Beyond financing: crowdfunding as an informational...
Embed Size (px)
Transcript of Beyond financing: crowdfunding as an informational...
Beyond financing: crowdfunding as an informational mechanism
Jordana Viotto da Cruz*
Université Paris 13 (CEPN, Labex ICCA) and Télécom ParisTech
This version: April 2016
Abstract Over the last few years, crowdfunding has financially supported projects that were later endorsed by institutions such as the TIME 25 Best Inventions of the Year and the Museum of Modern Arts. Such acknowledgement provides evidence that crowdfunding can secure monetary resources to high-quality projects. But this new model of access to capital may bring additional advantages to entrepreneurs. In this paper, we test the hypothesis that crowdfunding also serves as an informational mechanism for project owners. It analyzes whether information from crowdfunding projects impacts their decision to release a new product in the market. We use a unique dataset built with publicly available data from several Internet-based sources. In order to isolate the effects of access to capital from the impact of information on the subsequent project owner’s decision, we benefit from the “all or nothing” rule on crowdfunding platforms which conditions access to capital to achieving a given financing threshold. Such rule creates two subsamples – one of project owners who access capital and information, and another of project owners who access information but not capital. Our results show that the probability of project owners who only access information to release the product in the market increases with the crowd’s valuation. Keywords: crowdfunding, information, uncertainty, e-commerce, new product release JEL Codes: L82, D83, G23
* I thank my supervisors, Marc Bourreau and François Moreau, for their guidance and support. Thank you to Paul Belleflamme, Martin Peitz, Christian Peukert, Greg Taylor, Yutec Sun, Marianne Lumeau, Maya Bacache, Lukasz Grzybowski, Paul Jensen, Stephane Lemairé, Annette Schminke, Patrick Wikström, Dainis Zegner and David Zvilichovsky, and the participants of the 13th Conference on the Economics of Information and Communication Technologies, 8th ICT Paris Conference, XXX Jornadas de Economía Industrial, 2015 ESNIE, 3rd Summer School of Digital Economics, 2nd Workshop on Industrial Organization and Digital Economics for their valuable comments and suggestions. All errors are my own. I gratefully acknowledge the financial support of Labex ICCA (Industries Culturelles et Création Artistique). Contact: viotto[at]enst.fr
Crowdfunding is an alternative mode of financing that has provided monetary support for
projects whose high-quality was later endorsed by institutions such as the TIME 25 Best
Inventions of the Year,1 the Oscars,2 the Grammy Awards,3 and the Museum of Modern
Arts (MoMA)4. Besides monetary resources, entrepreneurs presenting their ideas on
crowdfunding platforms may obtain additional benefits from their campaigns. For
example, they can collect information about the public’s valuation of their project.
Producers face great uncertainty preceding the release of new goods in the market
(see, for example, Asplund and Sandin, 1999). Crowdfunding offers an investment
opportunity associated with a consumption experience (Schwienbacher, 2015) where
contributors choose the amount they give to a project, in a sort of incentive-aligned
mechanism (Agrawal, Catalini and Goldfarb, 2013) that allows individuals to reveal their
valuation about a certain idea. From this perspective, the aggregate contributions can be
considered as the “crowd’s valuation” and may provide a proxy about the market’s interest
on the new good, which in turn can help reducing the entrepreneurs’ uncertainty prior to
the release of the product.
Such hypothesis was evoked on past research (Agrawal et al., 2013; Belleflamme,
Lambert and Schwienbacher, 2014; Belleflamme, Omrani and Peitz, 2015), but to the best
of our knowledge not yet tested in the context of new product release on retail channels,
1 Information from the pages dedicated to “The 15 Best Inventions” in 2013 (http://ti.me/17TRn1m), 2014 (http://time.com/3594971/the-25-best-inventions-of-2014), and 2015 (http://time.com/4115398/best-inventions- 2015/). Last consulted on January 6, 2015. 2 Samantha Murphy. “Oscar Win Is a First for Kickstarter-Funded Film”. Mashable, February 25, 2013. Available at http://mashable.com/2013/02/24/inocente-oscar-kickstarter/#egRpHGYu_8q4. Last consulted on December 5, 2015. 3 Jazz musician Maria Schneider was nominated to four Grammy Awards and won in one category with her album “Concert in the Garden” (2004), which was financed through ArtistShare. Information from the artist’s website (http://www.mariaschneider.com/albuminfo.aspx?ID=60) and the Grammy Awards (https://www.grammy.com/). Last consulted on December 5, 2015. 4 Margaret Rhodes. “A CFL Bulb That Is As Practical As It Is Sculptural”. FastCoDesign, January 13, 2014. Available at http://www.fastcodesign.com/3024738/wanted/a-cfl-bulb-that-is-as-practical-as-it-is-sculptural. Last consulted on December 5, 2015.
which is the objective of the present paper. We frame our research question as: how do
project owners respond to information from their crowdfunding campaigns? As
“response”, we consider the decision to release the corresponding good in the market after
a crowdfunding campaign. We posit that the probability of an entrepreneur to release the
corresponding product in the market increases with the crowd’s valuation.
One issue to test this hypothesis is that crowdfunding project owners, when
successful, access simultaneously capital and information. To disentangle both effects, we
utilize the “all or nothing” rule on crowdfunding platforms that conditions access to capital
on the achievement of a certain financing threshold during the campaign.5 Due to this rule,
even project owners that receive support remain unfinanced if their initial target is not
reached. In this case, their respective contributors are reimbursed at the end of the period.
The “all or nothing” rule creates two subsamples of project owners – those who
receive support and obtain access to the capital raised through their campaign and those
who receive support but remain unfinanced. We expect that if information is important
enough to reduce the entrepreneurs’ uncertainty, the probability of releasing the new
product among those who did not succeed will increases with the amount of contribution.
We highlight that failed projects do not necessarily lack quality. Past research
demonstrates that failure is strongly correlated with high financing goals (e.g., Mollick,
2014) and that under the uncertainty context of crowdfunding, only a part of high-quality
projects get financed, particularly if the project owner fails to inform a relative number of
potential contributors at a very early stage of the campaign (Li and Duan, 2014).
In order to test our hypothesis, we focus on projects aiming at producing music
albums. Music is one of the main categories on crowdfunding (the second on the platform
we study in terms of number of projects) and about 40% of projects within the category
5 As explained later, the financing goal and the duration of the campaign are two characteristics determined at the beginning of the campaign and that cannot be changed once the project is online.
aim at creating an album. More importantly, the music industry confronts the same
information asymmetries issues as other markets, particularly concerning uncertainty prior
to the release of a new product. Finally, as in Bacache-Beauvallet, Moreau and Bourreau
(2014) and Agrawal, Catalini and Goldfarb (2015), we consider musicians to be artists-
entrepreneurs who need to access capital in order to release a new product in the market.
Thus we believe we provide insights to other categories.
Our analysis uses a unique dataset built with information collected from different
Internet-based sources. The main one is Kickstarter, considered one of the prominent
crowdfunding platforms worldwide.6 Kickstarter allows project owners to offer early
access to the good or service being developed as well as prizes and “community benefits”
(Belleflamme et al., 2014) in exchange for financial support. Other data sources include
Facebook, Amazon and iTunes.
The final sample contains 707 observations, with both successful and unsuccessful
projects, and we remark that 20% of unsuccessful projects release the respective product in
the market after the crowdfunding campaign. The analysis focuses on the probability of
such project owners to release the corresponding product in the market given the “crowd’s
valuation” (i.e., the amount collected). Our results demonstrate that the likelihood of
releasing after an unsuccessful campaign increases with the amount collected. This result
allows us to conclude that entrepreneurs use not only the capital they raise on
crowdfunding but also the information obtained. In other words, crowdfunding goes
beyond financing: it is also an informational mechanism for entrepreneurs.
Our dataset reveals other interesting characteristics of crowdfunding projects. First,
while the general assumption is that project owners would start production only after
ending the campaign, if successful, our data suggests that, at least for a part of the projects,
6 We detail the four main crowdfunding models on Section 2.
crowdfunding and production happen simultaneously, and not sequentially. Second, our
study shows that some successful projects are not released in the market. We will briefly
discuss these questions in Section 5.
We expect our paper to contribute to the growing literature on crowdfunding,
particularly the stream dedicated to the informational mechanisms arising on these
platforms. This stream has mostly dedicated to understanding the impact of information on
the demand side (contributors) and only recently started investigating the supply side
(project owners). Besides the academic contribution, we expect to provide insights to
entrepreneurs about crowdfunding as an informational mechanism.
To the best of our knowledge, ours is the first paper to study the response of
entrepreneurs to information provided by crowdfunding campaigns in the context of new
product release. It is also the first one to follow projects beyond the crowdfunding
campaigns, connecting projects to online retail outlets (e-commerce).
The paper is organized as follows. Section 2 defines crowdfunding and its main
existing financing models, and reviews the relevant literature. Section 3 describes the
theoretical framework and the hypothesis. Section 4 details the dataset and the variables
used. Section 5 presents the results, and Section 6 brings the concluding remarks.
2. Crowdfunding overview
2.1. Definition and financing models
Crowdfunding can be defined as an “open call through the Internet for the provision of
financial resources either in form of donation or in exchange for some form of reward
and/or voting rights in order to support initiatives for specific purposes” (Belleflamme et
al., 2014). A complementary definition points to “the efforts by entrepreneurial individuals
and groups – cultural, social, and for-profit – to fund their ventures by drawing on
relatively small contributions from a relatively large number of individuals using the
Internet, without standard financial intermediaries” (Mollick, 2014). Both encompass a
broad range of activities developed on distinct types of platforms, which are categorized in
four main models.
In the reward-based model, which is the focus of the present paper, contributors can
receive non-monetary compensations for their financial support, including advanced copies
of products, preferential prizes, and appreciation tokens (Belleflamme et al., 2014). The
donation-based crowdfunding facilitates private contributions to public goods, and
contributors are expected to donate out of altruism or warm glow (Andreoni, 1990). In the
lending-based crowdfunding, also referred to as peer-to-peer lending or social lending,
investors supply funds to individuals, groups or small companies, expecting to be
reimbursed after a given period, with or without interest rates. Finally, in equity-based
crowdfunding, investors become shareholders and receive dividends according to the
companies’ performance and the amount invested.
While some projects occur on the project owner’s website (Belleflamme et al.,
2013), most of the crowdfunding activity concentrates on dedicated platforms
(Belleflamme et al., 2015; Schwienbacher, 2015; Viotto, 2015). Platforms adopt specific
rules such as the “fixed funding” rule (“all or nothing”), which conditions access to capital
to the achievement of a financing threshold that is determined by the project owner at the
beginning of the campaign. Alternatively, some platforms offer the option of “flexible
funding” (“keep it all”), allowing project owners to be financed even if they do not reach
the financing goal.
The overall crowdfunding market has been growing exponentially over the last
years. According to the consulting firm Massolution, in 2015 there were 1,250 platforms
worldwide where individuals and firms transacted US$ 30 billion – twice the volume of
2014. The rapid development has attracted the attention of policymakers regarding a
number of issues, including the potential of crowdfunding to alleviate budget constraints
for new businesses, culture and media. For example, in 2015 the European Commission
signaled the need to understand the role of crowdfunding in the cultural and creative
sectors in order to “support the implementation of political objectives regarding the access
to alternative sources of finance”.7 Such interest demonstrates the relevance of
crowdfunding to regulators.
2.2. Literature review
Crowdfunding has motivated a growing body of academic literature over the past few
years. Our paper is closely related to the stream analyzing informational mechanisms
arising on crowdfunding. Concerning the demand side, research investigates how data
about past contributions (and past contributors’ characteristics) influence future
contributions and projects’ outcomes (Zhang and Liu, 2012; Burtch, Ghose and Wattal,
2013; Lin, Prabhala and Viswanathan, 2013; Parker, 2014). Regarding the supply side,
research focuses on the entrepreneurs’ learning process on crowdfunding platforms by
examining project owners who submit projects more than once (Xu, 2015). Our work also
relates to studies that theoretically evoke the possibility of entrepreneurs to use information
from crowdfunding platforms to infer the crowd’s valuation (Agrawal et al., 2013;
Belleflamme et al., 2014).
Each one of the four main models of crowdfunding present distinct dynamics, and
attract different types of participants on both sides. In this sense, we position our paper
7 See “Call for tenders EAC/03/2015 - Pilot project on “Crowdfunding for the cultural and creative sectors: kick- starting the cultural economy””. Accessible at http://ec.europa.eu/culture/calls/general/0315_en.htm. Last consulted on December 5, 2015.
within the stream particularly interested in the reward-based crowdfunding environment.
One of the central findings in these papers is the importance of the social network for the
success of campaigns. Mollick (2014) reports that the number of project owners’ Facebook
friends (used as a proxy for the size of the social network) is positively correlated with the
probability of success of a project. Agrawal et al. (2015) finds that friends and family tend
to be the first to contribute to a project, revealing private information about the project
owner and signaling quality to “distant” investors8. Colombo, Franzoni and Rossi-
Lamastra (2015) conclude that the internal social capital is crucial in the crowdfunding
Another question concerns the project quality inferred by contributors, and how such
inference influences the project outcomes (success/failure). Using proxies such as the
choice of the model (e.g., “all or nothing” vs. “keep it all”, Cumming, Leboeuf and
Schwienbacher, 2015), videos and spelling (Mollick, 2014) authors conclude that quality
signals are positively correlated with the probability of success. But if quality and success
are positively correlated, failure does not necessarily imply lack of quality. First, success is
also correlated with lower financing goals (e.g., Mollick, 2014). Second, in an uncertain
environment such as crowdfunding, even high-quality projects may fail, particularly if the
project owner cannot inform a relatively high number of potential contributors at a very
early stage of the campaign (Li and Duan, 2014).
One interesting characteristic of reward-based crowdfunding is the timing of
contributions. The distribution of daily contributions is U-shaped, which is explained as the
result of two phenomena – a crowding out in the beginning, and a deadline effect at the end
(Kuppuswamy and Bayus, 2013). The literature also analyzes in the drivers for the
decision of contributors, and reports that intrinsic motivations (Zvilichovsky, Inbar, and
8 The paper analyzes a hybrid form of crowdfunding in the music industry, where investors could receive royalties and rewards.
Barzilay, 2013; Boudreau, Jeppesen, Reichstein and Rullani, 2015) play a great role for
success of projects.
Finally, papers suggest that crowdfunding mitigates geography-related frictions
(Agrawal et al., 2015) facilitating the access to capital to individuals living in areas that are
less served by the traditional systems, but that it still concentrates in big cities and
industrial clusters (Kim and Hann, 2014).
2.3. Financing on reward-based crowdfunding
As we focus on reward-based crowdfunding campaigns, we find it useful to present its
process. A project owner (firm, entrepreneur, association or artist) decides to raise money
through a crowdfunding platform. She sets an account on the platform and creates a
project, describing its expected outcomes and offering rewards to each level of
contribution. As a hypothetical example, a project owner can offer a “thank you” message
on Twitter for contributions of $1 and a special dinner for contributions of $1,000 (see
Figure A.1 in the Appendix for a concrete example of a crowdfunding campaign with some
of its respective rewards).
When setting a crowdfunding project on Kickstarter, the project owner establishes
a financing goal (minimum of $1 with no upper bound) and the duration of the campaign
(1 to 60 days), two features that cannot be changed once the campaign is submitted to the
appreciation of potential online contributors. The project owner invites contributors to
participate on the project, generally using social media like Facebook. At the end of the
campaign, the project owner accesses capital if the campaign is successful, and remains
unfinanced otherwise – in which case contributors are reimbursed.
Figure 1: Illustration of the timeline of reward-based crowdfunding
projects under the “all or nothing” model
3. Theoretical framework and hypothesis
We treat crowdfunding as a sort of incentive-aligned mechanism (Agrawal et al., 2013)
where contributors reveal their valuation on an idea by choosing how much to allocate on a
project and committing to future consumption (Schwienbacher, 2015), potentially offering
entrepreneurs information that reduces their uncertainty prior to a product release. Our
research question is: how do project owners respond to information from their
crowdfunding campaigns? The response we consider in our context is the decision to
release the corresponding good in the market after a crowdfunding campaign. We assume
that each participant expresses her valuation of the project when choosing the amount to
contribute with. Thus the total amount collected can be considered as the “crowd’s
valuation”. In this context, we consider that the probability of an entrepreneur to release
the corresponding product in the market increases with the crowd’s valuation.
The main issue in performing the empirical analysis is that the entrepreneurs’
decision can be affected simultaneously by the access to information and the access to
Call for contributions
Project owner’s decision
Project owner sets its project on a crowdfunding platform. She decides the financing objective, the campaign duration, and the rewards.
Project owner invites individuals to contribute through social network.
Contributors decide whether or not to participate, and how much to allocate on the project, thus revealing their valuation.
If the goal is achieved, the project owner accesses the capital raised. Otherwise, contributors are reimbursed.
Project owner decides whether or not to release the product in the market.
capital. In order to disentangle both effects, we benefit from the “all or nothing” rule on
crowdfunding platforms, which conditions the access to capital to the achievement of a
certain financing threshold during the period of the campaign. It implies that even if the
project owner has received support, she remains unfinanced if failing to reach the target,
and their respective contributors are then reimbursed. The condition creates a subsample of
project owners that do not have access to capital, but that can still use information on the
public’s valuation to reduce their uncertainty. The decision of this group will enable our
understanding about the role of information to reduce uncertainty. We formalize our
hypothesis as: products from unsuccessful projects will be released in the market if the
crowdfunding campaign yields information signaling public’s interest in the project. We
expect that if information is important enough to reduce the entrepreneurs’ uncertainty, the
probability that products from failed projects will be released increases with the crowd’s
The hypothesis may sound counterintuitive so we provide the example of singer
Robyn Landis, who received $8,477, an amount within the average of projects on
Kickstarter (around $7,825; see Mollick, 2014), but with a $14,400 goal she could not
access the capital raised.9 Nevertheless, the corresponding album “Waterproof” was
released in the market.
Our model is as follows:
Pr(release=1| failure, information, previous product, production phase) =
Φ(β1failι + β2 informationι + β3failι*informationι + β4previous_productsι +
β5fail*previous_productsι + β6production_phaseι + β7fail*production_phaseι + ψι)
9 Information from the artist’s Kickstarter campaign, available at http://kck.st/1aYcCAE. Last consulted on December 5, 2015.
The dependent variable is a dummy taking the value 1 if the product corresponding
to the crowdfunding project is identified on online retail channels after the campaign, and 0
otherwise; Φ is the cumulative distribution function of the standard normal distribution; the
variable fail equals 1 if the project was not successful on crowdfunding, and 0 otherwise;
information is the logarithmic form of the total amount collected – which we consider as
the “crowd’s valuation”; 10 previous products is the number of products the project owner
had in the market previously to the crowdfunding campaign; and production phase is a
proxy for the stage of production when the campaign ends (see explanation on Section 4).
The strategy to test our hypothesis relies on the use of the interaction term between the fail
and information, and the corresponding coefficient β3 yields the effect of the information
proxy only to failed projects. We expect that if entrepreneurs respond positively to
information, β3 will be positive.
To ensure that our results are not biased by other factors that may lead an
entrepreneur to release a product in the market after a failed campaign, we include two
other interaction terms that aim at testing alternative explanations. The first one is that
project owners who already have other products in the market may use revenues coming
from sales of these to finance the new one (Van Auken and Neeley, 1996; Ebben, J., &
Johnson, 2006). In this case, project owners with other products in the market would be
more likely to release a new product after a crowdfunding failure. We expect that the
interaction term between fail and previous products to capture this effect if it exists. In this
case, we assume that project owners incur fixed costs to maintain the product on retail
channels; an assumption that is backed by evidence from the industry – to have an album
on Amazon and iTunes, an artist-entrepreneur must pay access fees to digital distributors,
10 Alternatively we test the interaction of fail with number of contributions and average of the amount contributed. The corresponding results are reported on Tables A.1 and A.2 in the Appendix.
for example. 11
The second alternative explanation is the stage of production of the corresponding
good. The general assumption is that crowdfunding and production happen sequentially,
but theses processes can also occur simultaneously. In this case, project owners incur fixed
costs of production, and cover such costs ex post with capital coming from the
crowdfunding campaign, if successful. Thus, if unsuccessful, once fixed costs are already
incurred, the project owner may as well release the product. We control this possibility by
introducing a third interaction term between fail and production phase.
The term ψ is a set of control variables including genre and the number of
Facebook fans, which we use as a proxy for potential public, similarly to previous works
(e.g., Mollick, 2014).
Our primary source of information for the empirical analysis is Kickstarter, considered
one of the main reward-based crowdfunding platforms worldwide, having attracted more
than US$ 2 billion in transactions coming from 10 million contributors up to 2015. The
funding is based on the “all or nothing” principle, and the platform transfers the funds to
the project owners at the end of the funding period if the project collected at least the pre-
established financing goal. A project that cannot achieve its goal is considered
unsuccessful and the contributors are reimbursed.
All project pages on Kickstarter publicly display information on the characteristic
and the performance of the project: initial financing goal, amount collected, location (city,
state and country), number of contributors, category (e.g., film, music, theater),
subcategory (e.g., genre, in the case of music), number of updates on the campaign (by the
11 Information from digital distributor CDBaby.
project owner), comments (by the project owner and contributors), the period of each
campaign (initial and final dates of the project), the description of each reward and the
minimum amount to access it, number of contributors choosing each type of reward, and
the estimated delivery of rewards (see Figures A.1 and A.2 in the Appendix for an
example of a crowdfunding campaign on Kickstarter with some of its respective rewards).
Once the project ends, the pages of successful and unsuccessful projects stay online with
all information as the last day of campaign.
Figure 2: Distribution of categories on Kickstarter according to the
number of projects
Reward-based crowdfunding is largely used for projects with creative, innovative
and/or technological appeal. For example, Kickstarter presents ideas related to games,
design, films, and music, among others. In order to be able to compare projects, we chose
to focus on one category offering outcomes presenting some level of similarity in terms of
product characteristics. We find the music category, more specifically on projects aiming
at producing a music album, to be suitable for our purpose. Music is one of the main
categories in crowdfunding and the second category on Kickstarter in terms of number of
projects (see Figure 2), and about 40% of projects are dedicated to create a music album
(the word “album” appears in 18,466 projects in the Music category on Kickstarter).
As in Bacache-Beauvallet et al. (2014) and Agrawal et al. (2015), we consider
musicians to be artists-entrepreneurs who need to access capital in order to release a new
product (album) in the market. After the crowdfunding campaign, these artist-
entrepreneurs can release their products on traditional channels where consumers normally
purchase other similar goods such as Amazon or iTunes. These two channels lead the
distribution of recording music distribution the last years at the same time they impose
some barriers to non-professional artists. The distribution is done either by vertically
integrating with an incumbent (a label) or independently, through specialized distributors.
In our sample, all artists adopt the latter option, approaching an entrepreneurial attitude.
Although independent distribution imposes low barriers to artists, distributors require
fixed fees to place albums on online retail channels, thus we consider only artists-
entrepreneurs expecting to sell will have incentives to release through these channels.
The music industry faces similar issues regarding information asymmetries as other
markets. For example, on the demand side, consumers can only know the utility of the
consumption ex post. On the supply side, producers are uncertain about the public’s
response to each product until after the release. Thus we believe we provide insights to
We obtain a sample of 1,505 US-based projects aiming at producing a music album
with estimated delivery of rewards between August 2014 and May 2015. Similarly to
previous work (Mollick, 2014; Cumming, Leboeuf and Schwienbacher, 2015), we
eliminate very low or very high financing goals. To decide the upper and lower bounds, we
consider the specificities of the music industry – project owners setting goals of less than
$3,000 will be likely to expect contributions mainly from friends and family, and those
determining goals above $200,000 seem to be unrealistic about the market. Finally, we
discard the projects where the project owner does not have a Facebook page. The final
sample contains 707 observations from unique project owners (i.e., there is no project
owner with more than one project).
Figure 3: Distribution of projects according to the duration of campaigns (in days).
Figure 4: Distribution of projects according to the percentage obtained.
Table 1: Summary statistics
All Sample Obs Mean Std. Dev. Min Max Released 707 0.61 0.49 0 1 Fail 707 0.26 0.44 0 1 Supporters 707 127.55 258.87 0 3305 Collected 707 9958.02 18571.26 0 278486 Average Collected 707 85.88 69.45 0 956.25 Goal 707 10247.20 11915.44 3100 175000 Previous Albums 707 1.26 2.51 0 34 First Album 707 0.51 0.50 0 1 Production Phase 707 2.79 3.00 0 21 Facebook Fans 707 4738.29 22510.51 2 444214
Obs Mean Std. Dev. Min Max
Released 522 0.74 0.44 0 1 Supporters 522 166.08 291.35 10 3305 Collected 522 12960.06 20734.42 530 278486 Average Collected 522 93.79 53.04 28.15432 467.7333 Goal 522 10159.66 12789.06 3100 175000 Previous Albums 522 1.45 2.65 0 34 First Album 522 0.44 0.50 0 1 Production Phase 522 2.65 2.83 0 21 Facebook Fans 522 5724.72 25658.51 14 444214
Obs Mean Std. Dev. Min Max
Released 185 0.25 0.43 0 1 Supporters 185 18.83 25.99 0 193 Collected 185 1487.42 2882.52 0 23815 Average Collected 185 63.57 99.31 0 956.25 Goal 185 10494.21 9030.81 3150 60000 Previous Albums 185 0.72 1.98 0 17 First Album 185 0.70 0.46 0 1 Production Phase 185 3.20 3.43 0 18 Facebook Fans 185 1954.96 8340.71 2 108278
We complement the data from Kickstarter with information from other sources.
First, we consider the number of previous albums from the artists, which is a proxy for
experience and for a source of financial resources (as explained in Section 3). Then, we
obtain the number of Facebook fans, assuming that social media is the main promotion
channel for independent artists (Bourreau, Maillard and Moreau, 2014), and that
information from social media provides a proxy for the social network of the project owner
(Mollick, 2014), or, in our case, of potential public. Finally, we follow these projects on
online retail outlets (Amazon and iTunes)12 from August 2014 until November 2015,
leaving room for delays within the period reported by previous research (Mollick, 2014;
Hauge and Chimahusky, 2016).
Table 1 brings the summary statistics with the main characteristics of the whole
sample, for successful and for unsuccessful projects, and Table 2 displays the main
variables used in the analysis. Consistent with previous research on Kickstarter (e.g.,
Mollick, 2014), campaign duration concentrates around the period of 30 days (Figure 3),
and projects succeed by small margins and fail by large margins (Figure 4). We highlight
the fact that 25% of unsuccessful projects in our sample release a product in the market
after crowdfunding while 26% of successful projects do not. We will discuss both in the
Table 2: Main Variables
Released = 1 if project releases on retail channels (Amazon and iTunes), 0 otherwise. Fail = 1 if project does not reach the financing objective, 0 otherwise. Information log(TotalCollected+1)13
(1) Number of previous albums, (2) dummy for “first album” (=1 if the artist is releasing the first album through crowdfunding, and 0 otherwise).
Period between the end of a campaign and the estimated delivery of main rewards, informed on project page (in months).
We consider it is worthy to provide details on the variable production phase, which
controls for the stage of production. Each project page on Kickstarter provides the
estimated delivery of rewards expressed in months and years. Considering that rewards
12 Figures A.1 and A.2 in the Appendix show examples of projects matching products released in the market. 13 Alternative models using log(AmountCollected+1) and log(AverageCollected+1) are reported in the Appendix.
include the main crowdfunded good, we expect that the project owners to account for the
end of production when setting an estimated delivery. Therefore we would expect that the
period between the end of the crowdfunding campaign and the estimated delivery is an
approximation of the time needed by the project owner to access capital, launch and finish
the production, and deliver the good. The variable “production phase” is expressed in
months – a project with a “production phase” of one month is closer to completion than
another one with a “production phase” of six months.
Figure 5 displays the distribution of the production phase, showing that most
crowdfunding campaigns set the estimated delivery in the same month or one month later
than the end of the campaign. According to industry sources, a music album takes from six
months to one year to be produced, which suggests that most project owners ran the
campaign and the production simultaneously.
One potential concern in our setting is that project owners may have raised the
money outside the crowdfunding platform, but the literature suggests that it would be less
likely as the entrepreneurs come to crowdfunding as an alternative when lacking access to
traditional sources (Hann and Kim, 2014). Another potential concern regards the
possibility of raising money on other platform, but we posit that project owners do not
have incentives to hold the same projects in multiple platforms – while it is not a forbidden
practice, it implies opportunity costs due to the amount of work required to run a campaign
(Hui, Gerber and Greenberg, 2012), and it can also hurt the potential of success as publicly
available information about previous support influences future contributions
(Kuppuswamy and Bayus, 2013; Burtch et al., 2014; Agrawal et al., 2015). Still, we check
for double projects in other platforms, but cannot find any from the final sample.
Figure 5: Percentage of crowdfunding campaigns by period between the end of
the campaign and the estimated delivery of rewards, in months.
Our study relies on the regression of the Probit model presented on Section 3. Table 3
reports the main model with the information proxy “total amount collected”. Two
alternative models with the information proxies “number of contributors” and “average
contributed” are reported in the Appendix (Tables A.1 and A.2) as robustness checks. The
first column on each table presents the results of the estimation using the variable previous
albums as a proxy for previous products. This variable is substituted by the dummy first
album on the second column. On the third column, we use the same model as in the first one
excluding the projects that had zero contributions – project owners failing after having
attracted some contribution may behave differently than project owners that failed for
receiving no contributions at all.
Table 3: Probit estimation results for the decision to release an album in the market. Dependent variable: dummy indicating the release on e-commerce channels.
Information proxy: total amount collected.
(1) (2) (3) Fail -1.329*** -1.161*** -1.259***
(0.253) (0.286) (0.262) Fail*Log(Collected+1) 0.228** 0.178* 0.335***
(0.106) (0.103) (0.116) Log(Collected+1) -0.167* -0.123 -0.158
(0.099) (0.095) (0.100) Previous Albums 0.065*
(0.034) (0.0340) Fail*PreviousAlbums 0.060
(0.057) (0.0612) First Album
(0.259) Production Phase -0.056*** -0.056*** -0.055**
(0.0213) (0.0214) (0.0213) Fail*ProductionPhase 0.005 0.010 0.019
(0.0457) (0.0441) (0.047) Log(FB Fans) 0.067 0.0653 0.0625
(0.042) (0.041) (0.043) Constant 0.611 0.792 0.625
(0.543) (0.566) (0.549)
Observations 707 707 689 Wald chi2 (17) 168.3 (17) 166.7 (17) 164.6 Prob>chi2 0.0000 0.0000 0.0000 Pseudo R2 0.2105 0.2081 0.2111
Robust standard errors in parentheses; *** p
statistically significant for all specifications, meaning that the probability of releasing an
album after a failed crowdfunding campaign increases with the public’s valuation. Such
result is reinforced by the alternative model using average amount collected (Table A.2),
where the interaction term fail*log(average_collected+1) is also positive and statistically
significant. Interestingly enough, the coefficient for the interaction term
fail*log(supporters+1) is only positive and statistically significant when eliminating the
projects that had no contributions (third column of Table A.1). We infer that project
owners care about the crowd’s valuation (measured by the proxy log(collected+1)) and the
individual valuation (measured by the proxy log(average_collected+1)), and less about the
total number of contributors (log(supporters+1)) in order to decide on the product release.
Our results lack evidence that releases after failing on crowdfunding can be explained
by alternative explanations. The coefficients of the interaction terms fail*PreviousAlbums,
fail*FirstAlbum and fail*ProductionPhase are not significantly different from zero.
As we are interested in the release of crowdfunded products in the market, it may
be interesting to note that 26% of the successful projects are not released (see Table 1).
We must highlight that the project owners only commit to deliver rewards, not to release
the product in the market – and we observe the release in the market, not the delivery of
rewards. The literature suggests that fraud is rare but delays are common (see, for
example, Mollick 2014 and Hauge and Chimahusky, 2016). Intrigued by such apparent
inconsistency, we analyzed some successful but not released projects from our sample and
we could identify two main reasons for the non-releasing. The first one is that project
owners deliberately did not place the corresponding product released in the market after
delivering the rewards.14 It is a puzzling finding given that a reward-based crowdfunding
14 For example, Grammy award winner Kenny Loggins raised US$121,797 to record an album of his band Blue Sky Riders, which was distributed to contributors in September 2015, according to updates on the crowdfunding campaign page (and not contested on the comments by contributors). Remaining physical copies of album were available on the band's website, but the product was not made available on Amazon or iTunes, even if the
campaign followed by a release on traditional channels can expand the market for the
project owner (Belleflamme et al., 2014). The second reason is that projects were more
than six months late in relationship to the estimated delivery.15 This situation suggests that
risks for contributors may be greater than what is accounted for, which can put into
question the reputation of artists and platforms. Such questions, though, go beyond the
scope of this paper and may be explored in future work.
In this paper, we provide empirical evidence that the benefits of crowdfunding to
entrepreneurs go beyond the provision of financial support to projects lacking access to
traditional sources of capital. It can also be a valuable source of information about the
“crowd’s” valuation on the presented idea, helping entrepreneurs to reduce their uncertainty
regarding the acceptance of a new product or service, and ultimately providing support for
their decision to release a product in the market.
We depart from the hypothesis that crowdfunding contributors reveal their valuation
about the presented project when deciding whether or not to contribute to the campaign, and
at which amount. The sum of allocations of all contributors provides then information about
the aggregated valuation of the project by the crowd.
The entrepreneurial decision after a crowdfunding campaign can come from two
simultaneous effects: information and access to capital. To ensure that the response we
observe is due to the latter but not the former, we need to isolate both effects. We benefit
from the “all or nothing” rule on crowdfunding platforms, which conditions the access to
previous albums were. The reasons for that go beyond the scope of this paper. 15 For example, musician Paula Fuga (who has worked with names like Jack Johnson, and thus has a reputation) raised $27,797 in June 2014, and posted an updated in November 2015 with the title “Hang in there guys!! I haven't forgotten and No, I didn't rip you off!!” (update only available to supporters). Information from the artist’s crowdfunding campaign available at http://kck.st/1jAq2UX. Last consulted on March 27, 2016.
capital to the achievement of a given financing threshold. In this setting, projects that
achieve or overcome this financing goal are “successful” and have access to the capital
raised while those that stay below this threshold are considered “unsuccessful” and remain
unfinanced, with contributors being reimbursed. Due to this rule, even if project owners
receive support, they remain unfinanced if they do not reach the financing threshold.
We focus our analysis on the decision of the subsample of unsuccessful project
owners, and how it is affected by the crowd’s valuation on their project (i.e., the amount
they collected). With this strategy, we isolate the effect of information from the effect of
access to capital on the likelihood of releasing a product. The results of our empirical
analysis show that the greater the valuation, the more likely the entrepreneur will be to take
the project to a next level, for example, releasing it in traditional retail channels after the
campaign. We thus conclude that that entrepreneurs respond not only to the access to
capital, but also to information from their crowdfunding campaign. In other words,
crowdfunding goes beyond financing: it is also an informational mechanism for project
We highlight that failing on crowdfunding do not necessarily mean lack of quality.
Past research shows that projects with smaller financing goals are more likely to be
successful and that under uncertainty, only a part of high-quality projects get financed,
particularly if the project owner fails to inform a relative number of potential contributors at
a very early stage.
In order to ensure that our interpretation is not biased by other factors, we test
alternative explanations for the release of products coming from unsuccessful crowdfunding
projects. The first one is the use of finance from existing products. We use the number of
previous albums as a proxy for this test. The second one is the fact that the project can be
actually close to completion – or completed – which means that fixed costs are incurred and
that the project owner would try to cover such costs with traditional sales. Our results do not
allow us to conclude that such reasons can be the explanations for a release of a product
coming from an unsuccessful crowdfunding project.
Although we focus on the music category on crowdfunding, we are confident that the
insights are extensible to other categories. Uncertainty prior to the release of a new product
is a traditional issue in many industries. We believe that our study provides contributions to
the understanding of the uses of crowdfunding other than its primary objective of funding
new projects, with empirical evidence to entrepreneurs that this mechanism can be used to
test and validate their ideas on the Internet. Finally, the paper follows the results of
crowdfunding projects in the market, connecting crowdfunding and e-commerce activities.
Agrawal, A., Catalini, C., & Goldfarb, A. (2013). Some Simple Economics of
Crowdfunding. In Innovation Policy and the Economy (Vol. 14).
Agrawal, A., Catalini, C., & Goldfarb, A. (2015). Crowdfunding: Geography, Social
Networks, and the Timing of Investment Decisions. Journal of Economics and
Management Strategy, 24(2).
Aguiar, Luis & Martens, B. (2013). Digital Music Consumption on the Internet: Evidence
from Clickstream Data. JRC Technical Reports, 24(2), 446–465.
Ahlers, G. K. C., Cumming, D., Günther, C., & Schweizer, D. (2015). Signaling in Equity
Crowdfunding. Entrepreneurship Theory and Practice, 39(4), 955-980.
Alexander, P. J. (1994). Entry Barriers, Release Behavior, and Multi-product Firms in the
Music Recording Industry. Review of Industrial Organization, 9(1), 85–98.
Allison, T. H., McKenny, A. F., & Short, J. C. (2013). The effect of entrepreneurial
rhetoric on microlending investment: An examination of the warm-glow effect. Journal
of Business Venturing, 28(6), 690–707.
Andreoni, J. (1990). Impure Altruism and Donations to Public Goods: A Theory of Warm-
Glow Giving. The Economic Journal, 100(401), 464-477.
Asplund, M., & Sandin, R. (1999). The Survival of New Products. Review of Industrial
Organization, 15(3), 219–237.
Bacache-Beauvallet, M., Bourreau, M., & Moreau, F. (2014). Digitization and
Entrepreneurship: Self-releasing in the Recorded Music Industry. Working Paper.
Belleflamme, P., Lambert, T., & Schwienbacher, A. (2013). Individual Crowdfunding
Practices. Venture Capital: An International Journal of Entrepreneurial Finance, 15(4),
Belleflamme, P., Lambert, T., & Schwienbacher, A. (2014). Crowdfunding: Tapping the
Right Crowd. Journal of Business Venturing, 29(5), 585–609.
Belleflamme, P., Omrani, N., & Peitz, M. (2015). The economics of crowdfunding
platforms. Information Economics and Policy. 33, 11-18.
Bernstein, S., Korteweg, A. G., & Laws, K. (2015). Attracting Early Stage Investors:
Evidence from a Randomized Field Experiment. Journal of Finance. forthcoming.
Boudreau, K., Jeppesen, L. B., Reichstein, T., & Rullani, F. (2015). Crowdfunding as
“Donations”: Theory & Evidence. (SSRN Scholarly Paper No. ID 2669545)
Bourreau, M., Maillard, S., & Moreau, F. (2014). Stars vs. Underdogs in Online Music
Markets: The Effect of IT on Visibility, Artists’ Broadcasting, and Fans' Activities.
(SSRN Scholarly Paper No. ID 2441976).
Burtch, G., Ghose, A., & Wattal, S. (2013). An Empirical Examination of the Antecedents
and Consequences of Contribution Patterns in Crowd-Funded Markets. Information
Systems Research, 24(3), 499-519.
Burtch, G., Ghose, A., & Wattal, S. (2014). An Empirical Examination of Peer Referrals in
Online Crowdfunding. ICIS 2014 Proceedings.
Burtch, G., Ghose, A., & Wattal, S. (2015). The Hidden Cost of Accommodating
Crowdfunder Privacy Preferences: A Randomized Field Experiment. Management
Science, 61(5), 949–962.
Byrne, D. (2012). How Music Works. Canongate Books.
Cassar, G. (2004). The financing of business start-ups. Journal of Business Venturing, 19,
Chemin, M., & De Laat, J. (2013). Can Warm Glow Alleviate Credit Market Failure?
Evidence from Online Peer-to-Peer Lenders. Economic Development and Cultural
Change, 61(4), 825–858.
Cholakova, M., & Clarysse, B. (2014). Does the Possibility to Make Equity Investments in
Crowdfunding Projects Crowd Out Reward-Based Investments? Entrepreneurship
Theory and Practice, 39(1), 145–172.
Colombo, M. G., Franzoni, C., & Rossi-Lamastra, C. (2015). Internal Social Capital and
the Attraction of Early Contributions in Crowdfunding. Entrepreneurship Theory and
Practice, 39(1), 75–100.
Connolly, M., & Krueger, A. B. (2006). Rockonomics: The Economics of Popular Music.
In Victor A. Ginsburg and David Throsby (Ed.), Handbook of the Economics of Art and
Culture (Vol. 1, p. 667-719). Elsevier.
Cumming, D. J., Leboeuf, G., & Schwienbacher, A. (2015). Crowdfunding Models: Keep-
It-All vs. All-Or-Nothing (SSRN Scholarly Paper No. ID 2447567).
Doshi, A. (2014). The Impact of High Performance Outliers on Two-Sided Platforms:
Evidence from Crowdfunding (SSRN Scholarly Paper No. ID 2422111).
Ebben, J., & Johnson, A. (2006). Bootstrapping in small firms: An empirical analysis of
change over time. Journal of Business Venturing, 21(6), 851–865.
Edelman, B. (2012). Using Internet Data for Economic Research. The Journal of Economic
Perspectives, 26(2), 189-206.
Essling, C., Koenen, J., & Peukert, C. (2014). Loudness War: How Record Labels
Compete for Consumer Attention. (SSRN Scholarly Paper No. ID 2444708)
Hauge, J. A., & Chimahusky, S. (2016). Are promises meaningless in an uncertaint
crowdfunding environment? Economic Inquiry, 2016.
Hildebrand, T., Puri, M., & Rocholl, J. (2013). Adverse Incentives in Crowdfunding.
SSRN Scholarly Paper No. ID 1615483.
Hui, J. S., Gerber, E. M., & Greenberg, M. (2012). Easy Money? The Demands of
Crowdfunding Work. Technical Report No. 4, (4), 1–11.
Kim, K., & Hann, I.-H. (2014). Crowdfunding and the Democratization of Access to
Capital: A Geographical Analysis. Robert H. Smith School Research Paper.
Kuppuswamy, V., & Bayus, B. L. (2013). Crowdfunding Creative Ideas: The Dynamics of
Project Backers in Kickstarter (SSRN Scholarly Paper No. ID 2234765).
Lehner, O. M. (2013). Crowdfunding social ventures: a model and research agenda.
Venture Capital, 15(4), 289–311.
Li, Z. & Duan, J. (2014). Dynamic Strategies for Successful Online Crowdfunding. (SSRN
Scholarly Papers 2506352.)
Lin, M., Prabhala, N. R., & Viswanathan, S. (2013). Judging Borrowers by the Company
They Keep: Friendship Networks and Information Asymmetry in Online Peer-to-Peer
Lending. Management Science, 59(1), 17–35.
Lin, M. & Wei, Z. (2015). Market Mechanisms in Online Crowdfunding. (SSRN Scholarly
Paper No. ID 2328468).
Loginova, O., Wang, X. H., & Zeng, C. (2012). Learning in Advance Selling with
Heterogeneous Consumers. NET Institute Working Paper.
Ly, P., & Mason, G. (2012). Competition Between Microfinance NGOs: Evidence from
Kiva. World Development, 40(3), 643‑655.
Maillard, S. (2013). Consumer Information in the Digital Age: Empirical Evidence from
the Spillovers in the Music Industry. Working Paper.
Mollick, E. (2014). The dynamics of crowdfunding: An exploratory study. Journal of
Business Venturing, 29(1), 1-16.
Onnée S. & Renault S. (2014), Crowdfunding : vers une compréhension du rôle joué par la
foule. Management & Avenir, 74, p. 117-133.
Parker, S. C. (2014). Crowdfunding, cascades and informed investors. Economics Letters,
Strausz, R. (2015). Crowdfunding, Demand Uncertainty, and Moral Hazard - a
Mechanism Design Approach. Working Paper.
Schwienbacher, A. (2015). Entrepreneurial Risk-Taking in Crowdfunding Campaigns
(SSRN Scholarly Paper No. ID 2506355).
Schwienbacher, A., & Larralde, B. (2010). Crowdfunding of Small Entrepreneurial
Ventures. Handbook of Entrepreneurial Finance, 2010, 1–23.
Van Auken, H. E., & Neeley, L. (1996). Evidence of bootstrap financing among small
start-up firms. The Journal of Entrepreneurial and Small Business Finance, 5(3), 235–
Viotto, J. (2015). « Competition and regulation of crowdfunding platforms: a two-sided market
approach », Communications & Strategies, 99, p. 33-50
Waldfogel, J. (2011). Bye, Bye, Miss American Pie? The Supply of New Recorded Music
Since Napster (Working Paper No. 16882). National Bureau of Economic Research.
Xu, T. (2015). The Informational Role of Crowdfunding. Working Paper.
Zhang, J., & Liu, P. (2012). Rational Herding in Microloan Markets. Management Science,
Zvilichovsky, D., Inbar, Y., & Barzilay, O. (2013). Playing Both Sides of the Market:
Success and Reciprocity on Crowdfunding Platforms. (SSRN Scholarly Paper No. ID
Table A.1: Probit estimation results for the decision to release an album in the market. Dependent variable: dummy indicating the release on e-commerce channels.
Information proxy: total number of supporters.
(1) (2) (3) Fail -1.101*** -0.936*** -1.071***
(0.253) (0.280) (0.262) Fail* Log(Supporters+1)
0.142 0.115 0.257* (0.124) (0.123) (0.132)
Log(Supporters+1) -0.008 0.0137 -0.009 (0.094) (0.091) (0.095)
Previous Albums 0.0494
0.0490 (0.0333) (0.0331)
0.0744 (0.058) (0.061)
Production Phase -0.059*** -0.059*** -0.057*** (0.0212) (0.0214) (0.0212)
Fail*ProductionPhase 0.00687 0.0105 0.0217 (0.0453) (0.0437) (0.0457)
Log(FB Fans) 0.0479 0.0442 0.0475 (0.0413) (0.0410) (0.0423)
Constant 0.639 0.841 0.641 (0.551) (0.572) (0.557)
Observations 707 707 689
Wald chi2 (17) 167.4 (17) 169.5 (17) 169.2
Prob>chi2 0.0000 0.0000 0.0000 Pseudo R2 0.2082 0.2071 0.2061 Robust standard errors in parentheses; *** p
Table A.2: Probit estimation results for the decision to release an album in the
market. Dependent variable: dummy indicating the release on e-commerce channels. Information proxy: average collected.
(1) (2) (3) Fail -2.805*** -2.493*** -3.545***
(0.744) (0.762) (0.834) Fail*Log(Av. Collected+1)
0.353** 0.323* 0.517*** (0.168) (0.167) (0.189)
Log(Av. Collected+1) -0.254* -0.245* -0.244* (0.140) (0.140) (0.140)
Previous Albums 0.047
0.047 (0.0322) (0.0322)
0.084 (0.0551) (0.0562)
Production Phase -0.0592***
(0.0215) (0.0217) (0.0215) Fail*ProductionPhase 0.0120 0.0152 0.0291
(0.0464) (0.0447) (0.0463) Log(FB Fans) 0.0477 0.0477 0.0459
(0.0389) (0.0378) (0.0397) Constant 1.785** 1.917** 1.759**
(0.838) (0.846) (0.842)
Observations 707 707 689 Wald chi2 (17) 169.3 (17) 169.7 (17) 162.1 Prob>chi2 0.0000 0.0000 0.0000 Pseudo R2 0.2107 0.2091 0.2089 Robust standard errors in parentheses; *** p
Figure A. 1: Toad the Wet Sproacket’s Kickstarter campaign
and examples of rewards in detail (right).
Figure A.2: Toad the Wet Sproacket’s crowdfunded album on Amazon.