:
Australian studies” won the prestigious Emerald/EFMD Outstanding Doctoral Research Award for
1904 2008 for the Knowledge Management category. He continues to research on the topic of intellectual
capital, sustainability reporting, innovation, research methods, and academic writing. His research
activities link closely to management accounting and scholarly practice.
Carlo Bagnoli is an Associate Professor of Business Policy and Strategy at the Department of
Management, Ca’ Foscari University of Venice. He received a PhD in Business Economics at the Ca’
Foscari University of Venice. He was a Visiting Research Fellow at the University of Florida. He is the
Downloaded by UNIVERSITAS TRISAKTI, User Usakti At 02:15 27 March 2018 (PT)

Scientific Director of the Innovarea Project funded by the Regional Italian Government. His research
interests include knowledge management, competitive strategy, and business model innovation.
Carlo’s research work has been published in various outlets, including the Journal of Business
Economics and Management, Industrial Management & Data System, Journal of Management and
Governance, and Journal of Intellectual Capital.

For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: permissions@emeraldinsight.com