The effect of the activities of banks on the online sentiment and the share price : a case study of the Triodos bank

(1)

University of Amsterdam

Faculty of Economics and Business

‘THE EFFECT OF THE ACTIVITIES OF BANKS

ON THE ONLINE SENTIMENT AND THE SHARE

PRICE: A CASE STUDY OF THE

TRIODOS

BANK’

Nicole Woldring

10453997

Supervisor: Mr J. Demmers

Second assessor: Dr W.M. van Dolen

(2)

1. Introduction

Nowadays it is more important than ever for a bank to understand how to promote themselves towards the public because communicating only good intentions is no longer enough. Firms are judged on their actions, their actual behaviour

which is the only measure of credibility and trust (Van der Jagt, 2002). Especially with the rise of social media, more and more clients are able to give online comments about the activities of banks and this forces banks to a whole different way of dealing with customers. To keep and

Abstract

In this study the effect of banking activities on the online sentiment is examined using the Triodos bank as a case study. The activities that are central in this research are campaigns and investments. Also it is researched if a positive online sentiment results in a rising share price of banks and vice versa. Furthermore this research investigates the effect of the campaigns and investments of the Triodos bank on the share price of the Green Fund using an event study. Daily collected online comments on

Facebook and Twitter about the Triodos bank are used to analyse these relationships. The online sentiment in this research consists of 2.812 positive messages and 709 negative messages that were collected in the period from January 1, 2013 till June 25, 2013. Results show that campaigns have a positive influence on the online sentiment whereas the relationship with the negative online sentiment is not supported. Investments have a negative influence on the online sentiment rather than a positive influence. Absolute positive messages and share price correlates considerably especially because a share price is hard to predict. The event study shows that there is no relationship between the activities of the Triodos bank and the share price. This means that press releases from the Triodos bank about investments or campaigns does not influence an investor’s assessment of the firm’s

performance which eventually leads to changes in Triodos Green Fund. Recommended future research and limitations are being discussed.

(4)

attract clients it is important to know what clients online think about your activities. The public opinion can increasingly determine which firm is allowed to go through the ‘next round’ and who is

not. The bankruptcy of the DSB bank in 2009 is a good example of the power of social media. Peter Lakerman, a financial expert, stated in a morning talk show that the bank was exploiting mortgage customers. He advised the clients to take their money out of the bank. The news and social media took over this message which created a snow ball effect of negative word of mouth, eventually this led to the bankruptcy of the DSB bank.

Social media is accessible to everyone and this can have a major impact on the reputation of banks due to the speed, convenience and the absence of pressure to exhibit socially desirable behaviour (Litvin, Goldsmith & Pan, 2008). Expressing thoughts, feelings and opinions about products and services via social media is a form of electronic Word – Of – Mouth (eWOM). eWOM can be described as: ‘any positive or

negative statement made by potential, actual, or former customers about a product or company,

which is made available to a multitude of people and institutions via the internet’ (Hennig –

Thurau, Gwinner, Walsh & Gremler, 2004). Billions of people communicate via social media, such as blogs and online social communities. Examples of online social communities are: Facebook, Twitter, Linkedin etc. These online social communities are observable for all kind of firms which enables the estimation of the impact of different activities on sales or other quantities of interest to the firm (Sonnier, McAlister and Rutz, 2011).

Previous studies show that word of mouth (interpersonal communication) is recognized as an influential source of information for consumers. 50% of consumers say that they are ‘likely – to – buy’ as a result of the information

they hear from others during a eWOM conversation (Sonnier, McAlister and Rutz, 2001; Keller & Fay, 2012). eWOM is so influential because consumers typically find other consumers more confidence than advertisers and the information is more trust worthy than information from a controlled source (Buhalis &

(5)

Law, 2008). Other studies also show the importance of the eWOM. Through the internet individuals can make their opinions more accessible to other internet users and are more up – to – date, enjoyable and reliable than

information provided by firms themselves (Zhu & Zheng; Gretzel & Yoo, 2008). Businesses and organizations concerned with reputation management (i.e. banks) find eWOM increasingly important because they are wrestling with the case how positive and negative eWOM can affect existing processes (Goldman, 2008).

Although it is still unclear how different activities of banks can affect the online sentiment, prior studies examined other forms of consumer’

communications like viral marketing. According to Hill, Provost & Volinsky (2006) viral marketing can also be called electronic word of mouth marketing, buzz marketing or network based marketing. Porter & Golan (2006) conclude that the success of a viral campaign depends on the amount of messages a campaign gets which gives the brand or company more exposure. This influences sales and revenues positively. The

challenge for marketers is to communicate in a normal way through the internet. For customers this is an easy way to stay in contact with a firm, it’s low cost, has a minimal response time and a

potential market impact. This all makes it attractive for firms to put effort in making a thoughtful viral marketing campaign (Dobele, Toleman & Beverland, 2005). Understanding the motivation to spread messages and people’s

reaction is very important to advertisers. They need to know what content suits best to reach a large group of people (Ho & Dempsey, 2008). Although there have been studies conducted on viral marketing and the intention of eWOM, there’s never been specifically looked at the effect

of various activities on the online sentiment.

Today large collections of messages on social network sites exist with all different kind of topics. With the help of sentiment analysis these messages can be divided into positive, neutral or negative sentiment to help predict customer behaviour. Multiple studies researched the prediction of online sentiment on real world outcomes. Fung, Yu and Lam (2002)

(6)

demonstrated that online sentiment can predict the stock market its future behaviours. Wolfram (2010) concluded that ‘using Twitter as a source

of near real – time information to predict the price ahead of time can be used to make reasonable profits before the market adjusts itself’. The

online sentiment can’t only predict stock markets

but more real world outcomes like political elections (Tumasjan, Sprenger, Sadner & Welpe, 2010), movie sales (Mishne & Glance, 2006), book sales (Gruhl, Guha, Kumar, Novak & Tomkins, 2005) and detecting influenza epidemics (Culotta, 2010).

To gain insights into the effect of activities of banks on the online sentiment, online comments will be analysed in terms of the sentiment they convey. The aim of this study is to analyse the effect of activities of banks on the online sentiment and whether the online sentiment can predict the share price. As there are many types of activities banks can be involved in it is not feasible to discuss all those in this study. The two activities that are central in this research are:

investments and campaigns. The following main

question is central in this research: ‘What is the

effect of investments and campaigns on the online sentiment and the share price of banks?’ The

study of this thesis is based on one case study of a bank that is from Dutch origin. The bank that is central in this case study is the Triodos bank. Triodos is a global pioneer of green banking of Dutch origin currently making money work for positive social, environmental and cultural change. As of the end of 2012, Triodos had more than 437.000 customers and was worth €5.3

billion. The relationship and operations between the customers and the bank are mainly web based, however in Spain they have physical offices that are preferred by the customers (Triodos,2013). The goal of the case study is to understand what the effect of investments and campaigns is on the online sentiment of the Triodos bank and eventually on the share price of this bank. In the discussion it will be examined whether the results

are generalizable to other banks.

In the following chapter the conceptual framework and the theory regarding the subject of this research will be presented and the

(7)

hypothesis are explained. In chapter 3 the research method will be presented. Subsequently, chapter 4 shows the results of the study and in chapter 5 the conclusion follows. Finally in chapter 6 the discussion and limitations of the study are presented.

(8)

2. Theoretical framework

2.1 eWOM & Online sentiment

As previously described, this online version of word – of – mouth communication between consumers is

called electronic word- of – mouth, i.e eWOM (Lee & Yuon, 2009). Hennig – Thurau et al. (2004) give the following definition of eWOM: ‘any positive or

negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions via the internet’. A eWOM message is not

a personal message that is directly expressed towards another person, it is a virtual message that is available to anyone with an internet connection (Litvin, Goldsmith & Pan, 2008). Consumers assume that fellow consumers generally have no commercial motives when they post a eWOM. The independence of the source ensures that a message is credible and therefore has more influence on the receiver because the source is considered to be authentically (Cheung, Lee & Rabjohn,2008).

The interest in sentiment, mood and affection analysis have been growing over the last couple of years. Especially with the rise of microblogging and the

launch of Twitter in 2006, this new form of social media is getting increasingly important due to the increasing number of potential customers using this medium. Microblogging is a new form of communication in which consumers can express their feeling, thoughts and interests in short messages about all kind of things (Jansen & Zhang, 2009). Microblogging directly impacts eWOM communication because it allows people to share sentiment (positive/negative) about anything, anywhere and to almost anyone that is connected via the internet (Jansen & Zhang, 2009). The motivation of organizations to find out what the consumers think about the brand on blogs and social networks is powered by the proven power of messages consumers can have on others. Consequently, firms have limited control on this development as comments are generated directly by the consumer (Berthon et al., 2008).

To understand what people think about your bank is important because brand choice, in this research the choice to become a customer of the Triodos bank, is greatly influenced by eWOM (Fiske, 1980). As consumers develop expectations about activities of

(9)

firms, a positive or negative sentiment towards an activity can leave the consumer with delight or frustration (Lee & Yuon, 2009). Especially for the banking industry it is critical to know what customers think about your bank because it is a service industry which must prioritize high quality service to its customers (Sarker, Bose, Khan, 2012). The online sentiment can be picked up by everyone on the internet, so it is important that the sentiment towards activities of your company is positive because eWOM often happens between actual and potential consumers (Arndt, 1967). Thus if the online sentiment about your activity is negative, it can be expected that it influences the potential consumer negatively and vice versa. Also it is researched that negative eWOM has a greater effect than positive eWOM (Park & Lee, 2009). Tools have already entered the market to help firms in managing expressions in the micro – blogging area by at least monitoring them (Jansen, Zhang, Sobel & Chowdury, 2009).

‘In the last few years, social networking has become increasingly popular with hundreds of millions of users worldwide. These new websites are not only

useful for keeping in touch with family and friends; it is a new powerful marketing tool (Eley & Tilley, 2009)’. So it is apparent that investigating the online

sentiment is very useful, to help rank the success of a campaign or launch of a new product. It can determine which versions of campaigns or investments are popular and which consumers like or dislike particular features (Vinodhini & Chandrasekaran, 2012). To achieve a high degree of customer satisfaction, and thus positive sentiment, is crucial for product differentiation and for developing a strong relationship with customers (Deng et al., 2009). Companies can respond to the consumers insights they generate through social media monitoring and analysis by modifying their marketing activities (Zabin & Jefferies, 2008). The social media has created a need for new online strategies, marketing actions and allocation of marketing resources (Kaplan & Haenlein2010).

2.2 Activities: Campaign & Investment

In the last five years banks have increasingly chosen social media as a communication channel to communicate product & service changes, marketing campaigns, investments and other happenings that are

(10)

of relevance to customers (Tsys, 2013). The benefit of having access to millions of social media users and the growing importance of social media has been widely acknowledged, banks are now faced with the challenge of how to execute their social strategies online (Senadheera, Vindaya, Warren, Matthew & Leitch, 2011). eWOM is not only a prediction of future sales, it is also a reflection of past and current activities (Duan, Gu & Whinston, 2008). There are limited to no studies at this time that investigated the impact of banking activities on the online sentiment. Blair – Goldensohn, Hannan, McDonald, Neylon, Reis and Reynar (2008) used Google Maps data to analyse consumer sentiments towards department stores, hotels and restaurants using positive/negative classifications. Therefore they could summarize the sentiment regarding different aspects of the services provided by the stores, hotels and restaurants. Godes and Mayzlin (2004) found that the number of postings is positively correlated with a TV show’s

performance. Chen, Wu & Yoon (2004) illustrated that past car sales are positively related to the number of online postings. Jansen et al. (2009) found that 19% of tweets contained mentions of products or brands and that an automated classification system

was able to extract significant differences of customer sentiment.

Previous research demonstrates the link between environmental initiatives and positive affective, cognitive and behavioural responses. The level of campaign relevance is considered to play a crucial role, it is activated when the message of the campaign meets the values, goals or needs of the customer and therefore this will influence the response towards the campaign (Vermeir & Verbeke, 2006). Because of a growing environmental consciousness among consumers and the above mentioned, it is considered that consumers develop a positive attitude effect on brands or firms that are perceived as environmentally sound, in this case the green Triodos bank. (Swenson & Wells, 1997). Also a positive social image among the public is important for companies with a high public visibility, e.g. banks. So one might consider that these companies are expected to share positive messages of their activities (Branco & Rodrigues, 2006). Based on the above mentioned it is expected that green campaigns will have a positive effect on the online sentiment.

(11)

H1: Green campaigns have a positive effect on the

online sentiment

Mood can affect the evaluation of a financial decision. Sometimes a decision of a person is influenced by emotions that have nothing to do with the situation. In other words, if someone is in a good mood it can clearly affect the way they think about an investment, it is likely that they will be more optimistic towards the investment (Johnson & Tversky, 1983). Triodos markets its products by not just being a bank where you can put your money at, but at the same time giving a social return to society, since the money is being used to support projects with social and environmental aims (Tridos,2013). Banks can report on what they are investing in to ensure that their lending and investment policies do not harm the environment (Branco & Rodrigues, 2006). The person – environment (PE) fit theory asserts that: ‘the

congruence between individual characteristics (e.g. values) and environmental characteristics (e.g. organizational values) is an important predictor of attitudes and behaviours such as satisfaction and commitment’ (Kristof-Brown, Zimmerman &

Johnson, 2005). Thus if there is a high PE fit, in this case social and environmental aims, it is likely that it

will affect someone positively and vice versa. From the data it is not clear which person has a high PE fit, therefore I assume that if they talk about the Triodos bank online they have a certain interest towards Triodos and thus towards social and environmental aims. Based on this knowledge, it is expected that if the Triodos makes an investment with social and environmental aims it will have a positive effect on the online sentiment.

H2: Green investments have a positive effect on the

online sentiment

2.3 Share price Triodos Green Fund

The enormity and high variance of information in case of social media presents an interesting opportunity for researching the data into a form that allows for predictions about particular outcomes (Asur & Huberman, 2010). According to Fama (1965) share prices are largely driven by new information, rather than present and past prices. This conclusion makes it hard for share prices to be predicted, it follows a random path and cannot be predicted with more than 50% accuracy (Qian & Rasheed, 2007). On the other hand, several studies have showed that share prices do not walk a random

(12)

path but can indeed be predicted (Gallagher & Taylor, 2002)(Kavussanos & Dockery, 2001). The sentiment can be extracted from online social media, i.e. Twitter, Facebook, to predict changes in various measures.

The share market itself works as a sort of meter of social sentiment. The share market reaches a high when the social mood is high, and a rising share market indicates an increasingly positive social mood trend (Nofsinger, 2005). Findings in Gilbert & Karahalios (2009) research show that the sentiment of millions in a large online community can affect real world setting, like financial markets. Luo (2009) concludes that: ‘the impact of WOM on firm stock

value is often implicitly assumed among practitioners and academics’. He researched the long term impact

of negative word of mouth on cash flows and stock prices. He finds that the higher a firms’ historical

NWOM is, the more shortfalls its future cash flow and share returns will have and vice versa. Chevalier and Mayzlin (2006) and Luo (2007) both find some short – term, immediate significant impact of WOM on sales performance. Checking Twitter or other social media websites for emotional outbursts of any

kind can help to gauge the following day’s share

market mutation (Zhang et al.,2010). The effect of consumer sentiment on financial numbers has also been researched in the early days, Richins (1983) concludes: ‘if the number of consumers experiencing

dissatisfaction is high enough, such responses may have lasting effects in terms of negative image and reduced sales for the firm’. Luo (2007), Gupta and

Zeithaml (2006), Luo and Homburg (2007), Keller (2003), Luo and Bhattacharya (2006), Gruca and Rego (2005) all conclude that a negative sentiment results in unstable cash flows, leading to higher volatilities in stock prices in the long run and more shortfalls in stock returns. Sabherwal, Sarkar & Zhang (2008) researched 160.000 postings from a message board and concluded that positive posting volume positively correlates with share price. Based on the above knowledge, it is expected that if there is a positive sentiment on a particular day, the share price of the Green Fund on the next day will reach a high and vice versa.

H3 : A positive online sentiment results in a rising

share price of the Green Fund

H4: A negative online sentiment results in a falling

(13)

A few studies also addresses the link between the activities of firms and the effect of the share price value. Early research from Horsky and Swyngedouw (1987) showed that changing a company’s name caused stock –market reaction. Also Chaney, Devinney & Winer (1991) concluded that the announcement of new product caused changes in the stock price. Brand extensions and brand attitude has also been researched as indicators for changes in stock price (Lane & Jacobson, 1995; Aaker & Jacobson, 2001). Based on the results it can be expected that activities of banks have a direct impact on the stock price of the firm and thus this may also apply to the case study of this research. A press release of a green investment or campaign of the Triodos bank should influence an investor’s assessment of the firm’s performance which

eventually leads to changes in Triodos Green Fund. A McKinsey study from 2007 revealed that 87% of the respondents of that study are concerned about the social and environmental products they consume (Bonini & Oppenheim, 2007). Tareting green consumers can maximize profits which helps to increase a firm’s performance (Fraj – Adrés et al.

,2008). Therefore the short – term firm performance

should be directly positively affected by the green activities of the Triodos bank. Based on the above mentioned the following hypotheses can be stated:

H5: A green campaign will result in positive share

– price reactions of the Green Fund

H6: A green investment will result in positive share

– price reactions of the Green Fund

Hypothesis 5 and 6 test the direct effect of the activities on the share price without taking in consideration the online sentiment as a mediation factor. To date, no research has been conducted to assess the mediation role of online sentiment on the relationship between activities and share

price. Attitude towards an ad has been assumed to

be a mediation variable in the relationship between advertising and purchase intentions (Mackenzie, Lutz & Belch, 1986). Although this research differs from the conceptual framework presented in this study, the study shows some overlaps that can be used in this study as well. Therefore hypothesis 7 and 8 are stated as following:

H7: Online sentiment will fully mediate the

(14)

H8: Online sentiment will fully mediate the

relationship between investments and share price

In figure 1 the theoretical framework that is used in this study is given.

Campaigns Investments Online sentiment H1 H2 Shareprice H5 H6 H3 H4

(15)

2.4 Case study: Triodos Bank

In order to perform this research a case study is executed with regard to the Triodos bank. Below some background information about this bank is given.

Triodos is a contraction of the Greek words ‘tri’ and ‘hodos’, which means triple road. In their

interrelationships: the individual, society and the natural environment determines the quality of life and human dignity. The Triodos bank wants to contribute to this and that’s why it is reflected in the name.

Triodos bank is a bank originated from the Netherlands and has spread its operations in Germany, United Kingdom, Spain and Belgium (Triodos, 2013). Triodos is one of the most sustainable banks in the world and is named ‘Sustainable Bank of the Year’ by the Financial

Times in 2009 (Financial Times, 2009). Their mission is to make positively social, environmental and cultural changes in the world happen. One of the goals of Triodos is to contribute to a society, which promotes the quality of life and where human dignity plays a central role. Another goal they have is to make it possible for humans, companies and organizations to use their money in ways that it’s beneficial for the

people and the environment and to promote sustainable development (Triodos, 2013). Founded in 1980, it is an independent owned bank by public shareholders, but the shares are held through a trust fund which protects the social and environmental aims of the bank (Cowton and Thompson, 2000). In comparison with other banks, Triodos is still relative unknown in the banking sector but its recent performance has been impressive, with growth in profit of more than 51%. (Triodos, 2013).

(16)

3. Research Methods

In this chapter the methodology is discussed that is used to answer the research question. In addition, the independent, campaigns and investments and the dependent variables, online sentiment and share price, will be operationalized.

3.1 Design and Material

The research tool ‘Tracebuzz’ is used to measure the

online sentiment. This is a research tool for monitoring and analysing social media on different channels and site like: Twitter, Facebook, Youtube, LinkedIn, blogs, forums etc. ‘Tracebuzz’ collected online comments on a daily basis about Triodos on various social platforms from January 1, 2013 till June 25, 2013 (N=176). The online comments about the Triodos bank are in this research the unit of

analysis. In this research the focus is only on messages that are collected from Twitter and Facebook. The total data set consists of a total of 8.260 messages. These comments are analysed manually analysed by ‘Tracebuzz’ on their positive,

negative and neutral sentiments. These comments are separately measured by counting the amount of messages based on the sentiment. Of all the online posts and messages from Twitter and Facebook about the Triodos bank, 2.812 had a positive sentiment, 709 a negative sentiment and 4.739 a neutral sentiment. The data of the share price of the Green Fund is contracted from the website: www.iex.nl, also from January 1, 2013 till June 25, 2013. Quantitative research is applied using the statistical software SPSS and Excel 2013. In figure 2 the number of absolute negative and absolute positive per day is shown.

0 20 40 60 80 100 120 140 160 180 1-ja n 8-ja n 15 -ja n 22 -ja n 29 -ja n 5-fe b 12 -f eb 19 -f eb 26 -f eb 5-m rt 12 -m rt 19 -m rt 26 -m rt 2-ap r 9-apr 16 -a pr 23 -a pr 30 -a pr 7-m ei 14 -m ei 21 -m ei 28 -m ei 4-ju n 11 -ju n 18 -ju n 25 -ju n

(17)

3.2 Independent variables

3.2.1 Campaigns

The independent variable campaigns is measured on the basis of the total amount of campaigns of the Triodos bank on each day in the period from January 1, 2013 till June 25,2013. The messages of the campaigns are contracted from the Facebook and the Twitter account of the Triodos bank. In the dataset the variable, campaigns, have been noted as: 0 = no campaign, 1 = campaign. Figure 3 shows the number of positive and negative messages (online sentiment) in relationship with the campaigns of the Triodos bank.

3.2.2 Investments

The second independent variable investments is measured on the basis of the total amount of investments of the Triodos bank on each day in the period from January 1, 2013 till June 25,2013. The messages of the investments are contract from the Facebook and the Twitter account of the Triodos bank. In the dataset the variable, investments, have been noted as: 0 = no investment, 1 = investment. Figure 4 shows the number of positive and negative messages (online sentiment) in relationship with the investments of the Triodos bank.

Table 1 shows an overview of the expressions on investments and campaigns of the Triodos bank on the specific days.

(18)

3.3 Dependent variables

3.3.1 Online sentiment

The dependent variable online sentiment is measured on the basis of the amount of electronic messages about Triodos and the sentiment of the messages. Wysocki (1998) uses pure message counts to report that variation in daily message posting volume is related to news and earnings announcements. Also Tumarkin & Whitelaw (2001) use this method of counting positive and negative messages per day to find a relationship between internet postings and stock prices. Furthermore Antweiler & Frank (2004) and Sprenger & Welpe (2010) find that message volume can forecast real world outcomes. In the paragraph ‘Design and Material’ is explained how the

analysis of online social buzz by ‘Tracebuzz’ is done. The research tool was monitoring on a full and

continuous basis, 24/7. This measurement of online messages gives a clear picture of how people talk about the Triodos bank in the Netherlands. A distinction is made between, negative, positive and neutral comments which explain what people online think about Triodos’ investments and campaigns. A few examples of the three different kinds of messages are given below.

An example of a positive message recognized by the research tool of ‘Tracebuzz:

Twitter 03-01-2013 21:35: ‘Mooie nieuwe iPhone

app van Triodos!’

An example of a negative message recognized by the research tool of ‘Tracebuzz’:

Facebook 13-06-2013 18:16: ‘Ik haat de Triodos

Bank.’

Date Sort Link

3-1-2013 Campaign https://www.facebook.com/photo.php?fbid=459720394091635&set=a.179630652100612.48588.138873442843000&type=1&theater 8-1-2013 Investment https://www.facebook.com/photo.php?fbid=462085557188452&set=a.179630652100612.48588.138873442843000&type=1&theater 21-1-2013 Campaign https://twitter.com/chielblokvoort/status/293342484455620610 29-1-2013 Campaign https://twitter.com/TriodosNL/status/296231224140242945 11-2-2013 Investment https://twitter.com/TriodosNL/status/300921760143007744 15-2-2013 Campaign https://www.facebook.com/photo.php?fbid=482871461776528&set=a.179630652100612.48588.138873442843000&type=1&relevant_count=1 5-3-2013 Investment http://twitter.com/TriodosNL/statuses/308881784270094336 2-4-2013 Investment https://www.facebook.com/photo.php?fbid=502759719787702&set=a.179630652100612.48588.138873442843000&type=1&relevant_count=1 4-4-2013 Campaign https://twitter.com/TriodosNL/status/319717703059382273 24-4-2013 Investment http://www.triodos.nl/nl/over-triodos-bank/nieuws/persberichten/bnp-paribas-groenfonds-gaat-op-in-triodos-groenfonds/ 22-5-2013 Campaign https://twitter.com/TriodosNL/status/337222389920317440 29-5-2013 Campaign https://twitter.com/TriodosNL/status/340024759725203457 17-6-2013 Campaign https://twitter.com/TriodosNL/statuses/344201009473671170

(19)

An example of a neutral message recognized by the research tool of ‘Tracebuzz’:.

Twitter 23-06-2013 16:13: ‘Gisteren waren klanten

van de Triodos Bank te gast op Landgoed Welna en Butler Laurens verzorgde de lunch en receptie.’

The online sentiment is separately measured by counting the amount of messages based on their positive, negative and neutral sentiment.

3.3.2 Share price

The second independent variable share price green

fund is measured on the basis of the daily data of the

share price of the Triodos Green fund.

This data is obtained from the website: www.iex.nl. On this site it is possible to see historical share price data of the Green fund. So it was not necessary to collect the data per day, it has been jointly collected on November 7, 2013. The stock market is not open 7 days a week, so this caused a lot of missing data (28%) in SPSS for the variable share price. To fill up this missing data, the share price of the Fridays and the Mondays are used to set the average price for the missing data. This theory is viewed by many researchers as a viable method of dealing with missing data (Acock,2005).

An example of setting up the average price is given below.

Share price of Friday January 11, 2013: €56,64 Share price of Monday January 14, 2013: €56,59 Share price of Saturday 12 and Sunday 13 of January, 2013:€56,64+ €56,59 = €113,23 /2 = €56,62

This is done for every Saturday and Sunday in the period from January 1, 2013 till June 25,2013.

(20)

3.4 The Models

3.4.1 Linear regression

The hypotheses 1 and 2 are tested using a linear regression. The model of the hypothesis are as following:

Hypothesis 1 and 2:

Y = β0 + β1 * Investments + β2 * Campaigns+ 

Where Y is the online sentiment (positive / negative); β0 is the constant term; β1 and β2 describes the effect

of the regressors and is the error term.

In order to see if the investments and campaigns also have an effect on the online sentiment of the following day a time lag predicator is added.

Hypothesis 1 and 2:

Y = β0 + β1t-1 * Investments + β2t-1 * Campaigns+ 

3.4.2 Correlation & linear regression

The hypotheses 3 and 4 are tested using a correlation and a regression analysis. These measures are used to find a statistical relationship between the online sentiment and the share price of the Triodos bank. In general it is very hard to predict the stock price because there is simply no proven market predictive model. Therefore to analyse the correlation output use is made of the norm Cohen (1992). More information about this norm will be given in the results chapter.

The linear regression model of hypotheses 3 and 4 are as following:

Hypothesis 3:

Yt = β0 + β1t-1 * positive online sentiment+ 

Hypothesis 4:

Yt = β0 + β1t-1 * negative online sentiment+ 

Where Y is the share price; β0 is the constant term; β1 describe the effect of the regressor and is the error term.

3.4.3 Event study

For hypotheses 5 and 6 the event study is conducted to perform the statistical analysis. Investors may need some days to evaluate new information, therefore the share prices do not react on the actual day. The event study test if there is a statistical significant reaction to the activities in following days. Pruit et al. (2004):

‘The event study is the standard assessment metric for the measurement of the net economic value of any corporate event – marketing or otherwise – for which precise announcement dates may be obtained’.

Several things have to be taken into consideration. First, the market is efficient, it reacts rapidly to new information. So when this information influences the share returns shareholders will immediately respond

(21)

to it. With this event study it is possible to link positive and negative changes in share returns to specific events, in this case the campaigns and investments of the Triodos bank (McWilliams & Siegel, 1997). A second thing that needs to be taken into consideration is that the campaigns and the investments of the Triodos bank are publicly announced in the press. So market investors did not have earlier information on the event.

To test if the returns of the Triodos Green Fund differ from returns in the past, daily interest rates are needed. The stock price of the Triodos bank on the next day is used in order to calculate the expected (abnormal) return. Im, Dow & Grover (2001) give the following definition of abnormal returns: ‘unbiased

estimate of changes in the market value of the firm during the event period, which reflects the price reaction to the event’. If the abnormal return is small

or not significant, one can conclude that this event is not important for the market. To estimate the normal returns and calculate the abnormal returns, a period

1_Where:

Ṝit = mean abnormal return of firm i in the

estimation period

Rit = actual return for firm i on the event day t

before and after the event date will be calculated. MacKinlay (1997) conducted a time line for the event study (see figure 5). The estimation period is used to calculate the expected normal return on the event day and is compared to the return calculated on the announcement date. The difference between these two returns is called the abnormal return.

Figure 5 Time line

To determine the length of the estimation period and the event window it is important that there are no events that overlap because otherwise this can cause ‘confounding effects’ (McWilliams, 1997). The

window that is used in this research is derived from the work of Miyazaki and Morgan (2001). They use an estimation window of 120 days and an event window of 10 days.

The abnormal return is calculated as following:

(22)

The above formula indicates that the mean abnormal return and the actual return needs to be estimated in order to derive the abnormal return. In the literature there are several models used to calculate the abnormal return. In the research the market return model is used because this model is accepted as the dominant model to estimate abnormal returns according to Im et al. (2001). The market model takes the movement of the world market interest rate into consideration (MacKinlay, 1997).

In figure 6 the share price of the Triodos Green Fund and the world interest rate are shown in the period from January 1, 2013 till June 25,2013

αi = intercept

βi = systematic risk

Rmt = actual return for time period t

εit = error term

The normal return is calculated as following:

Ṝit = αi + βiRmt + εit 1 (2)

The actual return is calculated as following:

Ri = αi + β*Rm (3)

In formula (2) and (3) the market parameter is included, therefore the values of day t of the interest rates need to be collected.

In order to test statistical significance for H5 and H6 a t – test, displayed by equation 4, will be performed for every day in the event window. At three different

55 55,2 55,4 55,6 55,8 56 56,2 56,4 56,6 56,8 57 0 0,5 1 1,5 2 2,5

01-01-2013 - 25-6-2013

Interest rate Triodos Green Fund

(23)

significant levels, 10%, 5% and 1% is checked whether the results are different from zero.

𝑡 =AR

𝑆𝑒 (4)

One green campaign and one green investment are chosen to test the positive share price reaction of the Green Fund. The reason why these two activities are chosen is because the Triodos bank made an official press release about it and on Twitter and Facebook it created a lot of comments. The green campaign that is central in this event study is the press release of the Triodos bank on 22-05-2013 ‘Nukuhiva wint Hart – Hoofdprijs 2013’ (figure 7).

The green investment that is central in this event study is the press release of the Triodos bank on 24-04-2013 regarding the take –over of the Green Fund of BNP Paribas (figure 8). They do this by issuing Triodos Green shares to BNP Paribas Green Fund, after this the latter fund will be eliminated. This transaction will be effected on May 31,2013 (Triodos, 2013). Because the effect of this take – over took

place on 31-05-2013, this date (t=0) is chosen for the event study instead of the actual press release on 24-04-2013.

3.4.4 Mediation analysis

A mediation analysis is performed to measure hypothesis 7 and 8. The first step in this analysis is to perform a linear regression analysis with

campaigns or investments as the independent

variable and online sentiment as dependent variable. The second step is also to perform a linear regression but now changing the dependent variable to share price. The third step is to perform a stepwise regression with online

sentiment and investments or campaigns as

independent variable and share price as dependent variable. The last step is to perform a Sobel test to see if there is a mediation effect (Baron & Kenny, 1986).

Figure 7 ‘Nukuhiva wint Hart – Hoofdprijs 2013’

(24)

4. Results

The following chapter represents the results regarding the above mentioned hypotheses and methods in order to reject or accept the hypotheses.

4.1 Positive effect of campaigns versus investment In hypothesis 1 it was investigated if campaigns had a positive influence on the online sentiment. To test this relationship, 2 regression analysis were used. First the independent variable campaigns was tested on the dependent variable negative online sentiment (β= 1,405 ; P=0,613). Secondly, the independent variable campaigns was tested on the dependent variable positive online sentiment (β= 27,125 ; P= 0,001). The relationship between campaigns and the

positive online sentiment is supported (P= <0,05),

whereas the relationship with negative online

sentiment is not supported. Thus this is in line with

hypothesis 1: ‘Green campaigns have a positive influence on the online sentiment’. In order to see if

the campaigns also have an effect on the online sentiment of the following day the same analysis including the time lags were conducted. The relationship between campaigns and the negative

online sentiment the next day is not significant (β=

1,643 ; P=0,555). Also the relationship between

campaigns and positive online sentiment the next day

is not significant ((β= 17,493 ; P=0,433). It can be concluded that green campaigns have a positive influence on the same day whereas there is no effect on the next day.

For hypothesis 2 I also used 2 regression analysis to examine the relationship between investment and

online sentiment. First the independent variable investments was tested on the dependent variable negative online sentiment (β=4,318 ; P=0,051).

Secondly, the independent variable investments was tested on the dependent variable positive online

sentiment (β= 10,506; P=0,385). Based on these

findings, hypothesis 2 is not supported. It can be said that the relationship between investments and

negative online sentiment is almost supported

(P≠0,05), so it can be concluded that investment have a negative influence on the online sentiment rather than a positive influence.

In order to see if the investments also have an effect on the online sentiment of the following day the same analysis including the time lags were conducted. The relationship between investments and the negative

(25)

3,012; P=0,438). Also the relationship between

investments and positive online sentiment the next day

is not significant ((β= 23,224 ; P=0,084). It can be concluded that there is no effect of green investments on the next day online sentiment.

4.2 Correlation and regression online sentiment and share price

In order to test hypothesis 3 and 4 a correlation analyses was conducted. The online sentiment is split up in the positive, negative and neutral variables for a clear interpretations of the results. Table 2 presents an overview of the correlations that were founded. The result of the correlation between share price and absolute positive and negative suggests significant positive correlations (P= 0,015 ; P= 0,016) at a 5% significant level. According to Cohen (1992) there are different standards for the interpretation of correlation coefficients. From Cohen his article it can be seen that the correlations (significance of product moment r) of 0.184 and 0.180 can be called ‘small to medium’. However looking at the outcome of the variable ‘share price’ an outcome of 0.184 and 0.180

is considerable especially because a stock price is very difficult to predict.

Variable Shareprice Absolute _Positive Absolute _Negative Absolute _Neutral

Shareprice - Absolute Positive 0,184* - Absolute Negative 0,180* 0,341** - Absolute Neutral -0,176 0,440** -0,212** -

*. Correlation is significant at the 0.05 level **. Correlation is significant at the 0.01 level

A regression analysis is also performed to test hypothesis 3 and 4. The independent variable share

price was tested on the dependent variable positive online sentiment (β= 10,925 ; P= 0,010) and negative online sentiment (β= 3,198 ; P= 0,020). It can be

concluded that a positive online sentiment results in a rising share price of the Green Fund of the Triodos bank and vice versa, which is in line with the correlation analysis.

4.3 Event study

To test hypothesis 5 and 6 an event study was conducted. The different parameters in equation 1,2,3 and 4 were estimated and used to compute the abnormal return (AR) and the t – test. In hypothesis 5 it was researched if a green campaign will result in a

(26)

positive stock – price reaction of the green fund. In hypothesis 6 it was researched if a green investment will result in a positive stock – price reaction of the green fund. The results of this test, together with the results of hypothesis 5, are presented in table 3. On the event day of the campaign the absolute return is negative (AR = -0,00057) which is in line with hypothesis 5. Also t+1 and t+2 of the campaign event are negative. Looking at the AR, of the event day of the investment, this is negative (AR = -0,00077). Further, looking at t+1 and t+2 the absolute return of these days are both positive, which can insinuate a later positive effect of the investment on the stock price value of the Triodos Green Fund. This is in line with hypothesis 6, namely the AR of t+1 and t+2 are both positive.

However, to say something meaningful about these results and in order to support or reject the hypothesis 5 and 6, t –tests are performed to test the abnormal returns on significance. The results of the t-test are also presented in table 3. All the t – test values are checked at a significance level of both 10% and 5%. The t-test for the AR of H5 (campaign) is not significant (t= -0,28783) at both significant levels, so

the null hypothesis regarding the effect of campaign on the stock price of the Triodos Green Fund can’t be rejected. The AR of t+1 and t+2 of H6 are both positive, however to be able to rejected the null hypothesis, both absolute returns need to be significant, which is not the case at both significant levels (t+1=0,555688 ; t+2= 0,62861). Thus this means that both hypotheses are not supported (MacKinlay, 1997).

4.4 Mediation analysis

The first step is to test the relationship between activities and share price in order to examine whether there is a possibility that there is a mediator in this context (Baron & Kenny, 1986). The result of this regression analysis is not significant (β= 32,182 ; P= 0,537), so it is certain that there is no mediation effect in this study.

(27)

(28)

5. Conclusion & Discussion

This study researched if the activities of banks had any influence on the online sentiment and eventually on the share price. A case study of the Triodos bank was used to test these relationships. The online sentiment in this study consists of 2.812 positive messages and 709 negative messages that were collected by ‘Tracebuzz’ in the period from January

1, 2013 till June 25, 2013. The following section will discuss the results and provides the limitations and suggestions for future research.

Hypothesis 1, which stated that green campaigns had a positive influence on the online sentiment, is accepted. This is in line with previous research that stated that campaign relevance is considered to play a crucial role, this is activated when the message of the campaign meets the values of the customer which results in a positive attitude towards the campaign (Vermeir & Verbeke, 2006 ; Swenson & Wells, 1997). So there is a reason to assume that when a campaign is aligned with the expectations and values of the consumers, it will created a positive online sentiment towards the campaign.

It was also expected that if a bank makes an investment it will have a positive effect on the online sentiment. However this study shows that the green investments of the Triodos bank have a negative influence on the online sentiment rather than a positive influence. This is a remarkable outcome, because when the investment meets the social and environmental aims of a person, it is likely that it will affect someone positively (Kristof – Brown, Zimmerman & Johnson, 2005). However this is not the case in this study, a reason for this outcome may be that the financial industry is still in the aftermath of the financial crisis and this is affecting the trust between society and financial services (Bonson & Flores, 2011). Especially because many people blame the banks for the crisis this can result in feelings of anger and hence negative sentiment to the investment activity. It is wise for banks to develop a marketing plan in a way that oppose negative media coverage (Coombs & Holladay, 2007). Also negative eWOM has more impact than positive eWOM (Park & Lee, 2009), it is therefore important for banks to do an investment that fits the expectations and values of the customer which will create a positive online sentiment towards the investment.

(29)

For both hypothesis 1 and 2 a time lag predictor was added to the regression formula to test if there was an effect on the online sentiment of the following day. Both hypothesis that included a time lag were not significant, whereas the above results shows that the hypothesis without a time lag predictor were significant.

Subsequently this study looked at the correlation and the regression analysis between the online sentiment and the share price of a bank. The results of this study show that a positive online sentiment results in a rising share price of the Green Fund of the Triodos bank and vice versa. This was also expected by researchers that state that the higher a firm’s negative

word of mouth is, the more shortfalls its future cash flow and share returns will have and vice versa (Luo, 2009 ; Luo and Homburg, 2007 ; Sabherwal, Sarkar & Zhang, 2008). Checking Twitter or other social media websites for emotional outbursts of any kind can help to gauge the following day’s share market mutation (Zhang et al.,2010). The result of the correlation between share price and absolute positive sentiment and negative sentiment is positive and significant. So this study gives prove of the fact that

the share price of the Green Fund will rise regarding a positive online sentiment and will fall regarding a negative online sentiment.

Another aspect of this study looked at the press release of an investment or campaign of the Triodos bank in relation with share price. An event study was used to research if a press release would influence an investor’s assessment of the firm’s performance

which eventually would lead to changes in the Triodos Green Fund. It was stated that a green campaign or a green investment leads to a positive share price reactions of the Triodos Green Fund but this is not the case according to this study. So this study finds opposite findings to the studies that did find an effect (Devinney & Winer, 1991 ; Aaker & Jacobson, 2001). This non – effect could be explained because the Triodos Green Fund is a fund. This means that funds are in general less volatile and react slowly on activities of banks so that’s why it could be

possible that the short – term effect is not visible (Pástor & Stambaugh, 2012).

It was the first time that the mediation effect of online sentiment on the association between activities and share price was researched. No mediation effect was

(30)

found in this study. Comparable previous research did find a mediation effect between advertising and purchase intentions with attitude as the mediation factor. The reason that there was no mediation effect found in this study can be due to the above mentioned problem of the Triodos Green Fund being a fund.

In spite of the fact that there were some outcomes of this study that were not significant, there are some important implications for the Triodos bank. It has been found that people do respond differentially to different activities. It appears that when Triodos talks online about campaigns it has a positive effect on the online sentiment which also correlates with a higher share price of the Triodos Green Fund. It is also found that when Triodos talks about investments this has a negative effect on the online sentiment. So an implication for the Triodos bank is to talk about campaigns on social media and not so much about investments due to the reason of the aftermath of the financial crisis. An investment costs most of the time money, which can set people off talking about this subject online.

For banks in general it is important to know what customers online think about your activities and

brand, because the public opinion can increasingly determine which bank is allowed to go through to the ‘next round’ and who is not. Also it is important that

the online sentiment is positive because eWOM often happens between actual and potential consumers (Arndt, 1967). From this study it is clear that activities of banks should be aligned with the expectations and values of the consumers, it will create a positive online sentiment towards the bank. This is especially important because the online sentiment can cause a positive or negative effect on the share price of a bank.

6. Limitations & future research

Despite the interesting findings that this research has yielded, it also has some limitations. A first possible criticism to this study is that use is made of existing data that is collected by the Triodos bank itself via ‘Tracebuzz’. It is better to collect own data that fits

the conceptual model.

Furthermore the online messages were collected in a six months period. It would be interesting to use a longer period and see if differences appear in the way people react to different kind of activities.

(31)

The dataset is only focused on messages of the Triodos bank. When more banks would have been compared, on the basis of the online sentiment they convey, a comparison study could have been made about the different kind of sentiment towards different banks. This would have made this research more reliable for statements about generalization. Also for this research use is made of only the reactions on Twitter and Facebook. Other reactions of different social media sites founded by ‘Trazebuzz’ were not reliable.

In this study the variable ‘online sentiment’ was tested on the variable ‘share price’. It would have

been interesting to turn this around and see what kind of online sentiment arises from the share price of different banks. This is a topic that can be researched in the future.

So follow up studies about this subject will need to focus on a longer period of data collection and more banks to find some interesting differences. Also more social media sites should be involved in follow up studies to provide a broader picture.

(32)

References

ACOCK, A.C., 2005. Working with missing values. Journal of Marriage and Family, 67(4), pp. 1012-1028.

ARNDT, J., 1967. Word of mouth advertising. Advertising Research Foundation.

ASUR, S. and HUBERMAN, B.A., 2010. Predicting the future with social media, Web

Intelligence and Intelligent Agent Technology (WI-IAT), 2010 IEEE/WIC/ACM International Conference on 2010, IEEE, pp. 492-499.

BERGER, J. and SCHWARTZ, E.M., 2011. What drives immediate and ongoing word of mouth? Journal of Marketing Research, 48(5), pp. 869-880.

BERTHON, P., PITT, L. and CAMPBELL, C., 2008. Ad Lib: WHEN CUSTOMERS CREATE THE AD. California management review, 50(4), pp. 6-30.

BIELAK, D., BONINI, S.M. and OPPENHEIM, J.M., 2007. CEOs on strategy and social issues.

McKinsey Quarterly, 10(1),.

BLAIR-GOLDENSOHN, S., HANNAN, K., MCDONALD, R., NEYLON, T., REIS, G.A. and REYNAR, J., 2008. Building a sentiment summarizer for local service reviews, WWW

Workshop on NLP in the Information Explosion Era 2008, pp. 14.

BONINI, S. and OPPENHEIM, J., 2008.

Cultivating the green consumer. Stanfold Social

Innovation Review, 6(4), pp. 56-61.

BONSÓN, E. and FLORES, F., 2011. Social media and corporate dialogue: the response of global financial institutions. Online Information

Review, 35(1), pp. 34-49.

BRANCO, M.C. and RODRIGUES, L.L., 2006. Communication of corporate social

responsibility by Portuguese banks: a legitimacy

theory perspective. Corporate Communications:

An International Journal, 11(3), pp. 232-248.

BUHALIS, D. and LAW, R., 2008. Progress in information technology and tourism

management: 20 years on and 10 years after the Internet—The state of eTourism research.

Tourism management, 29(4), pp. 609-623.

BUIL‐CARRASCO, I., FRAJ‐ANDRÉS, E. and MATUTE‐VALLEJO, J., 2008. Corporate environmentalism strategy in the Spanish consumer product sector: a typology of firms.

Business Strategy and the Environment, 17(6),

pp. 350-368.

CHANEY, P.K., DEVINNEY, T.M. and

WINER, R.S., 1991. The impact of new product introductions on the market value of firms. The

Journal of Business, 64(4), pp. 573-610.

CHEN, P., WU, S. and YOON, J., 2004. The impact of online recommendations and consumer feedback on sales.

CHEUNG, C.M., LEE, M.K. and RABJOHN, N., 2008. The impact of electronic word-of-mouth: The adoption of online opinions in online customer communities. Internet Research, 18(3), pp. 229-247.

CHEVALIER, J.A. and MAYZLIN, D., 2006. The effect of word of mouth on sales: Online book reviews. Journal of Marketing Research,

43(3), pp. 345-354.

COHEN, J., 1992. A power primer.

Psychological bulletin, 112(1), pp. 155.

COOMBS, W.T. and HOLLADAY, S.J., 2007. The negative communication dynamic:

Exploring the impact of stakeholder affect on behavioral intentions. Journal of Communication

Management, 11(4), pp. 300-312.

COWTON, C.J. and THOMPSON, P., 2001. Financing the social economy: A case study of

(33)

Triodos Bank. International Journal of Nonprofit

& Voluntary Sector Marketing, 6(2), pp. 145.

CULOTTA, A., 2010. Detecting influenza outbreaks by analyzing Twitter messages. arXiv

preprint arXiv:1007.4748, .

DOBELE, A., TOLEMAN, D. and

BEVERLAND, M., 2005. Controlled infection! Spreading the brand message through viral marketing. Business horizons, 48(2), pp. 143-149.

DOBELE, A., TOLEMAN, D. and

BEVERLAND, M., 2005. Controlled infection! Spreading the brand message through viral marketing. Business horizons, 48(2), pp. 143-149.

EAST, R., HAMMOND, K. and LOMAX, W., 2008. Measuring the impact of positive and negative word of mouth on brand purchase probability. International Journal of Research in

Marketing, 25(3), pp. 215-224.

ELEY, B. and TILLEY, S., 2011. Online

marketing inside out. Uitgeverij Thema.

FAMA, E.F., 1965. The behavior of stock-market prices. Journal of business, , pp. 34-105. FAURE, G. and SHAKUN, M., 1999.

Introduction to the Special Issue on Intercultural Negotiation. Group Decision & Negotiation,

8(3), pp. 183-185.

FISKE, S.T., 1980. Attention and weight in person perception: The impact of negative and extreme behavior. Journal of personality and

social psychology, 38(6), pp. 889.

FUNG, G.P.C., YU, J.X. and LAM, W., 2002. News sensitive stock trend prediction. Advances

in Knowledge Discovery and Data Mining.

Springer, pp. 481-493.

FUNG, G.P.C., YU, J.X. and LAM, W., 2002. News sensitive stock trend prediction. Advances

in Knowledge Discovery and Data Mining.

Springer, pp. 481-493.

GALLAGHER, L.A. and TAYLOR, M.P., 2002. Permanent and Temporary Components of Stock Prices: Evidence from Assessing

Macroeconomic Shocks. Southern Economic

Journal, 69(2),.

GAYO-AVELLO, D., 2011. Don't Turn Social Media Into Another 'Literary Digest' Poll.

Communications of the ACM, 54(10), pp.

121-128.

GILBERT, E. and KARAHALIOS, K., 2009. Predicting tie strength with social media,

Proceedings of the SIGCHI Conference on Human Factors in Computing Systems 2009,

ACM, pp. 211-220.

GODES, D. and MAYZLIN, D., 2004. Using online conversations to study word-of-mouth communication. Marketing Science, 23(4), pp. 545-560.

GOLDMAN, E., 2008. Brand Spillovers.

Harv.JL & Tech., 22, pp. 381.

GRUCA, T.S. and REGO, L.L., 2005. Customer satisfaction, cash flow, and shareholder value.

Journal of Marketing, 69(3), pp. 1-130.

GRUHL, D., GUHA, R., KUMAR, R., NOVAK, J. and TOMKINS, A., 2005. The predictive power of online chatter, Proceedings of the

eleventh ACM SIGKDD international conference on Knowledge discovery in data mining 2005,

ACM, pp. 78-87.

GUPTA, S. and ZEITHAML, V., 2006.

Customer metrics and their impact on financial performance. Marketing Science, 25(6), pp. 718-739.

HENNIG‐THURAU, T., GWINNER, K.P., WALSH, G. and GREMLER, D.D., 2004. Electronic word‐of‐mouth via consumer‐opinion platforms: What motivates consumers to

(34)

articulate themselves on the Internet? Journal of

interactive marketing, 18(1), pp. 38-52.

HILL, S., PROVOST, F. and VOLINSKY, C., 2006. Network-based marketing: Identifying likely adopters via consumer networks.

Statistical Science, , pp. 256-276.

HIRSHLEIFER, D. and SHUMWAY, T., 2003. Good day sunshine: Stock returns and the weather. The Journal of Finance, 58(3), pp. 1009-1032.

HO, J.Y. and DEMPSEY, M., 2010. Viral marketing: Motivations to forward online

content. Journal of Business Research, 63(9), pp. 1000-1006.

HORSKY, D. and SWYNGEDOUW, P., 1987. Does it pay to change your company's name? A stock market perspective. Marketing Science,

6(4), pp. 320-335.

IM, K.S., DOW, K.E. and GROVER, V., 2001. Research report: A reexamination of IT

investment and the market value of the firm—An event study methodology. Information systems

research, 12(1), pp. 103-117.

IM, K.S., DOW, K.E. and GROVER, V., 2001. Research report: A reexamination of IT

investment and the market value of the firm—An event study methodology. Information systems

research, 12(1), pp. 103-117.

JANSEN, B.J., ZHANG, M., SOBEL, K. and CHOWDURY, A., 2009. Twitter power: Tweets as electronic word of mouth. Journal of the

American Society for Information Science and Technology, 60(11), pp. 2169-2188.

JANSEN, B.J., ZHANG, M., SOBEL, K. and CHOWDURY, A., 2009. Twitter power: Tweets as electronic word of mouth. Journal of the

American Society for Information Science and Technology, 60(11), pp. 2169-2188.

JOHNSON, E.J. and TVERSKY, A., 1983. Affect, generalization, and the perception of risk.

Journal of personality and social psychology,

45(1), pp. 20.

KAMSTRA, M.J., KRAMER, L.A. and LEVI, M.D., 2003. Winter blues: A SAD stock market cycle. American Economic Review, , pp. 324-343.

KAPLAN, A.M. and HAENLEIN, M., 2010. Users of the world, unite! The challenges and opportunities of Social Media. Business

horizons, 53(1), pp. 59-68.

KAVUSSANOS, M.G. and DOCKERY, E., 2001. A multivariate test for stock market efficiency: the case of ASE. Applied Financial

Economics, 11(5), pp. 573-579.

KELLER, E.D. and FAY, B., 2012. Word-of-Mouth Advocacy: A New Key to Advertising Effectiveness. Journal of Advertising Research,

52(4), pp. 459-464.

KILBOURNE, W.E., 1998. Green marketing: A theoretical perspective. Journal of Marketing

Management, 14(6), pp. 641-655.

KRISTOF‐BROWN, A.L., ZIMMERMAN, R.D. and JOHNSON, E.C., 2005. CONSEQUENCES OF INDIVIDUALS'FIT AT WORK: A META‐ ANALYSIS OF PERSON–JOB, PERSON– ORGANIZATION, PERSON–GROUP, AND PERSON–SUPERVISOR FIT. Personnel

Psychology, 58(2), pp. 281-342.

LANCE, P. and GUY J, G., 2006. From subservient chickens to brawny men: A comparison of viral advertising to television advertising. Journal of Interactive Advertising,

6(2), pp. 4-33.

LANE, V. and JACOBSON, R., 1995. Stock market reactions to brand extension

announcements: The effects of brand attitude and... Journal of Marketing, 59(1),.

The effect of the activities of banks on the online sentiment and the share price : a case study of the Triodos bank

University of Amsterdam

Faculty of Economics and Business