Trust-generation in crowdfunding:
Exploratory research to trust-inducing aspects in lending-based
crowdfunding in the Netherlands
Name: Tymen Bijleveld Student number: 10001717
Final draft submission date: 28-08-2015
MSc. in Business Administration – Entrepreneurship and Innovation track University of Amsterdam
First supervisor: Dhr. Dr. G. T. (Tsvi) Vinig Second supervisor:
Statement of originality
This document is written by Student Tymen Bijleveld who
declares to take full responsibility for the contents of this
document.
I declare that the text and the work presented in this
document is original and that no sources other than those
mentioned in the text and its references have been used in
creating it.
The Faculty of Economics and Business is responsible
solely for the supervision of completion of the work, not
for the contents.
“It’s not whether you get knocked down, it’s whether you get back up”
- Vincent Thomas “Vince” Lombardi
Foreword and Acknowledgements
This thesis is written as part of the completion of my Master of Science in Business Administration for the track Entrepreneurship and Innovation. When asked a few months back to describe the process of writing a thesis in one word I would have answered
‘frustration’. Now that I am handing in my final draft I’m not supporting that vision anymore. Asked now I would answer that the key concept in the process of writing a thesis is ‘trial-and-error’, and this is a positive view. Writing a master thesis is, as it should be, a last test of your skills. One of the most important skills you learn during your study is excessively tested when writing your thesis. During your study you learn to fail and try again, find
different ways of achieving your goal and to cope with these short-time failures. This skill of being able fail and try again and finding different ways is, in my humble opinion, one of the most important skills, in work as well as in life. To quote one of America’s most successful football coaches Vince Lombardi, “It’s not whether you get knocked down, it’s whether you get back up”.
It is important to realize that the accomplishments you achieve are more than just a result of your own efforts. Among others, the support you get from people has contributed to these achievements. With this in mind I would like to thank some people for supporting me during this final achievement to complete my Master’s. First of all I would like to thank Dr. Tsvi Vinig for his supervision during the whole process and for his constructive feedback to improve my thesis. I would like to thank my interviewees for sharing their vision on the subject and my survey participants for sharing their opinions. Special thanks to the Dutch lending-based crowdfunding platform Kapitaal Op Maat for sharing my survey in their newsletter, helping me find enough participants. A last thanks goes out to all my friends and my family their enduring support, during this process and throughout my life.
Content
Foreword'and'Acknowledgements'222222222222222222222222222222222222222222222222222222222222222'4! Abstract'22222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'7! 1.'Introduction'22222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'8!
An introduction to lending-based crowdfunding'22222222222222222222222222222222222222222222222222222222'8!
Problem definition'2222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'10!
Academic relevance'22222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'10!
Social relevance'2222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'12!
Structure of the thesis'222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'13!
2.'Literature'review'22222222222222222222222222222222222222222222222222222222222222222222222222222222'15!
The theory of planned behaviour'2222222222222222222222222222222222222222222222222222222222222222222222222'15!
The elaboration likelihood model of persuasion'2222222222222222222222222222222222222222222222222222222'17!
Informational cascading and the herding-behaviour bias'2222222222222222222222222222222222222222222'19!
E-commerce and websites'2222222222222222222222222222222222222222222222222222222222222222222222222222222222'21!
3.'The'conceptual'model'on'crowdfunding'and'propositions'22222222222222222222222222'24! 4.'Methodology'22222222222222222222222222222222222222222222222222222222222222222222222222222222222222'30!
The mixed research approach'22222222222222222222222222222222222222222222222222222222222222222222222222222'30!
Semi-structured interviews'22222222222222222222222222222222222222222222222222222222222222222222222222222222'31!
Online questionnaires'222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'33! Data collection:!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!34! Data analysis:!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!34!
5.'Results'2222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'36!
Results of the interviews'222222222222222222222222222222222222222222222222222222222222222222222222222222222222'36! The theory of planned behaviour!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!36! The elaboration likelihood model!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!36! Herding-behaviour!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!37! The platform!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!37!
Results of the survey'2222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'38! Descriptive!statistics!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!38!
The!theory!of!planned!behaviour!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!38! The!elaboration!likelihood!model!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!39! Herding!behaviour!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!39! The!platform!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!39!
Correlation between dimensions!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!41! One-sample T-test!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!42!
Comparing findings: Experts vs. investors'2222222222222222222222222222222222222222222222222222222222222'43! The theory of planned behaviour!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!43! The elaboration likelihood model!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!44! Herding-behaviour!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!45! The platform!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!45!
6.'Discussion'22222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'47!
Propositions and model validation'22222222222222222222222222222222222222222222222222222222222222222222222'47!
Discussing theory with empirical findings'22222222222222222222222222222222222222222222222222222222222222'49! Herding!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!49! The!Elaboration!Likelihood!Model!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!51! The!Theory!of!Planned!Behaviour!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!53! The!WEB!&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&!55! Academic implications'22222222222222222222222222222222222222222222222222222222222222222222222222222222222222'56! Practical implications'222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'57!
Limitations and future research'22222222222222222222222222222222222222222222222222222222222222222222222222'59!
7.'Conclusion'22222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'61! 8.'References'22222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'63! 9.'Appendix'2222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'70!
Interview questions (translated from Dutch)'22222222222222222222222222222222222222222222222222222222222'70!
Survey questions and answer possibilities (translated from Dutch)'2222222222222222222222222222222'71!
General statistics'222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222'74!
Abstract
The use of crowdfunding to find funding for a project or company has increased rapidly in popularity over the last years. An important concept in crowdfunding is trust. The crowd has to trust an entrepreneur to life up to what is promised. This thesis has put focus on the concept of trust in a specific form of crowdfunding, lending-based crowdfunding in the Netherlands.
In this thesis a conceptual model was constructed existing of four dimensions. The theory of planned behaviour (TPB), a psychological theory on trust, was used to construct the first dimension. The second dimension was constructed around another psychological theory, the elaboration likelihood model (ELM), a dual-process theory about persuasion. The third dimension was build using the psychological construct of herding (HERD), a behaviour-bias that explains why people (blindly) follow the crowd. The last dimension existed of recent findings about trust in e-commerce and the web (WEB).
In this research a mixed research design is used. With the use of semi-structured interviews an exploration of the concept and a first validation check of the model were done. Through these interviews the model was adjusted to encompass more aspects, which resulted in adding the fourth dimension. Subsequently surveys were used to get quantitative data providing a more clear and reliable validation of the model and its sub-variables.
All dimensions were validated in their predictive ability of trust in crowdfunding. The fourth dimension, the WEB, had the biggest predictive ability. Herding was able to explain the least amount of trust compared to the other dimensions.
Findings in this research contribute to the, so far, scarce research in the academic field of crowdfunding and trust and provide valuable new insights. On a more practical note this research is of value to platforms, helping them improve their competitive position; entrepreneurs, showing what aspects to emphasize on when searching for funding; and governments, providing guidelines on how to cope with/build legislation around this new concept.
1. Introduction
An introduction to lending-based crowdfunding
Most of the time, new innovative business models or new inventions develop from people experiencing a problem and coming up with a solution. An infamous English saying, “necessity is the mother of invention”, describes this phenomenon best. Crowdfunding arose from the necessity for the project starter to ask for financing as well as for the investor to invest their money. Crowdfunding is a new source of financing for your project or start-up by tapping the crowd instead of specialized investors (Belleflamme, Lambert & Schwienbacher, 2014).
When starting your own company, getting the investment you need has always been a challenge. This challenge in gaining finance for your company fluctuates in its level of difficulty. An example is the dotcom bubble in the late 90’s; a time in which it was fairly easy for Internet start-ups to find financing, and the years after 2000, when generating funds became much harder (Wheale & Amin, 2003). The start of the economic crisis in 2008, including several big banks going bankrupt, made the remaining banks far more cautious with their business loans (Goodman, 2008). This burst of the dotcom bubble in 2000 followed eight years later, when the start of a big economic crisis made the challenge of getting finance to start a new business very hard, and for some start-ups even impossible. This is where the necessity for the entrepreneurs comes in. When the existing ways of getting finance where not sufficient enough, people created a new way: crowdfunding. Though the necessity for entrepreneurs to use crowdfunding is visible, the concept of crowdfunding on the other hand relied on having enough people to invest.
Investing in start-ups has always been an activity reserved for, among others, banks, venture capitalists and angel investors, (Mason & Stark, 2004) who can be classified as external sources. Beside the external sources, there are also internal sources that can be seen as investors, such as the entrepreneur with his savings, the friends and the family (Manolova,
Manev, Carter & Gyoshev, 2006). The obvious reason for this is that, with exception of the internal sources, these parties have the money to invest and possibility to take risk. The “average” person in society needs their money to pay for life’s expenses and sets the remaining money aside on a savings account, with a good interest rate and risk being minimal to almost none. The economic crisis of 2008 influenced the “average” person as well, in two important ways, making the newly designed concept of crowdfunding, especially lending-based crowdfunding, an attractive opportunity. A first way is an obvious conclusion that can be drawn; the risk people have when saving money, or at least perceive to have. Where people first felt safe saving their money on a bank, this trust in the banks has decreased significantly, and years after the crisis banks and financial institutions are still the least trusted sectors in the global economy (Denning, 2013). Among other reasons, the bankruptcy of some well-known banks during the crisis is at the base of this decrease in trust. This increase in risk (or at least people's perception of this risk) is the first incentive for people to look for alternative ways of investing or saving their money, instead of putting ait all on a savings account in the bank. The second incentive is related to the financial return people get on their money. Where savings accounts use to give a good interest rate, the economic crisis has had its influence on this as well. The interest rate on savings in the Netherlands diminished from little over 5% before the crisis (consumentenbond, 2015) to below 1% in 2015 (Lalkens, 2015). The advantages that saving gave people, being the acceptable interest rate and no risk, as shown, have been influenced in a negative way.
The acceptance and popularity of lending-based crowdfunding can thus be explained from two perspectives. Entrepreneurs who have a hard time finding financing for their projects see crowdfunding as a welcome alternative. When the bank rejects their request, they have crowdfunding as a last resort. Talk in the corridors at crowdfunding events even implied that, now they got comfortable with crowdfunding, some entrepreneurs do not even go to the
bank anymore and see crowdfunding as a first resource. The second perspective is that investors lost part of their trust in financial institutions and saw their interest rate decline over the last years to an ultimate low of less than 1%.
Problem definition
In this research we will focus on lending-based crowdfunding in the Netherlands. The biggest platform in lending-based crowdfunding is the oldest one in the Netherlands, namely geldvoorelkaar.nl (Hupkens, 2014), which started in 2011. Two other platforms, both approximately one year old, are the second and third biggest in the Netherlands: Kapitaalopmaat.nl and Collincrowdfund.nl. This research focuses on Dutch lending-based crowdfunding platforms because all three aforementioned platforms use a strategy that is not used by the big international crowdfunding platforms like Kickstarter or Indiegogo. All three platforms perform a risk-analysis before allowing a project to be posted on their platform. Since we are looking at trust in projects, the best way to analyse this is to use platforms that are in most ways comparable to each other. All this leads to the research question:
• What enhances trust in the investor in lending-based crowdfunding in the
Netherlands?
To answer this question two perspectives of lending-based crowdfunding are examined, which leads to two sub questions:
• What aspects of the project enhance trust in the investor? • What aspects of the platform enhance trust in the investor?
Academic relevance
Crowdfunding is a relatively young concept and therefore has not received extensive academic attention yet. An example of the lack of, and still growing research in crowdfunding is shown when academic search engines are used. When searching for “crowdfunding” a lot
of the papers that are found are working papers, proving the lack of published research. The research that is published focused primarily on arguably the most famous crowdfunding sites of the moment, 'Indiegogo' and 'Kickstarter'.
Indiegogo has two models, the 'keep-it-all' model, where the project starter gets to keep all the money regardless of whether the goal is reached, and 'all-or-nothing' model, where the money can only be kept if the goal is reached. The effectiveness of these models to find enough backers has shown to be dependent on the scalability and capital required. When the project was non-scalable and the capital required was high, an all-or-nothing model was found most effective. Scalable projects, with low capital required, were more effective when the 'keep-it-all' model was used (Cumming, Leboeuf & Schwienbacher, 2014). A pattern in funding behaviour of backers was found in different researches, this time focused on 'Kickstarter' projects. A U-shaped pattern was discovered where the funding started off with a boost, then dropped down, and at the end of a projects period an increase of backers was shown (Kuppuswami and Bayus, 2013). Though these researches provided valuable insights, they were also done using the two big crowdfunding platforms that are both reward-based.
Though the term crowdfunding has not been used that long, peer-to-peer lending, which can arguably be seen as a form of crowdfunding, has been around longer. Research in this field has given us some valuable insights. Research on the peer-to-peer lending platform 'Prosper' has shown that social capital, in other words having a social network and connections with other members, has had a positive influence on several aspects of peer-to-peer lending, such as the interest rate and likelihood of getting funded (Greiner & Wang, 2009; Lin, Prabhala & Viswanathan, 2013). Furthermore, Gerber and Hui (2013) built a framework describing the motivations and deterrents for project starters as well as for projects backers to use crowdfunding.
Even though all these researches provide insight in crowdfunding, an important issue has not yet been examined. The concept of trust has played an important role in many academic fields, such as philosophy, psychology, economics, management and marketing. With the coming of the Internet, this trust extended to include online trust (Wang & Emurian, 2005). This online trust is where this paper will put its focus on; the online trust that is generated with crowdfunding. By determining what aspects in crowdfunding enhance trust, this paper will help build a better understanding of the dynamics of lending-based crowdfunding and its participants.
Social relevance
This research has put the focus on a specific form of lending-based crowdfunding, a form in which a risk analysis is performed by the platform. The three biggest lending-based crowdfunding platforms in the Netherlands all apply this strategy. Since we are examining trust in lending-based crowdfunding, it is important to examine how much trust investors gain from the risk analysis and the platform, and how much they gain from project characteristics.
The first aspect of the created trust that is examined is from the characteristics of the project's point of view. This research focuses on examining several aspects of projects and investigates to what extent they enhance trust. When the aspects of a project that enhance trust are established, it might help project starters in deciding their strategy for publishing their project. It can provide project starters with guidelines on how to publish their project in such a way, that it is most efficient in finding sufficient funds. Questions related to what content or graphics to add to a project can be answered from the perspective on what enhances trust in the investor.
The second aspect of trust is focused on the platform. Knowing how many trust investors gain from the platform they are investing in can be helpful in two ways. New, as
well as established platforms can learn from these results and add to, and/or put emphasis on the characteristics of the platform that enhance trust.
Thirdly, the influence of the platform might be important regarding the legislation around lending-based crowdfunding. Around a year ago, the first lending-based crowdfunding project declared bankruptcy and could not pay off its debts to its investors, including the investors through crowdfunding (Bueters, 2014). The platform on which the project was funded said they wanted to mediate between their investors and the bankrupt company, but that was it. At this moment the platform only has a facilitating role and has no responsibility. If this research finds that a big portion of the people's trust is enhanced by the risk analysis, this might be reason to rethink the responsibility a platform has. To what extend is a platform responsible when a project goes bankrupt and cannot pay back the investors on the platform?
Structure of the thesis
Following this introduction to the subject, this thesis will discuss several theories and findings related to trust. This part covers two known psychological theories and a psychological construct that can be related to trust and discusses how this might apply to crowdfunding. Also, several findings on trust in e-commerce and the Internet, and how they might relate to crowdfunding, are discussed. In the third section of this paper a conceptual model is developed based on the findings discussed in the literature review in section two. For every dimension in the model several propositions are given. The next section will focus on the methodology used in this research. The chapter will start of with explaining why a mixed method was used, and the advantages it has for this research. Furthermore, an elaboration is given on how the qualitative and quantitative research has been conducted. Section five will outline the most important findings of the research. Discussing and elaborating on these findings will be done in section six of the research. The limitations of the research will also be
addressed here, and suggestions for future research are given. The research will end with a short summary of the most important findings.
2. Literature review
According to the American Psychological Association (APA), psychology is the study of the mind and behaviour. The field of psychology is broad, including the field of psychology and trust. Among other things, research has focused on inter-organizational and interpersonal trust (Zaheer, McEvily & Perrone, 1998), personality characteristics and trust (Mooradian, Renzl & Matzler, 2006), and trust in Internet applications (Bierhoff & Vornefeld, 2004). This broad area of research on trust does include the difficulty of finding a definition of trust that covers all aspects. This paper will focus on trust in online crowdfunding from the investors’ perspective. Rotter (1967) defined trust as “an expectancy held by individuals or groups that the word, promise, verbal, or written statement of another can be relied on” (p.651). This definition of trust will be used in this paper, for it can be applied to crowdfunding, where people trust a project creator to live up to his promise of producing a product and providing the investors with the associated award. Using two well-known psychological theories, one psychological construct, backed up by some more recent findings about trust in e-commerce and the web, a model is built to analyse trust in lending-based crowdfunding.
The theory of planned behaviour
A well-known theory in psychology that can be linked to trust is the Theory of Planned Behaviour (TPB) (Azjen, 1991). This theory is an extension of the Theory of Reasoned Action (TRA), proposed by Fishbein and Azjen (1975, cited by Azjen, 1991). The TRA states that the planned behaviour of someone can be predicted by the intention someone has to deliver that behaviour. The intention (behavioural intention, BI) in this theory is based on two things. Firstly, the attitude (A), which comes down to what consequences the person thinks the behaviour has, and how he values those consequences. The second part of someone's intention is about the subjective norm (SN) related to the behaviour. This can be translated to how other people or groups value the behaviour in combination with how much
one wants to comply with these people or group expectations. Thus, the model for this behavioural intention, comes down to BI=A+SN. The TPB is an extension of the TRA and has an added variable: the perceived behavioural control (PBC). The PBC is the level of which a person is seen as in control of the needed behaviour. When the PBC is high, it is a good reflection of the actual behavioural control, and therefore a good predictor. When the PBC is low its predictive value declines. Summarize the TP consists of the variables 'intention' and the 'perceived behavioural control' to predict a person's planned behaviour, or in short TPB=BI+PBC.
This paper only uses the variables 'intention' and PBC and not the sub-variables of intention, being 'attitude' and 'subjective norm'. Because the model already covers a broad area of dimensions, using the sub-variables would not add enough extra value and mainly has the chance of over-complicating the model.
Both dimensions of TPB relate to crowdfunding. Firstly, the intention can be found in what the project starter needs the money for. What are the plans the project starter is trying to realize? The second variable of TPB, the PBC, comes down to how the investors think the project starter is in control of the project and its outcomes. Since prior successful experience in a market can be seen as an indicator of how a person is in control of his behaviour in that market, this will be the variable that is analysed in crowdfunding, a project starter’s experience in the respective field of the project.
TPB is seen as fundamental in psychological research and has been used to explain and predict large varieties of behaviours in numerous studies (Armitage & Connor, 2001). Though the TPB is a well-known and validated theory, more recent research has also shown that the TPB on its own might be insufficient. Wang (2009) used the TPB, but added attitude functions. Park, Klein, Smith and Martel (2009) used TPB in combination with levels of perceived norms, and Wang and McClung (2011) combined TPB with both the theories of
attitude functions and perceived norms. Combining theories with other theories or adding levels can help do more accurate research. The TPB can help to explain and predict part of the effect that enhances trust in investors in crowdfunding, but leaves a lot of factors used in crowdfunding unexamined. Following the example of Wang and McClung this research will use the TPB in combination with one other psychological theory, a psychological construct, and findings of research in e-commerce and the web to get to a better model for predicting trust in crowdfunding.
The elaboration likelihood model of persuasion
Trust can be seen from many psychological perspectives. This is why multiple psychological theories can be related to trust. Besides the TPB, which explains trust from the perspective of what a person expects another person to do, and how it is done, there is also the Elaboration Likelihood Model of persuasion (ELM) (Petty & Cacioppo, 1986). This model can be linked to trust from another perspective compared to TPB. ELM is a dual-process theory that provides a general framework for understanding the basic process underlying the effectiveness of persuasive communications. It elaborates on how attitudes of people form and change through the processing of information. Petty and Cacioppo conclude that there are two main routes to persuasion. The central route of information processing drives on logical, issue-relevant arguments and thoughtful consideration. This way of processing takes up more time and cognitive effort. The second way of processing information is through the peripheral route. This route shapes the attitudes of people by associating certain attitudes to certain heuristic cues. Examples of heuristic cues are the attractiveness or credibility of an attitude or object. Petty and Cacioppo define attitude as “General evaluations people hold in regard to themselves, other people, objects, and issues” (p.127).
In crowdfunding, people try to persuade others to fund their projects. They attempt to create trust by use of persuasion. For this reason it is important to include the ELM in our model. Greiner and Wang (2010) used the ELM in their research to trust in online peer-to-peer (P2P) lending. They examined two websites in P2P lending, which can, in a way, be seen as lending-based crowdfunding sites. They distinguished several aspects of online P2P lending that either fell in the category of the central route or in the category of the peripheral route. For the central route, they took into account the credit grade a borrower has, the debt-to-income ratio, the verification of a bank account, home ownership, and having a previously successful loan. The peripheral route in their research existed of a group rating, the group leader reward rate, endorsements from other members, the length of a description and having an image. In their research, they found support for effects on trust in loans from all their predicted factors, except homeownership group and leader reward. This was based on the dependent variables ‘percentage funded of a project’ and ‘the spread above prime rate’, the latter being defined as the difference of borrower rate and Wall Street Journal Prime Rate.
Though ELM is already used in P2P lending, which is somewhat similar to crowdfunding, most of the categories used are not found on the crowdfunding websites this research examines. Looking at the variables used, only length of description and having an image are applicable in crowdfunding. Greiner and Wang's research is still useful though, for it shows that the ELM has a significant value to add in a model that can be used to show what enhances trust in a crowdfunding project. Through this research it is attempted to establish a model based on academic research. The ELM will be added to categorize the aspects in lending-based crowdfunding that either are heuristic cues or is information that requires thoughtful consideration. For the peripheral route this includes having an image and the length of the description but also the risk rating of a project. For the central route the solid facts, such as liquidity and solvability that are stated in crowdfunding projects will be used.
Informational cascading and the herding-behaviour bias
Dijksterhuis and Bargh (2001) wrote about how social perception of events can affect social behaviour. An example they use is that one of the authors was pulled over by the police for driving to fast. When asked if he had watched Formula One Grand Prix the author answered yes, to his own surprise. Apparently watching someone else’s behaviour had influenced his behaviour. Dijksterhuis and Bargh called this the perception-behaviour link. Simply put, it all comes down to imitation of behaviour of others. Though the perception of behaviour influences your behaviour, imitation of someone else's behaviour can influence their perception of you too. This copying of other's behaviour, mimicry as Lakin and Chartrand (2003) call it, is used to create affiliation with the other person, and can, according to them, be linked to the perception-behaviour link. To sum up, people want to affiliate with other people, and that’s why they imitate them. By imitating them they create a feeling of affiliation with the other person.
Though this mimicry and perception-behaviour link is based on visual perceptions, this link has also been found in online behaviour. Huang and Chen (2006) found that people have a tendency to follow other people when shopping online, guiding them in their product choice. They state that there are two social influences that lead people to imitate others, normative and informational influence. Normative influence is about imitating people to conform to the expectations of others. Informational influence is about accepting information received from others as an indicator of reality. They expect informational influence to be of more effect in online product choice. This is also based on the psychological construct of informational cascading, which is an effect that occurs when people disregard their own information and use information of people who preceded them to value the efficacy of that behaviour (Dholakia & Soltysinski, 2001). Other people's behaviour, such as which choice to make, how much to bid or how much to invest, provides an informational signal to a person. This informational cascading is a form of herding behaviour where people follow the crowd
without considering own information. Herding behaviour is seen in a lot of human activities such as auction biddings, but also in stock markets (Hirshleifer & Hong Teoh, 2003) and product choice (Huang & Chen, 2003). Though herding behaviour can have groups of people make irrational and not always the most efficient choices, which in its turn may harm stocks or stock markets, it can also be seen as a positive thing. Because “previous research has shown that people imitate others out of a desire not only to be accepted but also to be safe” (Huang & Chen, 2003, p. 414) herding behaviour might also be used as an advantage. People want to feel safe in the choices they make. Huang and Chen found that sales volume and customer reviews influence customer’s product choice where higher sales volume and positive reviews lead to a stronger tendency to choose that product. This might also apply for crowdfunding, for safety and lowering risk are important when investing.
Berkovich (2011) did research into herding in the P2P online lending market. The research used a model designed by Berkovich and Tayon (2009, cited by Berkovich, 2011) that was based on a large market with incomplete information and publicly visible actions to assess the herding behaviour. He found that loans with more bids where more valuable to bidders. The average size bid of other lenders was not used as an indication to the amount to bid. Shen, Krumme and Lippman (2010) also found a herding effect in lenders on a P2P lending platform. Both researches have been done on the US lending platform 'Prosper'. Herding is a social effect and the social effects Shen et al. state in their research on Prosper might not line up with the social effects found in crowdfunding. In Prosper there is a social effect of having a friendship on the platform, joining or creating a group and getting endorsements from others. These components are not used in the crowdfunding platforms that are analysed in this paper. Still, a herding effect is expected to be of influence in crowdfunding too its influence coming from other aspects. For this reason it is added to the
model that is designed and validated in this paper. Aspects of crowdfunding that can fall under this category are the amounts of bids and the comments of other investors.
E-commerce and websites
The psychological theories that are about trust or related to trust discussed earlier mainly put the focus on the trust in the entrepreneur or project starter and other investors on the crowdfunding platform. A project starter has to persuade the investors into investing in the project and has to show that promises made can be kept. Furthermore, the investors are expected to follow other investors, according to the theory. By interviewing several experts on lending-based crowdfunding in the Netherlands one important variable came up was not included in the model before. According to several experts a big part of the trust in a project arises from the platform that is used. To validate this view of the experts it is decided to add the extra variable of the platform, including sub-variables, to the model. In the next section it is tried to establish important factors of trust in websites and e-commerce that are expected to be applicable to crowdfunding too.
In lending-based crowdfunding in the Netherlands a risk-analysis of the project by the platform is an often seen aspect. This risk-analysis is based on the financials and the prospections of the project and entrepreneur. After carefully examining a project the platform provides a number or letter that indicates what level of risk of the project has. Ability of a platform to perform well and live up to what they promise is an important factor for trust in E-commerce (Gefen, 2002; Ang, Dubelaar & Lee, 2001; Chen & Dhillon, 2003). Gefen defined ability as “having the appropriate skills and competence” (p. 40), this definition of ability is applicable to crowdfunding as well. The platform is expected to have the right skills and competence to perform a good and trustworthy risk-analysis. Ability, therefore, will be added to the dimension of trust in the platform.
Lending-based crowdfunding platforms in the Netherlands promote themselves with the success rate they have on funding projects. Almost all platforms score a success rate of around or exactly a 100%. This high percentage is due to two factors. First it’s because of the risk-analysis that is done beforehand. With this analysis the gross of the projects that are submitted for crowdfunding are cut out, keeping only the good ones. Secondly, the success rate is based on the amount of crowdfunding projects that have met their funding goal. Though the word success rate might imply a success of the company, it is merely a success of the funding. This success rate is not related to the success of a company but investors might see it as a success of the platform. The amount of companies successfully funded, in a way, can be linked to the reputation of a platform. Reputation has been shown to be important in enhancing trust in online websites (Koufaris & Hamption-Sosa, 2004; Casaló, Flavián & Guinalíu, 2007). Even when they used a hypothetical website, offering legal advice and manipulated the level of reputation, reputation was found to be a strong indicator of trust (McKnight, Choudhury & Kacmar, 2002).
Reputation is based on several aspects like social media (Tang, Gu & Whinston, 2012), media (Kotha, Rajgopal & Rindova, 2001) and reputation systems (Jøsang, Ismail & Boyd, 2007). Turner, Zavod and Yurcik found that feelings of security and privacy are also influenced by reputation, where a lower feeling of security or privacy negatively affects the reputation. It might be argued that in crowdfunding the feeling of security and privacy is of more importance then it is in e-commerce, for people invest significant amounts of money and not uncommonly more than one time. Furthermore, research in e-commerce has shown that privacy and security both influence the trust consumers have in a platform or company (Suh & Han, 2003; Koufaris & Hamption-Sosa, 2004; Ang, Dubelaar & Lee, 2001). For this reason a separate sub-variable is established for privacy and security.
Finally to have a complete model it is important to include variables that say something about the intention of the investor and how they can achieve this. The technology acceptance model (TAM) by Davis (1989) has two important variables that are used by users of new technology. Davis defines the first variable, the perceived usefulness of a new technology, as “The degree to which a person believes that using a particular system would enhance his or her job performance” (p. 320). In the case of crowdfunding, the job of people can be seen as investing, and thus the question is if the people see crowdfunding as a way to perform better in investing. Performing better can be linked to getting a better interest rate and diversifying more. The second variable is the ease of use of technology, defined by Davis as, “the degree to which a person believes that using a particular system would be free of effort” (p.320). In crowdfunding this translates to the ease of use of the website for the investor to invest. The TAM has been used in explaining and examining trust issues in several online fields, including online tax services (Wu & Chen, 2005), repeat online shoppers (Gefen, Karahanna & Straub, 2003), and trust by new customers in online companies (Koufaris & Hamption-Sosa, 2004).
3. The conceptual model on crowdfunding and propositions
The conceptual model of trust in crowdfunding is built on a combination of two perspectives. The first three dimensions are used to examine the trust in a project starter and the last dimension to examine trust in a platform. The first three parts of the model will be based on two psychological theories, the Theory of Planned Behaviour and the Elaboration Likelihood Model, and the psychological construct of herding behaviour. The last part of the model is based on recent findings in e-commerce and the web. The sub-variables here serve to find the overall effect of the platform on trust (fig.1).
Fig 1. Conceptual model
The Theory of Planned Behaviour (TPB) part of the model is expected to explain part of the trust investors have in projects and exists of two sub-variables, the intention and the perceived behavioural control (PBC). The intention is seen as to what extend an investor
looks at and gains trust from the intention a project starter is claiming to have with the investment asked and the market the project operates in. The PBC is seen as the experience a project starter has in the respective field of the project.
Proposition 1: The theory of planned behaviour can explain trust investors have in projects.
Proposition 2.a: Investors gain trust from the intention a project starter has with the project. Proposition 2.b: Investors gain trust from the market a project operates in.
Proposition 3: Investors gain trust from the experience a project starter has in the respective field of the project with more experience resulting in more trust.
The Elaboration Likelihood Model (ELM) is expected to explain trust by analysing how people are persuaded, using two cognitive information-processing routes, the peripheral route and the central route. The peripheral route uses heuristic and subjective cues such as how attractive or good something is, this is mostly done subconsciously. In our model the peripheral route is analysed using six propositions considering the length of a description of the project, length of the description of the project starter, the presence of a picture of the entrepreneur, a logo or a video and the risk rating the website has given the project.
Proposition 4: The Elaboration Likelihood Model can explain trust investors have in projects.
Proposition 5.a: The longer the description of the project the more trust investors have in the project.
Proposition 5.b: The longer the description of the project starter the more trust investors have in the project.
Proposition 5.c: The presence of a picture of the entrepreneur enhances trust in the investors. Proposition 5.d: The presence of a logo enhances trust in investors.
Proposition 5.e: The presence of a video enhances trust in the investors.
Proposition 5.f: The more positive the risk rating, the more trust investors have in the project.
The central route uses a lot more of the cognitive workload and is about looking at solid arguments and facts. When using the central route people are actively weighing their options, calculating risks and are consciously involved. To analyse the central route from the ELM in our model, four propositions are used considering the solvability, profitability, liquidity and the own money invested by the project starter.
Proposition 6.a: The better the solvability the more trust an investor has in the project Proposition 6.b: The better the profitability the more trust an investor has in the project Proposition 6.c: The better the liquidity the more trust an investor has in the project
Proposition 6.d: Investors have more trust in projects where a project starter invested a significant part of their own money relative to the funding goal.
The psychological construct of herding behaviour is expected to explain trust through people relying on other people’s choices to make their own choice. In the extreme cases people can make irrational choices, disregarding their own preferences and using information and choices of preceding others to value an object or behaviour. In the case of crowdfunding
this can be linked to the amount of other bids at the time of investing, the amount of comments and the content of the comments. There are the three propositions used to validate the herding dimension of the model.
Proposition 7: The Herding behaviour bias can explain trust investors have in projects.
Proposition 8.a: The more bids a project has, the more trust investors have in the project. Proposition 8.b: The more comments a project has, the more trust investors have in the project.
Proposition 8.c: The better the quality of the content of the comments, the more trust investors have in the project.
Recent findings in the field of e-commerce and the Internet are expected to be of influence in trust from investors in projects as well. The e-commerce and website variables used to build propositions to validate this dimension are the ability and reputation of a platform, the percentage of successful projects, the privacy & safety regulations, and the usefulness & ease of use of the website.
Proposition 9: Research in the field of e-commerce and the Internet can explain trust investors have in projects.
Proposition 10: The more competent the people behind the platform are in performing a risk analysis the more trust an investor has in projects on the platform
Proposition 11: Investors gain trust from the reputation of a platform, with a better reputation resulting in more trust.
Proposition 12.a: Investors have more trust in projects of a platform when they feel their privacy is ensured.
Proposition 12.b: Investors have more trust in projects when they feel their money on their account and the money they invest is secured.
Proposition 13.a: Investors have more trust lending-based crowdfunding platforms compared to other forms of investing.
Proposition 13.b: Investors have more trust in lending-based crowdfunding platforms when they are easy to use.
Using these propositions it is tried to find validation for the applied conceptual model of crowdfunding (fig.2). This model is similar to the conceptual model only with the sub-variables defined based on the three major Dutch lending-based crowdfunding platforms.
4. Methodology
This paper tries to build, and find validation for a new conceptual model. The research methodology linked to this could be best classified as grounded theory research. Grounded theory is defined as “a general methodology for developing theory that is grounded in data systematically gathered and analysed” (Straus & Corbin, 1994, p.273). Straus and Corbin also state generated data can be used, but existing theories that seem appropriate, whether or not modified, can also be used. In this paper it is decided to use both. Existing psychology theories and generated data from interviews, substantiated with earlier research are used to build the model. Subsequently, surveys are used to find validation for the model.
The mixed research approach
Lending-based crowdfunding is a field of research that is relatively young. The first Dutch lending-based crowdfunding platform, geldvoorelkaar.nl, was founded in 2011, indicating that the research covering this field is still relatively small. Because this research examines a new phenomenon it can be classified as exploratory research. In this research the goal is to build and validate a new framework by using a mixed research approach.
When conducting exploratory research, mixed research is a common method to use. One of the strategies applicable is the sequential procedure (Creswell, 2013). This paper made use of this procedure by starting with interviews to explore the subject and the dimensions of the model, and, if needed, alter the model. As Thomas (2006) finds, the inductive approach is not that strong as some other strategies when developing a model. After the interviews the conceptual model was altered and surveys were used to validate the newly altered model and make it more generalizable to the population of investors.
The mixed method, or methodological triangulation, using the sequential procedure can be put to use in two ways. The qualitative part can precede the quantitative part and vice versa. Morse (1991) states that one of the characteristics indicating the problem is qualitative
is the “immaturity” of the field or concept. Given the lack of research and theories about lending-based crowdfunding, this is indeed the case. An inductive process drives qualitative problems, and according to Morse the qualitative method should, in that case, precede the quantitative one. Though it could be argued that an a priori technical framework drives the research, and Morse states that this is then classified as a deductive process; the framework was to guide the interviews. The final framework that is tried to validate in this paper was built and altered after the qualitative part.
When conducting and setting up both the qualitative as well as the quantitative research, high ethical standards were pursued. Participants were ensured complete anonymity and were debriefed at the end of the interview or survey. The survey was checked by several independent sources to make sure questions and format were clear, understandable and not too demanding.
Semi-structured interviews
Through literature research focusing on trust and psychology, and by examining several Dutch crowdfunding platforms, the first three dimensions of the model were built and altered to the extend that the focus of the sub-variables could be set on crowdfunding variables. A semi-structured interview (see appendix) was set up using this newly build model. Because the subject of trust in lending-based crowdfunding is relatively new it was made sure that there was room in the interview to have the interviewee give own input. Also the structure of the interview was not strictly followed, it was made sure the key questions were answered but besides that the interview was open.
All semi-structured interviews were audio-recorded to ensure identical replication of the content of each interview. Audio recording is frequently the method of choice when conducting semi-structured interviews, for it “provides a detailed insight into the performance
of both the respondent and the interviewer” (Barriball & While, 1994, p.332). Furthermore it was ensured to ask the general questions first and then the specific question, asking it the other way around would indicate more importance to the specific question (Hutchinson & Wilson, 1992). To ensure this, the interview guide designed has underlined the general questions that are most important to the research. The questions following were designed to ask only when necessary and time allowed it. Most of the time the interviewee answered the specific questions in the answer to the general question.
In total, five interviews were conducted in the Netherlands with duration varying between 30 to 50 minutes. The participants were found through the supervisor’s network and the use of snowball sampling. In snowball sampling the natural network of people is used (Noy, 2008). In this case this included asking a person who was interviewed if there were people in his/her network that would be willing to be interviewed as well. The interviewee’s were all Dutch and either working in lending-based crowdfunding or were experts on the subject. Though it was tried to get views from different social sources, having a completely generalizable group of participants in the interviews is difficult. This is due to the fact that the selection, most of the time, is done on the goodwill and selection of the powerful within an organisation (Diefenbach, 2009). They select the interviewees that get the opportunity to share their view. Unfortunately this was also the case in this research, all CEO’s of lending-based platforms had an employee do the interview. In our case this is not a significant problem, for Diefenbach also states that this problem is most apparent when only qualitative research is used to develop a theory. This paper uses quantitative research to validate the model, hence making earlier mentioned bias of less importance. Two interviews were held face-to-face but due to geographic location of some interviewees the remaining three were done using the video-call system 'Skype'. All interviewees were notified that the interview
would be recorded. Furthermore, the scope of the research and use of the data was explained. Lastly, anonymity was guaranteed.
Data analysis of the interviews was done by first transcribing all interviews. Secondly the raw data, the transcriptions, were analysed and the key themes and categories that overlap all interviews were established. Categorizing raw qualitative data is necessary when analysing qualitative data (Thomas, 2006). In our case it was either categorized in one of the existing three dimensions of the model or categorized in “outside the model”. After this, the data categorized as “outside the model” was put into four different categories, all having one overlapping category, namely “the platform”.
Online questionnaires
After having completed the conceptual model based on research and input from the interviews, the relationships between the variables and trust was measured. These relationships were measured using an online questionnaire. According to Lumsden (2007) an online questionnaire has multiple advantages compared to a paper-based questionnaire, with the most important advantages for this research being the cost, speed, appearance and usability. The questionnaire was designed to be short and focused, taking about 5 minutes, to make the threshold to enter as low as possible for participants. The questionnaire was pre-tested by several graduate students and a university professor to ensure quality as well as usability. One item was scored contra-indicative and for that reason was rescored. Using a descriptive analysis, the main descriptive were found. A reliability analysis was done to ensure the reliability of all items. Through use of a correlation matrix the correlations between the dimensions were examined. Lastly, two one sample T-tests were done to see if the found influence of trust is significant.
Data collection:
Participants were found using social media sources, namely LinkedIn, Twitter and Facebook. Through use of the own personal network several experts and coaches in crowdfunding, and a CEO of a lending-based crowdfunding network were approached with the request to share the questionnaire with their network. Furthermore, several crowdfunding groups on these social media platforms were used to spread the questionnaire more widely. Lastly, the questionnaire was posted in a newsletter of one of the large crowdfunding platforms in the Netherlands.
Data analysis:
Through a survey among investors on crowdfunding platforms (n=59) the 4 dimensions and its sub-variables were examined in their relation to the concept of trust. There were some missing values that were replaced with the series means of the corresponding item. The reliability of the TPB was too low, and for this reason the recoded item TPB_Starter2 (TPB_Starterrec) was deleted. Deleting this item increased the reliability to a level that is, according to Gliem and Gliem (2003), questionable but near to acceptable (α= .668). The Cronbach’s Alpha’s of ELM (α= .759) and the platform (α= .703) are over .7, which classifies the reliability as acceptable. Reliability of herding (α= .818) is over .8, which is classified as good (see fig 5.). Most participants were male, had followed a higher level of education, have had earlier experience with investing and were multiple investors on several platforms.
A correlation matrix is used to examine correlations between dimensions. To see if the dimensions were correlated, several assumptions have to be met. First the data was checked for outliers which, when found, were removed from the analysis. A second assumption is the normal distribution of the data. Using the Shapiro-Wilk test, normal distribution of the data of all four dimensions was confirmed (p > .05). Using a scatter plot between two dimensions has shown a linear relationship, which is a third assumption. The last assumption is that the data has to be scored on interval or ratio level. The survey was done using ordinal levels, but the
data from the dimensions, which includes all data from the sub-variables, is scored at interval level, thus the assumption is met.
A one-sample t-test is used to see if the mean scores of the dimensions differ significantly from a pre-set hypothesized value. All questions were scored on a 7-likert scale (1=completely disagree, 2=fairly disagree, 3=partly disagree, 4=do not disagree/do not agree, 5=partly agree, 6=fairly agree, 7=completely agree). The propositions regarding the dimensions state that each dimensions explains trust investors have in a project. The value 4, where people do not agree but also do not disagree, can be seen as explaining no trust. Our proposition is therefore based on the pre-set hypothesized value of 5 and higher, which indicate that the investor at least partly agrees that the dimensions and its sub-variables explain some trust. Therefore, using the one-sample t-test to see if the dimensions can explain trust significantly, the pre-set value of 4 as well as 5 are used to compare with the dimension means. The assumptions that have to be met are the same as when using a correlation matrix. The only exception is the assumption that the data is independent, meaning there is no relationship between observations. Since this data is built on objective investors that have no relationship with each other, this assumption is also met.
5. Results
Results of the interviews
All interviews were coded using a qualitative method developed by Strauss and Corbin (1998). First the interviews were open-coded, which comes down to transcribing the interviews. Secondly, axial coding was done; the parts of the interview that seemed relevant were set apart and were categorized. In this paper the axial coding is described as the sub-variables that are found. Lastly the selective coding was used to cluster the axial coding, or sub-variables, into one category. In the paper the selective coding is shown as the dimensions established.
The theory of planned behaviour
Both sub-variables of the first dimension, TPB, were acknowledged by all interviewees to be of importance. Beside the fact that several interviewees thought it important for generating trust to have a certain amount of experience in the respective field of the project, one platform did not even allow starters to put their project online. Furthermore, four out of five interviewees specifically mentioned the intention of a project to be of importance for trust too.
The elaboration likelihood model
The interviews were slightly less clear about the influence of the sub-variables associated with the ELM. Most interviewees came to the conclusion that there are different investors who have different views on projects. There were the investors that examined projects, looked at the finances and then made there decision, but this group is declining since projects are reaching their funding goal in a real short space of time, which does not allow careful examination of a project. Most investors at this moment, according to experts, use the peripheral route of information processing when looking for projects to invest in. Looking at the specific coding given to interview quotes most are defined as peripheral. Within this
