Success Factors in Reward-Based Crowdfunding in China and the US: A Comparative Study I. van Dijk
University of Amsterdam, Vrije Universiteit Amsterdam
11408529 / 2606070 Master Entrepreneurship
Thesis supervisor: dhr. L. (Liang) Zhao MSc Second reader: Dr. Tsvi Vinig
Table of Contents
Abstract ... 4
1. Introduction ... 6
2. Theoretical Framework ... 8
2.1 Crowdfunding and reward-based crowdfunding ... 9
2.2 Theories used in reward-based crowdfunding literature ... 10
2.2.1 Elaboration Likelihood Model. ... 10
2.2.2 Signaling Theory. ...11
2.2.3 Social Capital Theory. ... 12
2.2.4 Cultural differences and crowdfunding... 14
2.3 Hypotheses development ... 14
2.3.1 Collectivism of the Chinese culture. ... 14
2.3.2 Long term orientation of the Chinese culture. ... 15
2.3.3 Signals that affect the probability of success in Chinese reward-based crowdfunding. ... 17
3. Methodology ... 19
3.1 Research design and context ... 19
3.2 Data collection ... 19
3.3 Operationalization of variables ... 20
4. Results ... 22
4.1 Descriptive statistics ... 22
4.2 Hypothesis testing ... 23
4.2.1 Funding ratios in China and the US. ... 23
4.2.2 Regression analysis and hypotheses testing. ... 24
5. Discussion ... 26 5.1 Discussion of results ... 26 5.2 Theoretical contributions ... 29 5.3 Practical contributions ... 30 5.4 Limitations ... 31 5.5 Future research ... 32 6. Conclusion ... 33 References ... 34 Tables ... 37 Figures... 42
Abstract
Crowdfunding is an alternative form of financing that is rapidly gaining popularity. Unlike the crowdfunding landscape in the US, the Chinese crowdfunding landscape is not yet well
understood. This thesis aims to overcome this gap in the literature by investigating the factors of success of Chinese reward-based crowdfunding projects and comparing results to the US, based on Hofstede’s cultural six-dimensional model and signaling theory. Based on empirical evidence, this study shows that crowdfunding projects in China are generally not more successful than projects in the US. Results indicated that project duration, the number of backers, the funding goal, the number of updates, and the number of comments are positive indicators of project success in China. It was also suggested that the number of comments and updates are indicators of project success in both China and the US, and indicated signals that were different.
Statement of Originality
This document is written by Student Inigo van Dijk who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents
1. Introduction
Crowdfunding is an alternative financing approach for entrepreneurs who do not have access to traditional monetary resources such as loans from banks and venture capital.
Crowdfunding is an online aggregation of funds that helps organizations or individuals to bring their ideas to reality (Belleflamme et al., 2014). In general, crowdfunding involves raising funds from more than one participant, using a platform on the internet as a medium. Four different types of crowdfunding can be identified: rewards-based crowdfunding, lending-based crowdfunding, equity-based crowdfunding and donation-based crowdfunding (Kunz et al., 2016). In reward-based crowdfunding, participants (often called “backers”) finance projects to receive non-monetary returns, often in the form of products. This is the type of crowdfunding that is most widely known to the public, and is implemented by popular crowdfunding websites such as Kickstarter.com. Equity-based crowdfunding refers to financing of companies with capital investments by the crowd, similar to traditional equity investing. Collectively loaning money to project initiators is called lending-based crowdfunding. In donation-based
crowdfunding, participants give financial support to projects without expecting any return, as the name suggests. This study focuses on reward-based crowdfunding in China.
Even though reward-based crowdfunding is popular, success rates average below 40% (Kickstarter.com/help/stats, 2017). This makes the topic interesting to study, and it is one of the reasons for its growing attention in the literature. However, the scholarly knowledge on
crowdfunding remains quite limited (Short et al., 2017). Currently, the crowdfunding landscape in China is relatively new. To compare, Indiegogo, the first reward-based crowdfunding platform in the US, launched in 2008, where crowdfunding emerged in 2011 in China with the platform Demohour. However, the Chinese reward-based crowdfunding landscape is expected to be the
largest in 2017 with an estimated transactional value of over $5.5b, compared to almost $1b in the US (Statista.com, 2017). The Chinese crowdfunding landscape is also the fastest growing in the world, with an annual growth rate (CAGR 2017-2021) of 31.4% (Statista.com, 2017). This makes the Chinese reward-based crowdfunding landscape particularly interesting to study.
The scholarly knowledge utilizing data from platforms in the US is accumulating quickly, with many studies using data from Indiegogo or Kickstarter, the two largest reward-based
crowdfunding platforms in the US. The theoretical backing of the Chinese crowdfunding
landscape however, is lagging behind. This study aims to overcome this gap by first studying the impact of cultural differences on crowdfunding performance, and second studying success factors for reward-based crowdfunding projects hosted by the Chinese platform Zhongchou.com. Besides being theoretical relevant, this study is also of practical relevance because a better understanding of the determinants of reward-based crowdfunding success will allow project initiators to improve their chances of success.
The study is divided into two parts. First, it investigates the effects of cross-cultural differences on crowdfunding performance using Hofstede’s six cultural dimensions model (Hofstede, 1984; Hofstede, 1991). Hofstede scored many countries on his dimensional model and found that China scored very high on collectivism and long-term orientation, and the US scored much lower. This and the aforementioned are the reasons that this study will investigate differences between China and the US. The first part of the study will quantitatively test hypotheses linking collectivism and long-term orientation to crowdfunding performance.
The second part of this study attempts to replicate and extend a study by Kunz et al. (2017), who empirically investigate signaling in reward-based crowdfunding using a Kickstarter dataset. The authors developed a classification of quantifiable signals for reward-based
crowdfunding, which will be used in this thesis to test hypotheses on a Chinese dataset. This will increase the knowledge of the Chinese reward-based crowdfunding landscape. Hypotheses will be tested using binary logistic regression.
The research questions which will be answered by testing the aforementioned hypothesis are:
1. What impact does collectivism and long-term orientation of the Chinese culture have on crowdfunding performance?
2. What signals in Chinese reward-based crowdfunding can be used to predict crowdfunding performance? How are these signals different in China and the US? The rest of this thesis is structured as follows. A theoretical framework will be presented and hypotheses will be developed in section 2. Section 3 introduces the research methods used and section results will be reported in section 4. Section 5 tries to answer the research question, states the theoretical and practical contributions, discusses limitations, and states avenues for future research. Finally, this thesis is concluded in section 6.
2. Theoretical Framework
The theoretical framework will be discussed in the following manner. First, the concepts crowdfunding and reward-based crowdfunding will be defined and explained. Next, a literature review of reward-based crowdfunding follows, grouped by the theoretical models that back the studies. These models are often used in the reward-based crowdfunding literature and are therefore briefly discussed. Finally, based on Hofstede’s cultural dimensions model and signalling theory, hypotheses will be developed.
2.1 Crowdfunding and reward-based crowdfunding
Crowdfunding adds a new form of financing to traditional financing such as loans from banks, venture capital financing, and financial bootstrapping. Crowdfunding can be used to raise funds for launching a specific product, service, or initiative. Categories can include comics, dance, theatre, film, music, video, social causes such as education, environment and health related, and categories for commercial products or services such as sports, gaming, and technology. Besides allowing the initiator to raise funds, crowdfunding also allows to test the market for a product, build an early follower base, get valuable feedback, and pre-sell a product (Steinberg, 2012). Raised funds can be much higher than initially planned. In reward-based crowdfunding, if the project fails, nothing is lost besides time and public image. However, preparing a crowdfunding project requires a different skillset than traditional product pitches. Crowdfunding project initiators will mostly be doing marketing efforts. Another disadvantage of crowdfunding can be that there is a large chance of failing. Kickstarter projects have a success rate of only 37% (Kickstarter.com/help/stats, 2017).
A reward-based crowdfunding project contains generic information such as a title, description, images and possibly a video. Here the initiator tries to pitch the project to possible backers. The funding goal and progress is displayed, as well as the number of backers. Projects are posted in a certain category and often run a limited amount of days. Kickstarter and
Indiegogo have limited the duration to 60 days, where Zhongchou does not limit the duration of their projects. Project initiators can post updates to interact with the community. Often, social media statistics such as Facebook likes and shares are displayed. People can leave comments on the project page. The different levels of rewards are listed, starting with an option to choose no reward. The price increases as the size of the rewards increases. Initiators can limit rewards to
make them exclusive. Often, the funds of a project are released to the initiators only if they reach their goal. However, initiators can choose this setting. Interestingly, Zhongchou offers the option to ‘sell’ lottery tickets, where the reward goes to a random backer, every x number of tickets sold.
2.2 Theories used in reward-based crowdfunding literature 2.2.1 Elaboration Likelihood Model.
The first theory that is frequently used in reward-based crowdfunding literature is the Elaboration Likelihood Model (ELM). Since reward-based crowdfunding has similar
characteristics to e-commerce (Kunz et al., 2017), and investigating e-commerce concerns investigating consumer behavior, ELM can be used to explain backer behavior in reward-based crowdfunding. Several studies used Petty and Cacioppo’s (1986) ELM to explain backer
behavior. ELM proposes two distinct persuasion routes for evaluative processing: the central and the peripheral routes. The central route is used when a person has carefully evaluated the
information before making a decision. A person is focused and actively processes this central information. The peripheral route is used when the person has little motivation to process the information. This information is often in the smaller details. An example can be given using an advertisement: central information can be the product and price of the advertisement, and peripheral information can be the person who presents it or the color of the product.
Dey et al. (2017) link product related factors such as relevance and complexity to central cues, and video related factors such as perception of duration and audio/video quality to
peripheral cues. The authors adopted ELM to understand whether central and peripheral cues can explain the varying perceptions of backers while evaluating campaign videos on Kickstarter, for which they find support. Zheng et al. (2016) use the ELM to investigate the role of trust
management on fundraising performance in reward-based crowdfunding on Chinese data collected from Demohour.com. Results indicated trust management significantly promoted fundraising performance via central (entrepreneur’s creditworthiness) and peripheral
(entrepreneur-sponsor interactions) routes. Bi et al. (2017) also use the ELM to investigate the influence of online information on investing decisions of reward-based crowdfunding. Signals of project quality were linked to the central route and electronic word of mouth was linked to the peripheral route. The authors found that both routes have almost equal effects on funder investment decisions. The central route was significantly more important for the
Science&Technology and Agriculture projects, whereas the peripheral route was more important for Entertainment and Art projects.
2.2.2 Signaling Theory.
Signaling theory is useful for describing behavior where there is an information
asymmetry: when two parties have access to different information (Connelly et al., 2011). The sender must choose how to signal information, and the receiver must choose how to interpret the information. An e-commerce environment results in information asymmetry because buyers cannot physically evaluate the quality of products and easily assess the trustworthiness of sellers (Mavlanova, Benbunan-Fich & Koufaris, 2012). As stated earlier, reward-based crowdfunding and e-commerce have similar characteristics. In crowdfunding, signaling refers to carrying information from project initiators to backers. One of the problems of information asymmetry is adverse selection: the misrepresentation of the project initiators true characteristics(Mavlanova et al., 2012). This problem can be overcome by signaling high quality information, such as a
detailed description or video about the project and its initiator. In their framework, Mavlanova et al. (2012) describe three dimensions of website signals: the purchase time continuum
(pre-purchase, during purchase and post-purchase), the ease of verification (easy and difficult), and signaling cost (low and high cost).
Kunz et al. (2016) base their research on two out of three dimensions of website signals: purchase time continuum and signaling cost. The authors investigate different signals such as runtime, reward limit, and update count as success factors of project performance. Their results indicate that social ties, investment preparation and presentation, the supply of multiple rewards, and communicating with the crowd positively influences the chance of success. They also find that the funding goal, the project’s runtime, and the estimated time of delivery have a negative effect on the chance of success.
The literature covering signaling and backer behavior in reward-based crowdfunding is thin. I believe extending this knowledge is important for two reasons. First, much can be learned from the area of e-commerce, where signaling theory is important because of the similarities between e-commerce and reward-based crowdfunding. Second, signaling theory can help overcome problems associated with information asymmetry, as described above. This thesis will extend the knowledge of signaling and backer behavior by replicating the study by Kunz et al. (2016). More on this in section 2.4.
2.2.3 Social Capital Theory.
Social capital theory is multidimensional (Nahapiet & Ghoshal, 1998). Nahapiet and Ghoshal (1998) define social capital as “The sum of the actual and potential resources embedded within, available through, and derived from the network of relationships possessed by an
individual or social unit.”. Most scholars of management studies adopt this definition. The authors define three dimensions of social capital: structural, relational, and cognitive. The structural dimension refers to the configurations of ties between parties in a network and their
relationships. Development of network ties are necessary for the development and utilization of social capital resources. The relational dimension describes the type of relationship people have developed over time (Granovetter, 1992). It concerns the quality of the relationship like the obligations, expectations and trustworthiness of the social network (Zheng et al., 2014). The cognitive dimension implies that shared values and language can strengthen bonds in a social network. Interactions within its online community lie at the core of a crowdfunding platform (Skirnevskiy et al., 2017). Relationships can be strengthened by interactions, and as social capital is derived from the network of relationships (Nahapiet & Ghoshal ,1998), social capital theory can be applied to reward-based crowdfunding platforms.
Zheng et al. (2014) applied social capital theory on reward-based crowdfunding. They argue that entrepreneurs can leverage their social network ties (the structural dimension) to improve crowdfunding performance, for which they find support. On the relational dimension, the authors argue that the obligation to fund other entrepreneurs is positively associated with crowdfunding performance. They also find support for this hypothesis. They link shared meaning measured as the length of the project description to the cognitive dimension, and find support that shared meaning is positively associated with crowdfunding performance.
In their study, Skirnevskiy et al. (2017) leverage social capital theory to frame how internal social capital can develop through project track record. They also investigate how there can be a spillover from internal social capital to external online communities. The authors empirically assess whether long-term implications of social capital can increase funding success beyond a single campaign. They find support using both quantitative data analysis and
2.2.4 Cultural differences and crowdfunding
Finally, several studies investigate the effect of cultural differences on crowdfunding. Dushnitsky et al. (2016) investigate the variations across the different crowdfunding models of European countries. Burch, Ghose, and Wattal (2013) present evidence that lenders prefer culturally similar and geographically proximate borrowers in lending-based crowdfunding. The impact of social capital on crowdfunding can differ between cultures (Zheng et al., 2014). The authors in this study link the Chinese collective culture to reward-based crowdfunding. They suggest that the fundamental characteristics of crowdfunding could make it easier for Chinese entrepreneurs to find backers for their projects. The study hypothesizes that national culture moderates the effects of social capital and obligation. Comparing the effects between China and the US, support was found.
Of the abovementioned theories, signaling theory and cultural differences will be used for hypothesis development. To investigate cultural differences, Hofstede’s six cultural dimensions model will be used.
2.3 Hypotheses development
2.3.1 Collectivism of the Chinese culture.
In research on cross-cultural differences, Hofstede formed a model with six cultural dimensions: Power Distance, Uncertainty Avoidance, Individualism vs Collectivism and Masulinity vs Feminimity, Long Term Orientation and Indulgence (Hofstede, 1984; Hofstede, 1991). Hofstede scored many countries, among which China and the US, on each of these dimensions. A comparison of cultural dimensions of China and the US can be found in Figure 1. The two largest differences between these two countries are on individualism and long-term orientation. China has a much more collectivist culture and is more long-term oriented. The first
part of this thesis investigates whether these two differences could influence crowdfunding in China, and are now discussed in more detail
Chinese collectivism is sometimes described as guanxi. Guanxi literally means social connection. It is a synonym for special favors and obligations to the guanxi circle (Lee & Dawes, 2005). Yang (1994) describes guanxi as being implicitly based on mutual interests and benefits. Schwarz (1990) describes collective types as prosocial, favoring restrictive conformity, security, and tradition. Collectivistic people value that which benefits the group. It demands that the group is more important than the individual. One could intuitively argue that collectivism and
crowdfunding go hand in hand, similar to the suggestions of Zheng et al. (2014). Crowdfunding initiators ask for a collective trust and financial support from a large group, the crowd. Backer’s choice to participate in a crowdfunding project could thus be influenced by how collectivistic their culture is. This could therefore mean that the concept of crowdfunding is more successful in China. Thus, I argue:
H1. The success rate of reward-based crowdfunding projects in China is larger than
success rates of crowdfunding projects in the US.
2.3.2 Long term orientation of the Chinese culture.
In their study “long- versus short-term orientation: new perspectives”, Hofstede & Minkov (2010) show new support for Hofstede’s (1991) fifth dimension of national cultures: long-term orientation. The authors state that “Children [of cultures with a long-term orientation] learn thrift, not expecting immediate gratification of their desires, tenacity in the pursuit of whatever goals, and humility. Self-assertion is not encouraged. Old age is seen as a happy period, and it starts early ...” (Hofstede & Minkov, 2010, p. 496-497). In business, people invest in lifelong personal networks based on guanxi. The main work values are learning, honesty,
adaptiveness, accountability, and self-discipline. The authors sum up the values of long-term orientation as thrift, hard work, and persistence.
The cultural values of thrift, not expecting immediate gratification of desires, could have an influence on both crowdfunding backer and initiator behavior. As stated before, most
crowdfunding projects only receive their funding and are able to return rewards to the backers if the project is successful. Shorter projects are generally found to be more successful in
Kickstarter data (e.g. Kuppuswamy & Bayus, 2015). Contrastingly, Cordova, Dolci, and Gianfrate (2015) find that project duration increases the chance of success in technology projects. An explanation for the success of shorter duration projects could be that shorter duration signals legitimacy by setting modest, achievable expectations (Frydrych, Bock & Kinder, 2014). One could argue the need for immediate gratification could also influence the preference for, and success of, shorter duration projects. In long term oriented cultures, such as in China, this effect could be in the opposite direction. Thrift could result in a tolerance of projects with a longer duration, for both backers and initiators. This could be a reason for
Zhongchou.com not to incorporate the 60-day duration maximum of Kickstarter and Indiegogo. This study will test this assumption by comparing the duration of successful and unsuccessful projects from Zhongchou.com with data from Kickstarter.com, using results from the regression analysis as described in the next paragraph. The following is hypothesized:
H2. Longer duration is a predictor of success in Chinese reward-based crowdfunding
2.3.3 Signals that affect the probability of success in Chinese reward-based crowdfunding.
The second part of this thesis focuses on replicating and extending the study “An empirical investigation of signaling in reward-based crowdfunding” by Kunz et al. (2016), as discussed previously. The authors apply signaling theory to test how several types of signals affect the probability of success of a reward-based crowdfunding project, tested on
Kickstarter.com data. Please refer to section 2.2.2 of this thesis and the original study of Kunz et al. (2016) for theoretical backing of using signaling theory and theoretical backing for choosing each of the signals. The authors test the effects of the following signals: RunTime, RewardCount (amount of possible reward levels), RewardLimit, FacebookFriends (of the project initiator), BackedCount, AvgDelDateDiffGew (delivery period of rewards), DescriptionWordCount,
ImageCount, VideoCount, HpValue (having a homepage), Preptime (preparation time), FaqCount (the amount of frequently asked questions), UpdateCount, FacebookBuzz (amount of Facebook likes and shares) and StaffPicked (featured on the homepage). The study controlled for
FundingGoal, PledgedMoney and FundingRatio. The signals are grouped on two dimensions of the framework by Mavlanova et al. (2012): purchase time continuum and signaling costs. For the classification of signals by Kunz and his colleagues, please refer to Table 1 of their article.
This study will try to replicate their findings with the following signals: duration, rewardAmount, targetAmount, updateAmount, likeAmount, commentAmount and
backerAmount. LikeAmount is to replace FacebookBuzz. FacebookBuzz consists of the amount of likes and shares and can be seen as a positive indicator of feedback (Kunz et al., 2016). LikeAmount has a similar effect, and will therefore be used instead of FacebookBuzz. It should be noted that the effect of likeAmount will be smaller than that of FacebookBuzz, as no ‘shares’
are incorporated, only likes. The funding goal was measured by Kunz et al. (2016), but no hypothesis was tested. The results of their regression analysis for this signal can still be used for comparison, therefore the funding goal (targetAmount) is included as an independent variable in this study. The number of comments was not measured or tested by Kunz et al. (2016). However, it is also a during-campaign signal as that can be a success predictor like the number of updates and the number of likes. Therefore, it was also included in this study. Classifications among the dimensions of the framework by Mavlanova et al. (2012) can be found in Table 1. Runtime, the number of rewards and the funding goal are chosen by the project initiator before the project starts (ex-ante). Runtime is considered a high cost signal as more effort is required the longer a project is running. Designing a larger number of rewards also costs significant effort. The
funding goal is considered high cost as well because it requires effort to calculate the exact funds that need to be raised. During-funding phase signals updates, likes and comments are considered high costs because the initiators have no control over them. This is also the case for the number of backers, which is included in this study as a post-funding phase signal. Kunz et al. (2016) do not include post-funding phase signals. The number of backers as a signal of success could intuitively be argued, however, this assumes that there is no small group of large investors. This could be interesting to test; thus, I chose to include the number of backers as a signal in this study.
Hypotheses 3, 4, and 6 were taken from Kunz et al. (2016). Hypotheses 5, 7, and 8 were developed to extend their study.
H3. As the number of available reward levels increases, the probability of project success
H4. As the number of updates issued by the project initiator increases, the probability of
project success increases.
H5. As the number of likes increases, the probability of project success increases. H6. As the number of comments increases, the probability of project success increases. H7. As the number of backers increases, the probability of project success increases. H8. As the funding goal increases, the probability of project success increases.
3. Methodology 3.1 Research design and context
Even though the scholarly field of crowdfunding is novel, several theories have been formed and tested. The investigation of success factors of reward-based crowdfunding projects is becoming more mature, as is discussed in the theoretical background section of this thesis. Edmondson and McManus (2007) define this stage as an “intermediate stage of development” (pp. 1159). According to them, intermediate theory can benefit from both qualitative and
quantitative analysis. This thesis attempts to aid the maturation process by quantitatively testing several existing theories of the success factors of reward-based crowdfunding. Part 1 of this study uses existing theories to form new hypotheses (H1 and H2). Part 2 of this study will test both existing and new hypotheses on a Chinese dataset. Student’s t-test and binary logistic regression will be used to analyse the data. The research project is situated in the Chinese reward-based crowdfunding landscape, in particular the Zhongchou.com platform.
3.2 Data collection
Data was publicly available from the Zhongchou.com website at time of writing. A web crawler was used to collect project, reward, and support information. The crawler was written using the NodeJS programming language and data was stored in a MySQL database. The
Zhongchou.com website is structured as follows: every project has it’s own webpage link, in the form of http://www.zhongchou.com/deal-show/id-14630. The number at the end of the link
represents the identification number for that project on the website, and can be iterated on to find the projects. The crawler searched for projects from id 0 to id 999999. The highest id that was found was 713735. A total of 2598 projects were found, with 11,223 different reward options and a total of 242,679 support records.
For each project, the following data was collected: title, description, status, amount raised, number of backers, target amount, duration, and the number of likes, comments, updates, and rewards. It must be noted that the start and end date of the project were not available on the project page. Start date was estimated by taking the date of the first support record, or the date of collection if there were no support records. The time remaining of projects in progress could be retrieved. End date was estimated by taking the date of the last support record, or by adding the time remaining to the start date, if that resulted in a later date. Duration was then calculated as the difference in days between the start and the end date. This unfortunately limits the accuracy of the data set. The reward options of each project were saved, with their price and the number of backers that chose the reward. Finally, support records were saved, with a username, price, the type of reward, and a timestamp.
3.3 Operationalization of variables
The first part of this thesis investigates the cultural effects on crowdfunding performance. The dependent variable, crowdfunding performance, was measured using the funding ratio (fundingRatio). The funding ratio was measured as the percentage of final funding over the target amount of funding. Since projects that are in progress have incomplete funding ratios, 655
projects were excluded from the dataset, resulting in a total of 1944 completed projects. The duration of a project was measured in days, and calculated as described in the previous section.
The second part of this thesis is a replication of the study by Kunz et al. (2016) as
described in section 2.3. The dependent variable project success is measured as a binary variable, where success = 1 and failure = 0. The dependent variables are measured as follows. The number of available rewards is simply the count of reward levels for a single project (rewardAmount). Project duration is measured in the same way as described above. Initially, the scope of the description of a project was to be measured as the amount of words in the project description. However, many crowdfunding initiators chose to post the description of their projects as images. This would have resulted in over 50% of the project being excluded for analysis. Therefore, the description word count was not included in this study. UpdateAmount was measured as the number of updates the initiator published from the start of the project until the moment of measurement. The numbers of likes by members of the crowdfunding platform was measured (likeAmount), as was the number of comments (commentAmount).
3.4 Data analysis
The first hypothesis was tested by comparing the funding ratio mean of the Kickstarter.com dataset (Frydrich et al., 2014) with the funding ratio mean of the
Zhongchou.com dataset. This was done using a two-sample t statistic, which is considered appropriate when comparing two sample means (Field, 2005). Hypothesis 2 measured the effect
of duration on crowdfunding success. The coefficient for duration in the regression of part 2 of this thesis will be compared to the coefficient of Runtime in the study by Kunz et al. (2016). This will indicate a difference between the Kickstarter and Zhongchou dataset regarding the effect of duration on project success.
As the dependent variable of part 2 of this thesis is a binary variable (success), binary logistic regression will be used, in line with the original study. The regressions coefficients indicate the direct of influence between the dependent variable and independent variables, and will be used for testing hypotheses. The data was prepared for binary logistic regression. To remove outliers, projects with a funding target lower than 100$ (680 CNY) and higher than $1,000,000 (6,800,000 CNY) were excluded. This resulted in 148 projects with a low funding target and 3 projects with a high target being excluded. Histogram analysis showed no skewed data of the variables, therefore no logarithmic transformation was required (Benoit et al., 2011). The study included the target funding amount (targetAmount) in the regression analysis as an independent variable. Quantitative analysis was performed using SPSS version 23.
4. Results 4.1 Descriptive statistics
Descriptive statistics of the Zhongchou dataset are displayed in tables 2 to 5. The amount of funding raised and the target amount were also included in $, using a conversion rate of ¥1 = $0.147067 (retrieved at 28-06-2017). There were 2599 projects in total, of which 751 failed, 1193 succeeded and 655 were still in progress. The success rate of completed projects was a surprising 61.36%. The average funding ratios were 75.8%, 0.7%, 159.2% and 9.8% for total, failed, successful and projects in progress, respectively. The highest target amount was ¥5.2m or $764k. The lowest target amount was ¥720 or $105.89. The highest raised amount was ¥6.2m or $914k. The highest number of backers was 39,561 for all and successful projects, 402 for failed projects, and 584 for projects in progress. The highest number of likes was 4489, the highest number of possible rewards was 48, the highest number of comments was 1020, and the highest number of updates was 245. There were two projects who were successful with only 2 backers,
who raised ¥6100 or $897 and ¥1000 or $147 in 36 and 19 days. The longest project duration was 287 days for all and failed projects, 142 for successful projects and 162 for projects still in progress. For more descriptive statistics, please refer to tables 2 to 5.
4.2 Hypothesis testing
4.2.1 Funding ratios in China and the US.
H1 investigates the difference in success rates of projects in China and the US. This is measured using the funding ratios. The funding ratios for the Kickstarter sample were 77.18%, 133.31%, and 12.03% for all, successful and failed projects, respectively. For the Zhongchou sample, the funding ratios were 75.76%, 159.19% and 0.7%. An independent-samples t-test was conducted (alpha = 0.05) to compare the funding ratios of crowdfunding projects from
Kickstarter.com and Zhongchou.com, for every category (all, successful, failed). For successful projects, there was a significant difference in funding ratios for Kickstarter (M=133.31%, SD=79.85%) and Zhongchou (M=159.19%, SD=164.63%); t(976) = 2.2879, p = 0.0224. These results suggest that projects from Zhonchou.com that reach success, are more successful than successful projects from Kickstarter.com. For all projects, there was no significant difference found between Kickstarter (M=77.18%, SD=85.19%) and Zhongchou (M=75.76%,
SD=137.13%); t(2823) = 0.1534, p = 0.8781. For the projects that did not succeed, there was a significant difference found between Kickstarter (M=12.03%, SD=17.02%) and Zhongchou (M=0.71%, SD=7.51%); t(926) = 13.498, p = 0.000). This indicates that Kickstarter projects who fail, generally perform better than Zhongchou projects that fail. In other words, Zhongchou projects fail harder than Kickstarter projects.
To summarize, Zhongchou projects that succeed are more successful and Kickstarter projects, Zhongchou projects that fail, fail harder and over the entire sample, there is no
significant difference in funding ratios. Overall, Zhongchou projects succeed 61.36% and Kickstarter projects succeed 53.9%. The authors of the Kickstarter study (Frydrich et al., 2014) note that the the success rate they find is 10% higher than the average of Kickstarter, which is available on the Kickstarter website (www.kickstarter.com/help/stats). At the time of writing,
Kickstarter projects succeed only 35.85%. It must be noted that the sample size of the Kickstarter data collected by Frydrich et al. (2014) is relatively small with a total of 421 projects. This could mean that their sample is biased towards successful projects, or that the Kickstarter success rate has fallen over time. There is some indication that Zhongchou projects are more successful, looking at the funding ratios of succeeded projects and the overall success rate. However, the funding ratios of Zhongchou projects did not score better when measuring all and failed projects. H1 can only be accepted if for all categories (all projects, failed projects, and successful projects) the means of funding ratios are significantly higher. This was not the case. Therefore, H1 is rejected.
4.2.2 Regression analysis and hypotheses testing.
For H2 to H7, a binary logistic regression analysis was performed. Regression results are displayed in table 6. All variables except rewardAmount were significant at the level of α = 0.05. It must be noted that likeAmount is barely significant at the α = 0.05 level. The B represents the regression coefficient and indicate the direction of influence between the dependent and
independent variables. These coefficients can be used as a method for hypothesis testing (Kunz et al., 2016). Duration, updateAmount, commentAmount, backerAmount and targetAmount all had a significant positive effect on project success. LikeAmount had a significant negative effect on project success. RewardAmount had no significant effect on project success.
In table 7, hypothesis 2 to 7 are listed, each with a signal as independent variable. Each hypothesis has an expected direction of effect. Using the binary logistic results, the hypotheses are tested. The outcomes of hypotheses are also listed in table 7. The direction of effect for rewardAmount is not important, since there was no significant effect found. This is marked by putting the direction of effect in parentheses.
Confidence intervals can be used to assess the direction of influence of the effect
coefficients (Exp(B)). Confidence intervals closer to the effect coefficients are generally seen as stronger (Burns & Burns, 2008). Looking at table 6, only the confidence intervals of
updateAmount are relatively widespread (+-20%).
Logistic regressions can suffer from multicollinearity (Farrar & Glauber, 1967). In line with
testing for multicollinearity by Kunz et al., (2016), standard errors and a correlation matrix were used. Standard errors were < 1 and the highest correlation between independent variables was -0.48, between likeAmount and commentAmount. This is below 0.7 that the original study used as a ceiling value.
The model Chi-square is used to test the goodness of fit of the model (Burns & Burns, 2008). H0 in this test says the model is a good fitting model, and H1 says the model is not a good fitting model, which means the predictors have a significant effect. The null hypothesis is
rejected with a Chi-square of 2289.158, df = 7, p < 0.01. Therefore, the model predictors have a significant effect. A Nagelkerke R-squared of 0.940 indicates a strong relationship of 94% between the predictors and the prediction.
Hosmer and Lemeshow is an alternative to model chi-square and is used as a goodness-of-fit test (Burns & Burns, 2008). Well-fitting models should reject the null hypothesis that there is no difference between observed and model-predicted values. This study did not pass the
Hosmer and Lemeshow test with a chi-square of 1.759 * 10^13, 8 degrees of freedom and a significance of p < 0.01. However, concerns with using this test have been raised when using larger samples. For example, Allison (2013) voiced concerns about the way the test relies on choosing the right amount of groups, and sometimes for larger samples the right number can be hard to find. As this regression passes all other tests discussed, failing the Hosmer and
Lemeshow test will not invalidate the results, but it should be noted.
5. Discussion 5.1 Discussion of results
The research question of the first part of this thesis was “What impact does collectivism and long-term orientation of the Chinese culture have on crowdfunding performance?”. It attempted to answer the first sub question regarding collectivism by investigating the difference of success rates between projects in China and the US.
This study does not find compelling evidence that Chinese reward-based crowdfunding projects have more success than those in the US (H1). The average success rate of all projects from Zhongchou (61.36%) was higher than the compared average using Kickstarter data
(53.9%). However, when comparing the funding ratios for each category (all, successful, failed), the projects from Zhongchou did not score significantly higher. As stated in the results section for this hypothesis, this could be due to a small sample size of the study that is used as
comparison. Additional research is required using larger samples, which may show clearer results.
Results for Zhongchou projects showed that projects that are successful, are more successful compared to Kickstarter projects. Kappuswamy et al. (2017) find that crowdfunding contributions will significantly decrease after the target goal is reached. Findings of my study
suggest that this effect might be smaller in China than in the US. This should be tested in future studies to confirm. Results also showed that projects that fail, fail harder at Zhongchou compared to Kickstarter. This effect has been studied by the literature and is sometimes called the
“Kickstarter effect”: the closer a project gets to reaching its target, the more support it rallies (Kappuswamy & Bayus, 2015). Results of this study indicate that the “Kickstarter effect” could be stronger in collectivist societies.
The second sub question regarding long-term orientation was attempted to be answered by investigating the difference between the relationship of duration and project success in China and the US.
Results of the regression analysis indicated support that a longer duration is an indicator of success for crowdfunding projects in China. This contradicts findings of Bi et al. (2017), whose regression results indicate the opposite direction of the effect. This difference was not expected because both studies use the same data source. This difference could be accounted to the difference in operationalization of the dependent variable. Bi et al. (2017) used the number of backer as measurement for project success, and this study used the binary variable “success”. The authors also used a much smaller sample size, almost a third compared to this study. A more probable explanation is the fact that the authors “collect each project’s information on the day of it’s deadline” (Bi et al., 2017, p. 4). This study focuses on the entire runtime of a project, also after the target amount was reached. This could be the reason for the difference in results and should be noted. In line with results from this study, Zheng et al. (2014) found a positive significant relationship between duration and success in China, and found that the direction of this relationship is negative in the US. Kunz et al. (2016) similarly find that there is a negative relationship between duration and project success in the US. As described in section 2.3,
successful US projects are generally of shorter duration. Combining these findings with results from this study, one could argue that cultural differences (i.e. long-term orientation) could affect the relationship between duration and project success. However, to the ambiguous results of this study and that of Bi et al. (2017), I suggest that more research is required to clarify the direction and strength of the relationship between duration and success.
The second part of this thesis investigated “What signals in Chinese reward-based crowdfunding can be used to predict crowdfunding performance?” and how these signals are different compared to projects in the US. Signals were categorized over the purchase time continuum and signaling costs of the framework of Mavlanova et al. (2012). The pre-funding phase (ex-ante) signals duration and funding goal were significant positive predictors of project success. The number of rewards was no significant predictor of success. In the study by Kunz et al. (2016), the authors find different results for Kickstarter data. Project duration and funding goal were negative significant predictors of success, and the number of rewards was a positive significant predictor. This comparison suggests that there exist differences between Chinese and US projects.
In the funding phase, the number of updates was a significant positive predictor of project success, in line with the findings of Kunz et al. (2016). However, the strength of the prediction was stronger in the Chinese dataset. This indicates that the number of updates is a stronger prediction in China than in the US. The number of likes was a significant negative predictor of project success, unlike Kunz and his colleagues find. These results suggest that the number of likes is less important in China than it is in the US. However, it must be noted that this difference could be due to the different operationalization of variables. Kunz et al. (2016) also took into
account Facebook shares. The number of comments was a significant positive predictor of success in China.
In the post-funding phase, results indicated that the number of backers was a significant positive predictor of project success. This is in line with the results of Kunz et al. (2016), and suggests that the number of backers is an indicator of success for both China and the US. This also suggests that there are no small groups of large investors in Chinese crowdfunding projects, which should have affected the results.
The results discussed above suggest that there are numerous differences between Chinese and US reward-based crowdfunding projects. Results also uncovered similarities: both the number of updates and the number of backers is a signal of success in both countries. Could these signals be universal signals of success for reward-based crowdfunding project? Future research could investigate this question, using this study as a starting point.
5.2 Theoretical contributions
Findings of this study have several theoretical contributions. The study showed that the Chinese crowdfunding landscape is to be the largest in terms of transaction value and yearly growth. Following this I noticed that scholarly literature on crowdfunding in China is still in a nascent stage, and suggested that it should deserve more attention. Further, it was pointed out that most reward-based crowdfunding studies use Kickstarter or Indiegogo as a data source. Results of this study expanded the existing knowledge base on Chinese reward-based
crowdfunding in two parts. The first part of this study attempted to link Hofstede’s collectivism and long-term orientation to success factors in reward-based crowdfunding in China. This was a new approach in investigating success factors. It showed that cultural differences can affect the
concepts of reward-based crowdfunding, and allows for future research. The second part of this study noted that signaling theory can be important for understanding backer behavior. The study used a Chinese dataset to retested a study by Kunz et al. (2016) that utilized signal theory to predict project success in the US. Differences and similarities between signals of success for two different cultures were described. This opened up new fields of research that have not yet been properly addressed.
5.3 Practical contributions
This study also provides several practical contributions for crowdfunding platform creators and project initiators. By showing room for improvement of success rates of projects in both China and the US, this study signals platform creators to take action aiding project initiators to become successful. It showed that signals that can predict project success in the US, might not be applicable in China, and the other way around. Some crowdfunding platforms in China seem to straight up copy the platform structure of Kickstarter and Indiegogo. Chinese platform creators should be aware that they have to deal with possible cultural differences and structure their platform accordingly. This study suggested that projects that fail in China, fail harder. The exact reason for failure should be investigated by platform creators and acted upon. Positive
determinants of success were found for Chinese projects: duration, the number of rewards, the number of updates, the number of comments, the number of backers and the funding goal. Project initiators can leverage this information to improve their chances of success. Long term orientation could mean that Chinese backers tolerate longer duration of projects, which would suggest that project initiators should not be scared to choose a longer duration for their projects. Project initiators should encourage friends, family, and (potential) backers to leave comments on their project page, which could increase their chances of success. Project initiators that give
regular updates are associated with a higher probability of success. Project initiators should not be limited by the height of their funding target. Finally, project initiators should not spend too much time gathering likes on their projects, results of this study suggest.
5.4 Limitations
This empirical study is subject to several limitations. The dataset that was used is only a subset of the entire set of projects of Zhongchou. The study did not discriminate between project categories. Results could differ between project categories. The scope of this study was limited to reward-based crowdfunding. Results are not directly applicable to donation-based, equity-based, or lending-based crowdfunding. The first part of this study compared data that might not be representative for Kickstarter projects. The low sample size and difference in success rate of the sample and the reported success rate of Kickstarter as seen on their website suggests the aforementioned. As Kunz et al. (2016) discuss, the regression estimation model only takes into account a limited amount of variables. There could be a lot more variables that have significant effect on project success such as project quality and initiator experience. These variables were not measured in this study. The regression results failed the Hosmer-Lemeshow test, even though the test’s accuracy is sometimes doubted. Please refer to section 4.2.2 for more information on this issue. Another limitation of this study is the way the duration of a project was measured. This was estimated as described in section 3.3. The exact date could not be measured as the start and end dates are not displayed on the project page. This could be avoided by measuring new projects as they appear and collecting information when they end, in line with the methods of Bi et al. (2017). However, this would have resulted in a significantly smaller dataset as the time frame for this study was limited. Future studies are recommended to use a more accurate
approach than used in this study for more accurate results. Finally, the dependent variable “success” of the regression analysis was binary coded. Binary variables contain less information than continuous variables. A continuous variable such as funding ratio (as used in H1) could have been used for better predictions.
5.5 Future research
Several directions for future research are opened by this study. It makes a start
investigating the differences of determinants of success in reward-based crowdfunding projects between different cultures. Future studies could build on and further test the effects of
collectivism, long term orientation, and other cultural differences as discovered by Hofstede (1981; 1991; 2010) on crowdfunding success. Results showing that Chinese crowdfunding projects succeed and fail more extremely than their US counterparts raised the question “what could have caused this?”. This study suggested that cultural collectivity could be of influence. Future research could test this assumption and uncover more interesting results concerning Chinese crowdfunding projects. As suggested by this study, there should be a shift of focus to the Chinese crowdfunding landscape, as it is relatively new but growing rapidly, but scholarly knowledge is lagging behind. Additional research is required to clarify the relationship between project duration and success, as ambiguous results were discussed. Finally, this study presented a start for studying universal factors of success in reward-based crowdfunding. Future studies could build on this knowledge and investigate success factors in different cultures and countries besides China and the US.
6. Conclusion
Crowdfunding has evolved over the years as a successful alternative for raising capital. Reward-based crowdfunding in China has grown to the largest of the world and is still rapidly growing. However, less is known about the crowdfunding landscape in China, compared to the US. This thesis attempted to close that gap. Links were made between cultural differences and crowdfunding performance. Support was found by comparing US data to Chinese data using quantitative analysis. Several success factors of success for projects in China were found and compared to the success factors in the US. This study made progress in understanding the crowdfunding landscape of China. However, there is still much to be learned in this area of research.
References
Allison, P. (2013, March 5). Hosmer-Lemeshow Test for Logistic Regression. Retrieved June 23, 2017, from https://statisticalhorizons.com/hosmer-lemeshow
Belleflamme, P., Lambert, T., & Schwienbacher, A. (2014). Crowdfunding: Tapping the right crowd. Journal of business venturing, 29(5), 585-609.
Benoit, K. (2011). Linear regression models with logarithmic transformations. London School of Economics, London, 22(1), 23-36.
Bi, S., Liu, Z., & Usman, K. (2017). The influence of online information on investing decisions of reward-based crowdfunding. Journal of Business Research, 71, 10-18.
Burns, R. P., & Burns, R. (2008). Business research methods and statistics using SPSS. Sage. 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.
Connelly, B. L., Certo, S. T., Ireland, R. D., & Reutzel, C. R. (2011). Signaling theory: A review and assessment. Journal of Management, 37(1), 39-67.
Cordova, A., Dolci, J., & Gianfrate, G. (2015). The determinants of crowdfunding success: Evidence from technology projects. Procedia-Social and Behavioral Sciences, 181, 115-124 Crowdfunding - China | Statista Market Forecast. (n.d.). Retrieved June 24, 2017, from https://www.statista.com/outlook/335/117/crowdfunding/china
Crowdfunding - United States | Statista Market Forecast. (n.d.). Retrieved June 24, 2017, from https://www.statista.com/outlook/335/109/crowdfunding/united-states
Dey, S., Duff, B. R., Karahalios, K., & Fu, W. T. (2017, February). The Art and Science of Persuasion: Not All Crowdfunding Campaign Videos are The Same. In CSCW (pp. 755-769). Dushnitsky, G., Guerini, M., Piva, E., & Rossi-Lamastra, C. (2016). Crowdfunding in Europe: determinants of platform creation across countries. California management review, 58(2), 44-71. Edmondson, A. C., & McManus, S. E. (2007). Methodological fit in management field research. Academy of management review, 32(4), 1246-1264.
Farrar, D. E., & Glauber, R. R. (1967). Multicollinearity in regression analysis: the problem revisited. The Review of Economic and Statistics, 92-107.
Field, A. (2010). Comparing Two Means [pdf]. Retrieved June 27, 2017 from
Frydrych, D., Bock, A. J., Kinder, T., & Koeck, B. (2014). Exploring entrepreneurial legitimacy in reward-based crowdfunding. Venture Capital, 16(3), 247-269.
Granovetter, M. (1992). Problems of explanation in economic sociology. Networks and organizations: Structure, form, and action, 25, 56.
Hofstede, G. (1991). Cultures and organizations: Software of the mind. McGaw-Hill, London. Hofstede, G., & Bond, M. H. (1984). Hofstede's culture dimensions: An independent validation using Rokeach's value survey. Journal of cross-cultural psychology, 15(4), 417-433.
Hofstede, G., & Minkov, M. (2010). Long-versus short-term orientation: new perspectives. Asia Pacific business review, 16(4), 493-504.
Kickstarter. (n.d.). Retrieved June 26, 2017, from https://www.kickstarter.com/help/stats Kunz, M. M., Bretschneider, U., Erler, M., & Leimeister, J. M. (2016). An empirical
investigation of signaling in reward-based crowdfunding. Electronic Commerce Research, 1-37. Kuppuswamy, V., & Bayus, B. L. (2015). Crowdfunding creative ideas: The dynamics of project backers in Kickstarter.
Kuppuswamy, V., & Bayus, B. L. (2017). Does my contribution to your crowdfunding project matter? Journal of Business Venturing, 32(1), 72-89.
Lee, D. Y., & Dawes, P. L. (2005). Guanxi, trust, and long-term orientation in Chinese business markets. Journal of international marketing, 13(2), 28-56.
Mavlanova, T., Benbunan-Fich, R., & Koufaris, M. (2012). Signaling theory and information asymmetry in online commerce. Information & Management, 49(5), 240-247.
Nahapiet, J., & Ghoshal, S. (1998). Social capital, intellectual capital, and the organizational advantage. Academy of management review, 23(2), 242-266.
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. Advances in experimental social psychology, 19, 123-205.
Schwartz, S. H. (1990). Individualism-collectivism: Critique and proposed refinements. Journal of cross-cultural psychology, 21(2), 139-157.
Short, J. C., Ketchen, D. J., McKenny, A. F., Allison, T. H., & Ireland, R. D. (2017). Research on Crowdfunding: Reviewing the (Very Recent) Past and Celebrating the Present. Entrepreneurship Theory and Practice, 41(2), 149-160.
Skirnevskiy, V., Bendig, D., & Brettel, M. (2017). The influence of internal social capital on serial creators’ success in crowdfunding. Entrepreneurship Theory and Practice, 41(2), 209-236. Steinberg, S., & DeMaria, R. (2012). The Crowdfunding Bible: How to raise money for any startup, video game or project. Read. me.
Yang, M. M. H. (1994). Gifts, favors, and banquets: The art of social relationships in China. Cornell University Press.
Zheng, H., Hung, J. L., Qi, Z., & Xu, B. (2016). The role of trust management in reward-based crowdfunding. Online Information Review, 40(1), 97-118.
Zheng, H., Li, D., Wu, J., & Xu, Y. (2014). The role of multidimensional social capital in crowdfunding: A comparative study in China and US. Information & Management, 51(4), 488-496.
Tables Table 1.
Characteristics of signals in reward-based crowdfunding in the dimensions purchase-time continuum and signal costs according to Mavlanova et al. (2012).
Pre-funding phase
(ex-ante)
Funding phase (during-funding)
Post-funding phase (ex-post)
Low cost - - -
High cost Runtime
Number of rewards Funding goal Number of updates Number of likes Number of comments Number of backers Table 2.
Descriptive statistics for all projects.
Table 3.
Descriptive statistics for failed projects.
Table 4.
Table 5.
Descriptive statistics for projects in progress.
Table 6.
Table 7.
Testing hypotheses using binary logistic results.
Hypothesis Independent variable Expected direction Significant effect Direction of effect Outcome
H2 Duration Positive Yes Positive Accepted
H3 rewardAmount Negative No (Positive) Rejected
H4 updateAmount Positive Yes Positive Accepted
H5 likeAmount Positive Yes Negative Rejected
H6 commentAmount Positive Yes Positive Accepted
H7 backerAmount Positive Yes Positive Accepted
H8 targetAmount Positive Yes Positive Accepted
Notes: For every hypothesis, the independent variables are displayed and the expected direction of the hypothesis. Results from the binary logistic regression are displayed in the fourth and fifth column. The outcome of hypotheses are displayed in the last column.
Table 8.
Figures
.
Figure 1. China and the US scored on Hofstede’s six-dimensional model of cultural differences. Countries can score between 0 (very low) and 100 (very high) on each dimension (Hofstede, 2017).
