Network Effects in Crowdfunding
The advances in FinTech are concomitant with the development of digital platforms, such as Ethereum, LendingClub, or PayPal. The decisions that users make on such platforms are highly interdependent, insofar as these decisions jointly condition the value that users will obtain from interacting on the platform. Hence, users of FinTech platforms care about participation and usage decisions of other users, which amounts to say that network effects are at work. Although observers agree on the crucial importance of network effects for FinTech platforms, there has been so far little systematic evidence of their incidence.
In a recent working paper (available here), we attempt to fill this gap using a crowdfunding platform (referred to hereafter as CFP) as a laboratory for our analysis. CFPs facilitate the interaction between entrepreneurs in need for funding and backers interested in financing projects. Therefore, the success of CFPs strongly depends on how network effects emerge and are managed. Against this background, we pursue two goals. First, we map the web of network effects at work on CFPs. We distinguish between two sources of network effects: increased participation (by additional backers or entrepreneurs who decide to join the platform) and increased usage (mainly by existing backers who decide on which project to contribute). Second, we show evidence identifying the presence of these network effects in crowdfunding and assess their relative intensity. Specifically, we proceed by focusing on backers’ contributions on the universe of projects listed on Ulule, one of the leading reward-based CFPs in Europe. Between 2010 and 2016 (our sample period), Ulule attracted more than 1.3 million of backers on about 24 thousand entrepreneurial projects.
Our data reveal clear evidence of the prevalence of network effects in crowdfunding. We are interested primarily in network effects that arise within the group of backers. The basic question here is: How do backers affect one another through the decisions they make? In this regard, we find that current backers’ contributions to a particular project are positively influenced by previous backers’ contributions to that project. We confirm thereby the existence of positive ‘intra-project network effects’, which have already been documented in the literature. We also show an entirely novel result, namely the existence of positive ‘inter-project network effects’: current contributions to some project increase with past contributions to other contemporaneous projects on the CFP. Our central estimates indicate that the number of contributions generated by a project on a daily basis is approximately 2% higher following a 10% increase in the number of contributions within the same project (i.e., positive intra-project network effects) and approximately 0.5% higher following a 10% increase in the number of contributions in the other projects on a daily basis (i.e., positive inter-project network effects).
In order to establish more precisely the causal impact of inter-project network effects, we utilize ‘fast starters’, which are projects having generated an unexpectedly high number of pledges during the very first day of their campaign. In a difference-in-differences research design, we examine inter-project network effects on project funding outcomes surrounding fast starters’ first campaign day. Our difference-in-differences estimates indicate that inter-project network effects account for 2-3% increase in the number of contributions that a particular project obtains on a daily basis. We also find that inter-project network effects produced by fast starters are more pronounced on projects belonging to the same category than the fast starter. In the latter case, they account for 7.7% increase in the number of contributions on a daily basis.
Our empirical setting also allows us to get to grips with network effects that arise across the groups of backers and entrepreneurs. Here, the question is: How do backers value the participation of additional entrepreneurs and vice versa? We first observe that the combined impact of these two ‘cross-group network effects’ is positive: participation on the platform generates a positive feedback loop (more entrepreneurs attract more backers, who in turn attract more entrepreneurs). This fuels the growth of the platform and explains how positive inter-project network effects can arise. In economic terms, we document a compound daily growth rate of backers’ contributions on the CFP of 0.25% (or of 172% on an annual basis).
Further analyses allow us to identify the positive cross-group network effect that entrepreneurs exert on backers. Using backer-level data, we examine whether backers’ propensity to pledge again on the CFP relates to the size of the group of entrepreneurs. These tests aim at disentangling the value that backers extract from the participation of additional entrepreneurs. Our estimates show that the probability that backers will contribute again increases by 42.1% as the size of the group of entrepreneurs increases by a one standard deviation. In addition, our backer-level data allow us to disentangle the value that backers generate from the participation of other backers, that is, the intensity of ‘within-group network effects’. Interestingly, we find that within-group network effects go in the opposite direction. Our estimates indicate that backers’ probability to contribute again drops by 49.3% for a one standard deviation increase in the size of the group of backers. In short, cross-group (from entrepreneurs to backers) network effects increase the participation of backers and so the size of the CFP, whereas within-group (from backers to backers) network effects tend to rein in the size of the CFP.
Our findings have significant implications for CFP management and competition. From a managerial perspective, our analysis suggests that the success of a CFP depends not only on the quality and quantity of the projects that are proposed to potential backers, but also on the way these projects are mixed. Because synergies exist between projects (as evidenced by the presence of positive inter-project network effects), CFPs can increase total contributions by choosing the right mix of projects. In this regard, the detailed analysis that we perform at the level of project categories provides CFP managers with useful indications. On Ulule for instance, we show that the ‘Music and Art & Photos’ category is the one that generates the largest synergies; the platform may then want to give more visibility to projects in this category, as they are more conducive to stimulate platform growth. Another important lesson that CFP managers can draw from our work is that recurrent backers behave quite differently from new backers. In particular, we show that projects having a higher fraction of recurrent backers appear to generate more contributions, suggesting that retaining existing backers may yield larger returns than acquiring new backers.
From a competition point of view, our results suggest that reward-based crowdfunding is a ‘winner-takes-all’ type of market: the several sources of positive network effects that we identify create positive feedback loops, which tend to make strong CFPs stronger and weak CFPs weaker. Hence, a CFP that manages to grow faster than its rivals may acquire a self-sustaining competitive advantage, leading eventually to market domination. This also means that the only survival prospects for smaller CFPs are to be found in specialization (finding the right niche) or in consolidation (merging with other CFPs). These implications for CFP competition thus resonate with the heated debate about the dominance of big tech platforms (Amazon, Facebook, Google) and the way network effects may raise entry barriers.
This post is co-written with Paul Belleflamme and Armin Schwienbacher and also published in Coalition Theory Network.
