- 1 - Entrepreneur’s Facial Trustworthiness, Gender, and Crowdfunding Success November 2018 Abstract: This study examines how entrepreneurs’ facial trustworthiness is associated with the success of their crowdfunding campaigns. We adopt a novel dataset collected from Kickstarter crowdfunding platform and employ machine learning-based facial detection techniques to construct a comprehensive facial trustworthiness index for our investigation. Our results show that entrepreneur’s facial trustworthiness is positively associated with crowdfunding campaign success. Specifically, trustworthy-looking entrepreneurs receive 22.8% more amount pledged and attract 9.1% more backers in the crowdfunding campaign compared with untrustworthy- looking entrepreneurs. We also find that female entrepreneur’s facial trustworthiness plays a more prominent role in determining project success than that of male entrepreneur’s. Our study provides the first empirical evidence on how individual facial trustworthiness affects the outcomes of reward-based crowdfunding. Keywords: Facial Trustworthiness; Crowdfunding; Gender; Entrepreneurial Finance Paper #890363
45
Embed
Entrepreneur’s Facial Trustworthiness, Gender, and ... · - 1 - Entrepreneur’s Facial Trustworthiness, Gender, and Crowdfunding Success . November 2018 . Abstract: This study
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
- 1 -
Entrepreneur’s Facial Trustworthiness, Gender,
and Crowdfunding Success
November 2018
Abstract: This study examines how entrepreneurs’ facial trustworthiness is associated with the
success of their crowdfunding campaigns. We adopt a novel dataset collected from Kickstarter
crowdfunding platform and employ machine learning-based facial detection techniques to
construct a comprehensive facial trustworthiness index for our investigation. Our results show
that entrepreneur’s facial trustworthiness is positively associated with crowdfunding campaign
success. Specifically, trustworthy-looking entrepreneurs receive 22.8% more amount pledged
and attract 9.1% more backers in the crowdfunding campaign compared with untrustworthy-
looking entrepreneurs. We also find that female entrepreneur’s facial trustworthiness plays a
more prominent role in determining project success than that of male entrepreneur’s. Our study
provides the first empirical evidence on how individual facial trustworthiness affects the
This study investigates the role of entrepreneurs’ facial trustworthiness in crowdfunding success.
It is widely acknowledged that entrepreneurial activities contribute to the economic growth,
while financial constraint is still the major concern that impedes the action of entrepreneurship
(Chatterji and Seamans 2012). Although professional venture capitalists and angel investors are
only be accessed to a small number of seemingly successful ventures, crowdfunding provides a
new financing solution for a wide range of entrepreneurs who enter into the venture market with
limited product information and tracked records that may help attract commercial funding
sources. In recent years, crowdfunding has grown exponentially in market size and attracted
great attention from both practitioners and scholars. According to Statista (2018), the total dollar
amount of reward-based crowdfunding1 reached USD 919.3 million in the U.S. and USD 6,547.0
million globally in 2017.
Different from traditional commercial funding, such as venture capital, crowdfunding
allows a large group of individuals to provide funds directly to the entrepreneurs, on various
crowdfunding platforms (e.g., Kickstarter), without standard financial intermediaries involved
(Mollick 2014). However, the information about the early stage ventures on the reward-based
crowdfunding platforms is generally self-reported, with limited or no tracked records to justify
their credentials to help funders make their funding decisions (Calic and Mosakowski 2016). Due
to the opaque information environment in crowdfunding and the lack of monitoring from
financial intermediaries, crowdfunding may suffers from severe information asymmetry issues
(Belleflamme, Omrani, and Peitz 2015; Chemla and Tinn 2017; Kalayci, Ekenel, and Gunes
1 In reward-based crowdfunding, entrepreneurs promise investors the product that the proposed project intends to develop rather
than seeking for equity investments. Statista (2018) acquire the total reward-based crowdfunding statistics by focusing on the
reward-based crowdfunding and pre-financing of products, art, music and films, software or scientific research, while excluding
traditional venture capital investment, equity-based crowdfunding, or lending-based crowdfunding.
Paper #890363
- 2 -
2014; Strausz 2017). Thus, other than the desirability of the rewards, funders need to evaluate the
trustworthiness of the information disclosed by the entrepreneurs to mitigate their concerns of
the asymmetric information (Misztal 2013). Entrepreneur’s facial trustworthiness, which is
defined as a funder’s perceptions of an entrepreneur’s ability, benevolence, and integrity (Mayer,
Davis, and Schoorman 1995) based on the entrepreneur’s facial appearance, can be influential
for funder’s funding decisions. This paper aims to investigate whether an entrepreneur with more
facial trustworthiness are more likely to experience fundraising success in crowdfunding.
The psychology and neuroscience literature suggest that people are efficient in judging
the trustworthiness of others based on their facial features and also tend to incorporate such
facial trustworthiness judgement into their subsequent social decision-making (Borkenau, Brecke,
Möttig, and Paelecke 2009; Todorov, Loehr, and Oosterhof 2010; Todorov, Olivola, Dotsch, and
Mende-Siedlecki 2015). People could rapidly develop their perceptions of the facial
trustworthiness (Porter, England, Juodis, Ten Brinke, and Wilson 2008), and this appearance-
based trustworthiness perceptions are highly correlated with each other (Blankespoor, Hendricks,
and Miller 2017; Oosterhof and Todorov 2008; Rule, Krendl, Ivcevic, and Ambady 2013). Prior
research documents the implications of facial trustworthiness in various business settings
(Blankespoor et al. 2017; Duarte, Siegel, and Young 2012; Rezlescu, Duchaine, Olivola, and
Chater 2012) with a general impression that people are more likely to trust an individual who
appears more trustworthy when making business decisions. Trust can mitigate expected moral
hazard under asymmetric information associated with business transactions (Al-Najjar and
Casadesus-Masanell 2001) and may serve as a “lubricant” to facilitate social transactions
(Misztal 2013). Given the information asymmetry concerns in crowdfunding market, trust on
entrepreneurs can alleviate funders’ concerns about the credibility of entrepreneurs’ self-reported
Paper #890363
- 3 -
project information. Therefore, we hypothesize that funders are more likely to fund a project
developed by an entrepreneur who appears more trustworthy.
In addition, gender stereotype in entrepreneurship advocates masculine over feminine
(García and Welter 2013; Ogbor 2000). Women are often perceived as less competent with their
innovation and business development skills and tend to receive less venture resources compared
with men in entrepreneurship (Lerner and Almor 2002; Mitchelmore and Rowley 2013; Thébaud
2010). Due to such stereotype, funders on crowdfunding platforms may exhibit higher
uncertainty concerns when evaluating entrepreneurial ventures developed by women. Moreover,
Seidman and Miller (2013) find that individuals tend to pay more attention to the physical
appearance of female individuals than that of male individuals when browsing the information
on online social networking sites. One possible explanation is that men are often evaluated based
on the combination of different personal traits, while women’s appearance tends to be the focal
point when being evaluated (Brownmiller 1984; Kaschak 1992). Thus, compared with that of
male entrepreneur’s, female entrepreneur’s facial trustworthiness may play a more important role
to affect funders’ evaluation of the entrepreneurial ventures and their funding decisions. We
therefore hypothesize that the positive association between entrepreneur’s facial trustworthiness
and crowdfunding success is strengthened if the entrepreneur is female.
To examine our hypotheses, we adopt a web crawler to extract the detailed funding
information of 1,770 technology related projects on Kickstarter, one of the most popular and
successful crowdfunding platforms2 (Strausz 2017). We focus on the technology category among
all venture projects on Kickstarter because information asymmetry problems are more prevalent
for technology focused entrepreneurial ventures due to the secrecy of their core techniques
2 As of October 2018, the number of launched projects on Kickstarter amounted to 419,755, with the success rate among these
projects amounting to 36.4% and 3.92 billion U.S. dollars had been pledged to these projects (Kickstarter 2018).
Paper #890363
- 4 -
especially in early stage (Kousari 2011). We identify founding entrepreneurs’ headshot pictures
and employ machine learning-based facial detection techniques (Dalal and Triggs 2005; Kazemi
and Sullivan 2014; Sagonas, Tzimiropoulos, Zafeiriou, and Pantic 2013) to measure
entrepreneur’s facial trustworthiness based on four distinctive facial features, including inner
eyebrow ridge, angle of the chin, roundness of the face, and length of the lip-to-nose distance, as
suggested by prior neuroscience and psychology literature (Dotsch and Todorov 2012; Enlow
and Hans 1996; Robinson, Blais, Duncan, Forget, and Fiset 2014; Todorov, Baron, and
Oosterhof 2008; Vernon, Sutherland, Young, and Hartley 2014).
Our empirical results show that entrepreneurs’ facial trustworthiness is positively
associated with crowdfunding success, which is measured by 1) the likelihood of fundraising
success, 2) the total dollar amount pledged, 3) the percentage of total dollar amount pledged over
a campaign goal, and 4) the total number of backers. We also find that entrepreneurs’ gender
moderates the relationship between entrepreneurs’ facial trustworthiness and crowdfunding
success. More specifically, female entrepreneur’s facial trustworthiness plays a more important
role in determining fundraising success on Kickstarter after controlling other project related
factors. Our results are robust to both first-time entrepreneurs and seasoned entrepreneurs on
Kickstarter, and also to U.S. entrepreneurs’ sample only. Moreover, our results still hold after
controlling for entrepreneurs’ facial attractiveness (Graham, Harvey, and Puri 2016; Halford and
Hsu 2014).
Our paper has a main contribution to the growing literature on crowdfunding
(Belleflamme et al. 2015) by demonstrating that funders evaluate entrepreneurs’ pictures on
crowdfunding platforms, and incorporate their perceptions of entrepreneurs’ facial
trustworthiness in their funding decisions. Our study is most related to Duarte et al. (2012), who
Paper #890363
- 5 -
investigates the role of appearance-based impression plays in peer-to-peer lending, but differs in
1) funders on Kickstarter need to rely on the limited information voluntarily provided by the
entrepreneurs, while funders on peer-to-peer lending platform can easily gain access to hard
information, such as the fundraiser’s credit score range to assist their funding decisions; 2) we
report the novel evidence that female entrepreneur’ facial trustworthiness plays a more important
role than that of male entrepreneur’s in determining crowdfunding success; 3) we construct the
appearance-based measure using machine learning based facial feature point detection
techniques. The availability and use of the new methodology sheds light on fruitful directions for
future studies on facial trustworthiness in various business and economic decision settings. The
results of our study are important for financial regulators to regulate the information environment
in crowdfunding by developing more transparent and reliable disclosure policies. Additionally,
our study adds to the literature on how trust shapes financial decisions. Existing studies
document that trust increases households’ uses of trust-intensive contract instead of using only
cash (Guiso, Sapienza, and Zingales 2004), individual’s participation in stock market (Guiso,
Sapienza, and Zingales 2008), international trades (Guiso, Sapienza, and Zingales 2009), and
venture capital investments (Bottazzi, Da Rin, and Hellmann 2016). These studies rely on
measures of generalized societal trust based at the national level. Our study emphasizes
individual differences in the facial trustworthiness, thus expands societal trust to trust at the
individual level and reduces the concern that individual level trust may be heterogeneous within
a geographic region. Finally, our paper is related to finance literature on appearance in corporate
context. Appearance affects managerial compensation (García and Welter 2013), hedge fund
investment (Pareek and Zuckerman 2014) and shareholder value (Halford and Hsu 2014). We
Paper #890363
- 6 -
extend this line of research to the context of entrepreneurial finance and suggest that the
appearance of entrepreneurs with early-stage venture matters in a crowdfunding setting.
The rest of the paper is organized as follow. Section 2 reviews related literature and
develops testable hypotheses. Section 3 describes the sample construction, measurement of facial
trustworthiness, and empirical methodology. Section 4 and 5 present main empirical results and
robustness tests. Section 6 is the concluding remarks.
2. Literature Review and Hypothesis Development
2.1 Crowdfunding and Information Asymmetry
Crowdfunding provides an external financing tool for early stage entrepreneurial ventures to
raise funds on the internet from a large crowd.3 However, the funders (i.e., the crowd) may face
information asymmetry problems in crowdfunding. Entrepreneurs seeking external financing
through crowdfunding platforms tend to present very limited historical performance data and
tracked records of success (Belleflamme et al. 2015; Calic and Mosakowski 2016; Strausz 2017),
and are also reluctant to disclose their innovation secrets to the general public, especially to their
competitors, prior to selling their mature products (Agrawal, Catalini, and Goldfarb (2014).4
Funders, with the limited information being disclosed by entrepreneurs, are not able to exercise
their due diligence to evaluate the crowdfunding projects and therefore are less likely to have
access to the true ability of the entrepreneurs or the true quality of the products. Due to the lack
of necessary information to evaluate the proposed campaigns and an effective monitoring by
3 Compared with traditional financing with financial intermediaries, crowdfunding exhibits obvious strength that it allows
entrepreneurs to learn about potential customers’ demand before the project is launched. da Cruz (2018) argue that crowdfunding
serves as an informational mechanism that provides a valuable information on the crowd’s valuation about the presented idea,
helping entrepreneurs to mitigate the uncertainty regarding the acceptance of a new product or service. Based on the follow-up
survey data, Mollick and Kuppuswamy (2014) find that besides providing funded money, crowdfunding also helps provide
access to customers, press, employees, and outside funders. Crowdfunding effectively reduces social structural constraints
through the “democratization of access to capital” (Greenberg and Mollick 2017). 4 Funders often made their funding decisions mainly based on the information presented in the “pitch”, a textual, graphical and/or
video campaign description about the projects posted on Kickstarter (Mollick 2014).
Paper #890363
- 7 -
traditional financial intermediaries, entrepreneurs may engage in short-term opportunistic
behavior that results in moral hazard problem (Agrawal et al. 2014).
Chemla and Tinn (2017) model moral hazard problem in crowdfunding by arguing that
entrepreneurs may embezzle the funds raised on internet-based crowdfunding platforms, e.g.,
Kickstarter, because they are not legally responsible for guaranteeing product delivery. Strausz
(2017) develops a theoretical model of profit-maximizing contract, in the presence of
entrepreneurial moral hazard, consumers’ private information about the demand and
entrepreneurs’ private information about their production cost, to provide insights on how to
control moral hazard in crowdfunding. Hildebrand, Puri, and Rocholl (2016) highlight another
ethical consequence of information asymmetry in a lending-based crowdfunding based on the
evidence that group leader bids, which may have higher default rates, tend to be perceived as a
positive signal and are rewarded with lower interest rates. They argue that potential lenders are
taken advantage of by unscrupulous loan originators due to information asymmetry issues.
2.2 Trust and Its Role to Alleviate Information Asymmetry Problems
Trust refers to the subjective probability that individuals attribute to the possibility of being
cheated by evaluating the subjective characteristics of the person being trusted (Guiso et al.
2008). Align with the social capital literature, studies about the effects of trust on economic
efficiency have shown that social trust affects household’s financial decision (Guiso et al. 2004),
individual’s stock market participation (Guiso et al. 2008) and thus economic growth (La Porta,
Florencio, Shleifer, and Vishny 1997). Bilateral trust between nations also predicts the trade
relations and investments between two countries (Guiso et al. 2009). Trust may also positively
affect venture capital firms’ investment decisions (Bottazzi et al. 2016). These findings indicate
that financial markets function more efficiently when the degree of generalized trust is higher.
Paper #890363
- 8 -
One potential mechanism that trust facilitates inter-organizational efficiency is to reduce
information asymmetry between the contracting parties (Al-Najjar and Casadesus-Masanell 2001;
Chami and Fullenkamp 2002). The “crowdfunding contract” between entrepreneurs and funders
offers little tools for funders to fully understand the project and to track project information once
they commit their capital. Trust on entrepreneurs, ignited by potential funders evaluating
entrepreneurs’ facial trustworthiness, can help alleviate funders’ concerns about the asymmetric
information. Thus, funders may condition their funding decisions on how much they “trust” an
entrepreneur based on their observations of entrepreneurs’ facial trustworthiness.
In a recent study, Lin and Pursiainen (2017) document a positive association between
social capital of the entrepreneur’s home county and the success rate of crowdfunding campaigns.
The underlying assumption of this study is that the level of trust is determined by general prior
beliefs or stereotypes, which are heterogeneous across different geographical regions, but
homogeneous within a certain geographical region. Bottazzi et al. (2016) classify trust into two
different categories: (i) personalized trust which focuses on a specific trading partner; and (ii)
generalized trust which concerns regional “institutions” and call for further research that
disentangles the effects of these two types of trust in business settings. To the best of our
knowledge, however, very few prior studies examine the relationship between entrepreneur-
specific trustworthiness at the individual level in crowdfunding setting. Our study aims to
provide the first attempt to investigate how entrepreneur’s facial trustworthiness affects funder’s
funding decisions, as demonstrated by the performance of crowdfunding campaigns.
2.3 Facial Trustworthiness and Crowdfunding Success
People can quickly generate an impression of the personal traits (e.g., trustworthiness) of other
people by visually observing their facial appearance (Bar, Neta, and Linz 2006; Todorov et al.
Paper #890363
- 2 -
2015). Todorov et al. (2015) suggests that people are able to form a trustworthiness impression
based on the facial features of others with an exposure of as little as 34 milliseconds. Longer
exposures to a face would not alter people’s initial perception of trustworthiness, but rather
reinforce the confidence in prior judgments (Todorov et al. 2015; Willis and Todorov 2006).
Prior studies find that certain facial traits affect how individuals are perceived as
trustworthy or untrustworthy (Dotsch and Todorov 2012; Enlow and Hans 1996; Robinson et al.
2014; Todorov et al. 2008). For example, Enlow and Hans (1996) find that shallow cheeks and
low eyebrow ridge lead to low perceived trustworthiness. Todorov et al. (2008) suggest that
upper inner eyebrow ridge, pronounced cheekbones, wide chin, and shallow nose sellion are
correlated with high facial trustworthiness. The distance between the mouth and the nose is also
an important feature for trustworthiness judgments (Todorov et al. 2008), with longer lip-to-nose
distance tend to signal less trustworthy looking. Dotsch and Todorov (2012) show that a
trustworthy-looking face is associated with a smooth and small face, a smiling mouth, and open
eyes, while an untrustworthy-looking face is associated with a downturned mouth with thick lips,
angry-looking eyes, sagging cheeks, and a bold spot on top of the head. Using “Bubbles”
technique, Robinson et al. (2014) find that facial traits in human eye and mouth areas are
correlated with trustworthiness judgments.5
Laboratory studies show that people are less willingly to trust an individual who has an
untrustworthy-looking face, but more likely to trust people who look more trustworthy. For
example, in criminal cases, defendants who have untrustworthy-looking faces are more likely to
5 Emotional expressions, especially smiling, could influence the trustworthiness decisions (Eckel and Wilson 1998; É thier-
Majcher, Joubert, and Gosselin 2013; Krumhuber et al. 2007; Scharlemann, Eckel, Kacelnik, and Wilson 2001). Additionally,
face typicality is also a critical determinant in facial trustworthiness evaluations (Sofer, Dotsch, Wigboldus, and Todorov 2015).
In our study, we focus on static facial features to measure facial trustworthiness for two reasons. First, typical entrepreneurs’
picture from Kickstarter is wearing a smile, resulting in little variance in measuring facial trustworthiness via smiling facial
expression. Second, facial typicality is closely associated with a pool of faces that funders have ever seen and interacted with, so
that it may be impossible for us to define face typicality for every funder from the crowd.
Paper #890363
- 3 -
receive guilty verdicts (Porter, ten Brinke, and Gustaw 2010). In business settings, for example,
people tend to invest more money in trustworthy-looking business partners (Duarte et al. 2012;
Rezlescu et al. 2012; Tingley 2014). Firms with CEOs who are perceived as more trustworthy
also tend to receive higher valuations at all stages of the IPO (Blankespoor et al. 2017). People
also use facial trustworthiness to modify their wagering decisions (Schlicht, Shimojo, Camerer,
Battaglia, and Nakayama 2010). People’s inferences of facial trustworthiness tend to be highly
consistent with each other (Kim and Rosenberg 1980; Oosterhof and Todorov 2008; Rosenberg,
Nelson, and Vivekananthan 1968; Rule et al. 2013).
Entrepreneurs’ headshots on Kickstarter may provide important facial cues to assist
funders evaluate entrepreneurs’ trustworthiness.6
When potential funders are reviewing
information disclosed from fundraising webpage, they may feel uncertain about the true quality
of the proposed project due to high information asymmetry. If entrepreneurs’ pictures are
provided on the webpage, funders may naturally interpret the level of trustworthiness from
entrepreneur’s static facial features and incorporate such facial trustworthiness information into
their subsequent crowdfunding decision-making.7 Funders are likely to trust an entrepreneur with
trustworthy-looking facial appearance and tend to fund the campaign developed by a
6 We identified 16,122 Kickstarter projects in technology category from October 2009 to September 2017, among which about
11.52% (1,858 projects) uploaded entrepreneur pictures on the fund-raising webpage. We further exclude group photos and 1,802
high quality single pictures of entrepreneurs remained. The entrepreneur’s picture is located on the upper left corner of the
Kickstarter project webpage with the name of the entrepreneur well specified, so that we are confidently to argue that the picture
is the entrepreneur herself. Additionally, by single click the picture, the crowd can see the biographical information of the project
entrepreneur. By further clicking “See full profile”, the entrepreneur’s picture is clearly displayed. Therefore, the entrepreneur‘s
picture, if voluntarily provided by the entrepreneur, is part of funders’ information set that facilitates them to make crowdfunding
decisions. One may refer to the following weblink as an example: https://www.kickstarter.com/profile/966842631/about 7 We argue that funders’ trustworthiness perceptions based on entrepreneur’s facial features are made unintentionally. Prior fMRI
studies find that facial trustworthiness evaluation is correlated with the activation of amygdala (Winston, Strange, O'Doherty, and
Dolan 2002). Neuroscience literature suggests that people unconsciously make perceptions based on facial features (Evans 2008;
McClure, Laibson, Loewenstein, and Cohen 2004). Amygdala is partially responsible for System 1 thinking process described as
rapid, intuitive, and universal, compared with System 2 thinking process that is slower, controlled, and logical (Evans 2008).
McClure et al. (2004) find that limbic (including amygdala) activation may explain the impulsive behavior associated with the
sight, smell, or touch of a desired object. Since amygdala’s dual function enables people to rapidly interpret facial trustworthiness
(Todorov and Engell 2008; Winston et al. 2002) and to react rapidly and intuitively to such perceived trustworthiness (Evans
2008; McClure et al. 2004), we argue that on crowdfunding platforms, funders do not intentionally search for the information
from entrepreneur’s facial features, but rather are unconsciously influenced by entrepreneur’s facial trustworthiness.
trustworthy-looking entrepreneur. This discussion leads to our first hypothesis stated in an
alternative form as following:
Hypothesis 1: Entrepreneurs who look more trustworthy are more likely to be funded in
crowdfunding campaigns.
2.4 Gender, Facial Trustworthiness, and Crowdfunding Campaign Success
The gender disparity in the labor market has been widely recognized in both academic and
practice. For example, women receive 22% lower wages than man, and hold only 6% of the
Chief Executive Officer and other top executive positions in U.S. corporations (Tate and Yang
2015). Koenig, Eagly, Mitchell, and Ristikari (2011) argue that “stereotypes often are a potent
barrier to women’s advancement to positions of leadership”. Bird and Brush (2002) also indicate
that entrepreneurship has been recognized as a male dominated profession. Due to the gender
stereotypes in business ventures, entrepreneurial activities may be viewed from a gender-biased
perspective, which prioritizes masculine over feminine (García and Welter 2013; Ogbor 2000).
Funders on Kickstarter may be potentially biased by this gender stereotype when evaluating the
proposed projects and consider female entrepreneurs as less capable in terms of managing the
ventures and achieving successful outcomes. One possible strategy to alleviate funders’ concerns
about the ability of female entrepreneurs is to gather additional information, such as
trustworthiness indicators from entrepreneurs’ facial appearance, to evaluate the likelihood of
success for female entrepreneurs’ venture projects. Therefore, funders may put more weight on
the facial trustworthiness of female entrepreneurs than those of male entrepreneurs when
evaluating Kickstarter campaigns.
Moreover, evidence from Online Social Networking sites suggests that people tend to pay
more attention to females’ profile photograph than to that of males (Seidman and Miller 2013).
Paper #890363
- 5 -
For example, Seidman and Miller (2013) created eight Facebook profiles with eight photographs
including two attractive and two unattractive for both genders. The text profiles for these eight
profiles are rated as most equivalent on personalities. Participants’ eye movements were tracked
while they viewed each profile for 60 seconds. They found that participants spent significantly
more time viewing profile photos of their female targets than male targets.8 Such evidence is
highly consistent with feminist theory arguing that physical appearance is more essential for
achieving social status for female than for male, whereas males are more likely to be evaluated
on a broader spectrum of other traits (Adolphs, Baron-Cohen, and Tranel 2002; Brownmiller
1984; Kaschak 1992). Based on these prior studies, we expect that funders on Kickstarter are
more likely to extract and utilize the information from female entrepreneurs’ photos than from
male counterparts. Therefore, female entrepreneurs’ facial trustworthiness may play a more
important role in funders’ backing decisions than that of male entrepreneurs. We hence propose
our second hypothesis in the alternative form as following:
Hypothesis 2: The effect of entrepreneur’s facial trustworthiness on crowdfunding
success is more pronounced for female entrepreneurs than for male entrepreneurs.
3. Research Methodology
3.1 Sample Collection
We collect our sample on Kickstarter, one of the largest and most popular reward-based
crowdfunding platforms.9
We focus on technology-related projects because entrepreneurs of
technology-related projects tend to be reluctant to disclose information related to their core
8 In additional analysis, Seidman and Miller (2013) report insignificant associations between participant gender, Facebook profile
gender, and physical attractiveness and dependent variables. They hence exclude the alternative explanation that participant’s
attention toward the attractive individuals of the opposite gender is driven by sexual interest. 9 There are four prevalent types of online crowdfunding platforms in the market, including equity-based crowdfunding (investors
gain dividends, e.g., CircleUp, AngelList), donation-based crowdfunding (donors make benevolent contributions, e.g., JustGiving,
GiveForward), lending-based crowdfunding (lenders earn bonuses or interests, e.g., Prosper, Kiva), and reward-based
crowdfunding (funders receive rewards from creators, e.g., Kickstarter, RocketHub). Among different crowdfunding platforms,
finance researchers have increasingly focused on lending-based and reward-based types, and Kickstarter is among the
crowdfunding platforms that are most frequently studied.
Paper #890363
- 6 -
techniques, and thus are more likely to suffer from information asymmetry issues (Kousari 2011).
Technology-related projects also closely resemble entrepreneurial ventures in the conventional
venture capital markets. We adopt a web crawling algorithm to crawl data about the
characteristics of the projects, including the geographic locations, the dates of initiation and
deadline, pledged goals, number of backers and total amount pledged, as well as entrepreneurs’
names and profile photos from Kickstarter. Our algorithm initially extracts 16,122 Kickstarter
projects in the technology category that were active from October 2009 to September 2017.
Among them, only 1,858 projects (approximately 11.52%) provide high quality profile pictures
of the entrepreneurs who developed the projects. We further exclude 56 group photos (multiple
entrepreneurs are included in one picture), because it is hard to identify the most influential
person in the founding team from the picture. Our final picture sample includes 1,802 high
quality individual entrepreneur’s pictures. We also extract country specific characteristics that
may affect the development of the projects in their geographically located countries. These
country specific characteristics include the national annual gross domestic product per capita,
and the national social capital, which is a country level trust index provided by the World Values
survey. This process further reduced our sample to 1,770 projects from seventeen countries
(regions)10
from 2009 to 2017.
3.2 Facial Trustworthiness Measurements
We employ a machine-learning based face detector11
to identify the facial features of
each entrepreneur using their profile picture and measure their facial trustworthiness. Following
prior studies, we identify four specific facial features that are suggested to contribute to the
10 These countries/regions include Australia, Austria, Canada, Switzerland, Germany, Spain, France , United Kingdom, Hong
Kong, Italy, Mexico, Netherlands, Norway, New Zealand, Singapore, Sweden, and United States. 11 The face detector is built on top of the classic Histogram of Oriented Gradients (HOG) feature combined with a linear classifier
(Dalal and Triggs 2005; Zhu, Yeh, Cheng, and Avidan 2006). A facial feature point detection based on an ensemble of regression
trees is used to estimate the positions of the specific facial features and calculate the facial trustworthiness measurements
(Kazemi and Sullivan 2014; Sagonas et al. 2013).
Paper #890363
- 7 -
trustworthiness perceptions. Specifically, we calculate the angle of inner eyebrow ridge
(EYEBROW), which is suggested to be negatively associated with facial trustworthiness
(Todorov et al. 2008). FACE captures the roundness of the face, which is positively associated
with perceived trustworthiness (Berry and Zebrowitz-McArthur 1988; Gorn, Jiang, and Johar
2008; Livingston and Pearce 2009; Todorov et al. 2008). CHIN is defined as the wideness of the
chin on a face. People with a wider chin is likely to be perceived as more trustworthy (Todorov
et al. 2008). PHILTRUM measures the lip-to-nose distance scaled by the upper facial height.
People with a long PHILTRUM tend to be perceived as less trustworthy (Todorov et al. 2008;
Vernon et al. 2014). Appendix A describes the detailed procedures that we employed to measure
these four facial features.12
Prior literature suggest that when exposed to faces, people quickly build “holistic
representations”, which processes information from various facial feature as an integrated
perceptual whole (Taubert, Apthorp, Aagten-Murphy, and Alais 2011). Consistently, we
construct a composite facial trustworthiness index (TRUST) by aggregating EYEBROW, FACE,
CHIN, and PHILTRUM using the following procedure. First, we standardize each of the four
individual measures by deducting the sample mean and scaling by the sample standard deviation
for the ease of comparison across different facial features (Beatty, Ke, and Petroni 2002; Beaver,
McNichols, and Nelson 2007; Henry and Leone 2015). We then reverse the signs of EYEBROW
and PHILTRUM because prior studies suggest that these two measures are negatively associated
with facial trustworthiness. Third, we average the standardized values of FACE and CHIN, and
12 All of these four facial trustworthiness measurements are 2-D measurements, since the CEO’s and CFO’s pictures we collected
from Google Images are 2-D formats. Prior literature also indicates that some 3-D measurements, such as pronounced
cheekbones (Enlow and Hans 1996; Todorov et al. 2008; Vernon et al. 2014) and shallow nose sellion are associated with more
trustworthy-looking faces. However, as 2-D pictures do not have depth dimension, we cannot calculate 3-D trustworthiness
measurements using 2-D pictures.
Paper #890363
- 8 -
the reversed standardized values of EYEBROW and PHILTRUM to construct the composite
measure TRUST. A higher value of TRUST indicates a more trustworthy facial appearance.
3.3 Sample Descriptive Statistics
Table 1 Panel A depicts the number of Kickstarter technology projects and the successful
rate in our sample by year. The number of technology projects that provides entrepreneurs’
profile photos on Kickstarter increases from 6 projects in 2009 to 406 projects in 2017.
Interesting, annual success rates, which drops from 66.67% in 2009 to 13.79% in 2017, indicate
that backers are more conservative in funding technology projects13
. Table 1 Panel B plots
Kickstarter projects and success rate by country. The majority (67.2%) of our sample come from
the United States. The overall success rate of crowdfunding is 31.2% throughout the sample
period. Projects from some countries, e.g., Norway (57.4%), Sweden (40.0%), United States
(33.5%), and Hong Kong (33.3%), exhibit higher success rate in crowdfunding, compared with
projects from other countries.
[Insert Table 1 about Here]
Table 2 tabulates the descriptive statistics. 31.2% of sample technology projects are
successfully funded. The great majority (89.8%) of entrepreneurs are male, consistent with
Greenberg and Mollick (2017)’s argument that Kickstarter technology category is a “male-
dominated” domain. There are about 4.7% of the projects developed by experienced
entrepreneurs on Kickstarter, evidenced by their previous crowdfunding application records,
regardless of the funding results. Moreover, among these technology projects in our sample, 76.2%
of them provide video presentations of their products, which is viewed as a positive quality
13 The project success rate calculated in our sample is also consistent with that generated from Kickstarter project population.
According to Bidaux (January 26, 2018), compared with success rate of all Kickstarter projects in 2017 (36.7%, 19,348
successfully funded projects over a total of 52,741 projects), project success rate in technology category is much lower (20.4%,
1,253 successful projects over total 6,160 projects). Additionally, total number of technology projects in 2017 also declined by
13.1% compared with that in 2016 (6,160 projects in 2017 vs. 7,089 projects in 2016).
Paper #890363
- 9 -
indicator (Greenberg and Mollick 2017). Table 3 presents the Pearson correlation coefficient
matrix, which shows that our facial trustworthiness index TRUST is significantly and positively
correlated with the total dollar amount successfully pledged (0.063), the percentage of total
dollar amount pledged over project goal (0.063), and the number of backers (0.052). Variables
are defined in Appendix B.
[Insert Tables 2 and 3 about Here]
3.4 Research Design
3.4.1 Effect of entrepreneur’s facial trustworthiness on crowdfunding success
Hypothesis 1 predicts that entrepreneurs with trustworthy-looking facial features are more likely
to be successful in crowdfunding. We estimate the following regression model (1) to test this
make campaign less likely to be funded. Thus, we control for project goal (GOAL) and expect
that projects with lower goal are related to higher success rate. We also include the duration of
the projects (DURATION) as a control variable (Calic and Mosakowski 2016; Lin and Pursiainen
2018). Moreover, we control for entrepreneur’s experience in crowdfunding on Kickstarter
(PAST_EXPERIENCE) using a binary variable with 1 indicates an entrepreneur has at least one
prior project application on Kickstarter, and 0 otherwise. We anticipate that
PAST_EXPERIENCE is valued by funders and thus positively associated with crowdfunding
success. Greenberg and Mollick (2017) suggest that a video that describes project details is
usually a signal for a higher quality and well prepared project. We hence include an indicator
variable video (VEDIO), which equals 1 if the project webpage provides a video. In addition, we
control for country-level social capital index (SOCIAL_CAPITAL), as Lin and Pursiainen (2017)
suggest that crowdfunding projects originated in countries with higher social capital tend to have
14 Given a significant number of campaigns with a pledged amount of zero, using OLS regressions may yield biased estimates.
We further use Tobit models to estimate the three continuous variables and find our results still hold. 15 In crowdfunding platforms, ventures developed by women entrepreneurs tend to be viewed as more trustworthy and are more
likely to be successful than those developed by male entrepreneurs (Eckes 2002; Johnson, Stevenson, and Letwin 2018). Johnson
et al. (2018) document that female entrepreneurs in Kickstarter are more likely to be funded than their male counterparts. Johnson
et al. (2018) attribute the success of female entrepreneurs in crowdfunding context to trustworthiness perceptions. Lin and
Pursiainen (2018) argue that male entrepreneurs tend to overestimate the demand for their products and therefore set a higher
goal amount, resulting in more failures.
Paper #890363
- 11 -
higher success rate. Finally, we include country specific annual Gross Domestic Product per
capita (GDP) and expect that funders’ funding decisions may be influenced by macroeconomic
conditions. Country-fixed and year-fixed effects are controlled to capture the country variate and
time variate turbulence that may affect crowdfunding outcome. We also add fixed effects to
control for sub-categories of the projects under the technology category because the information
asymmetry issue may be heterogeneous across different sub-categories of technology.16
All
continuous variables are winsorized at 1% level. We cluster standard errors by country and by
year.
3.4.2 Moderating effect of entrepreneurs’ gender
H2 predicts that the positive association between entrepreneur’s facial trustworthiness and
crowdfunding success is moderated by entrepreneur’s gender. We therefore estimate the
following equations to empirically test our prediction: