Informational cues or content? Examining project funding decisions by crowdfunders

Abstract

We examine how crowdfunder experience affects their reliance on information available on projects. Drawing on elaboration likelihood model and using data from Kickstarter, we apply machine learning techniques and choice modeling to examine the information provided by creators, investigating not only the descriptions, but also the pictures and the videos. We found that more experienced crowdfunders react positively to descriptions exhibiting higher analytical thinking, while less experienced crowdfunders rely more on cues that arouse attention (e.g., number of pictures and positive emotions in videos). We highlight the importance of considering how experience influences crowdfunders’ interpretation of different types of information.

Introduction

Crowdfunding refers to efforts initiated by individuals as open calls, usually through the Internet, to fund initiatives by drawing on small contributions from the crowd [56]. It has become an important fundraising venue for innovative initiatives. Kickstarter, one of America's most prominent reward-based crowdfunding platform, has raised $4.7 billion for more than 170,000 projects as of December 2019. In reward-based crowdfunding, crowdfunders often face severe information asymmetry problems when making decisions, as they face high levels of uncertainty about the projects available for funding. Even though most reward-based crowdfunding platforms adopt an “all-or-nothing” funding mechanism, in which a project that fails to reach its funding goal before the deadline will return all previously pledged funds, project backers still face significant uncertainty. Research has shown that approximately 75% of Kickstarter projects face delivery delays [56]. Most projects on crowdfunding platforms are works in progress, and the true product quality or even their delivery is uncertain [6], owing to doubt on its chances for success, and its project creators’ abilities to realize the project.

Given the uncertainty of the project quality, many aspects of the reward-based crowdfunding platform accentuate the information asymmetry problem. First, a lack of screening mechanisms on most crowdfunding platforms means that the supply of poor-quality projects can be large relative to that of good quality projects. Second, crowdfunders have limited information gathering and monitoring mechanisms at their disposal, given the asynchronous and public nature of interactions with project creators. The project campaign page is the main (or even the only) source of information for a potential crowdfunder. The information exchange takes place online and is confined largely to the campaign page itself. This is in stark contrast to many other fundraising activities, such as venture capital and stock initial public offering, where fundraisers and fund providers can engage in offline activities for further information exchange and disclosure. It is difficult for a potential crowdfunder to ask for further information and clarification. Third, under the current reward-based crowdfunding model, crowdfunders have no visibility of and control over the funds after a successful crowdfunding campaign, making it more important to screen projects during the campaign stage [75]. Finally, crowdfunders do not possess the financial sophistication of angel investors or venture capitalists [2] and have to make decisions within a short funding duration of 30-60 days [18], making them more vulnerable to the uncertainties associated with investing in crowdfunding projects.

Many researchers are thus interested in examining how crowdfunders overcome information asymmetry to make their project funding decisions (e.g., [2, 79]). While prior research has provided many insights, there are several limitations. First, the studies from both the project creator and project backers’ perspectives assume that all backers are homogenously influenced by a set of informational cues or signals about the project. The literature so far has focused mainly on establishing whether a signal or a set of informational cues affects crowdfunders’ project backing decisions. Most studies have focused on the project level of analysis. This does not consider the heterogeneity of project backers [50]. The nature of crowdfunding is that it attracts a wider and more heterogeneous audience than traditional fundraising channels such as banks or venture capital companies. While studies in the crowdfunding domain have generally highlighted the importance of studying crowdfunder differences [3, 82], limited attention has been given to how different crowdfunders interpret and utilize information. The effects of project level information on individual decisions ultimately depend on whether individuals rely on that information, which differ by individual characteristics [34].

To examine how different individuals interpret various types of information when they make project backing decisions, we focus on one of the most important aspects of crowdfunder differences – their experience. The literature has shown that the experience of a decision-maker is a critical variable influencing decision-making in various contexts, including investments [23] and online shopping [88]. Heterogeneity in crowdfunder experience becomes more salient as many crowdfunders have limited experience in backing projects. So far, we still have limited understanding of how experience affects crowdfunders’ decision-making. In one notable exception, Kim and Viswanathan [41] focus on how crowdfunders with different experiences engage in herding behavior and find that inexperienced crowdfunders tend to follow experienced crowdfunders. Our paper builds on their observation of the importance of crowdfunder experience and examines how crowdfunder experience affects their reliance on different types of project information available on a project campaign page. Hence, our research seeks to answer the following research question: How do crowdfunders with different crowdfunding experiences differ in how they process the project information?

Prior research does not adequately examine information from different types of media provided by project creators. Leveraging on the theory of elaboration likelihood model (ELM), we provide a conceptual understanding of the types of creator-provided project information, in the form of both content and cues, which crowdfunders pay attention to when evaluating projects for their backing decisions. Project creators on reward-based crowdfunding sites often provide a project description, together with project video(s) and pictures on the project campaign page. We employ recently developed machine learning techniques, such as facial expression analysis and text analysis to examine project information – across different types of media (including text, images, and videos) to examine project information that can serve as project content and cues. This allows us to generate a more comprehensive understanding of how crowdfunders differing in their crowdfunding experience may pay attention to different types of information.

Past studies that examine crowdfunders’ decisions are largely conducted at the project level, focusing on projects’ funding performance (see Appendix A). In contrast, we collect detailed project backing information, which allows us to examine crowdfunders’ choices at the individual level using the choice model. This not only allows us to study crowdfunders’ decisions directly, but we can also examine how crowdfunders’ experience affect the way they process information and make project backing choices.

In the following sections, we first provide some background of the crowdfunding phenomenon, explaining the different types of crowdfunding. We then discuss the ELM and present our hypotheses. Following that, we describe our empirical investigation, as well as robustness tests. Finally, we discuss limitations of our study, as well as implications for both theory and practice.