实时的社会信息与互联网产品众筹的动态性——基于大数据的采集与挖掘研究

本文使用互联网大数据实时采集工具,针对我国三家典型的互联网产品众筹平台逐日抓取了2015年10月至2016年2月共130天的在线产品众筹项目的实时大数据,基于实时的社会信息探索了产品众筹过程中的动态性。本文依据社会责任扩散效应与目标梯度效应构建了理论预期,经检验发现,在产品众筹的起始阶段,新增的支持金额(人数)与累计支持金额(人数)的回归系数显著为正且呈明显的下降趋势,而在结束阶段,两者的回归系数又重新显著为正且呈上升的趋势,但上升趋势不够明显。此外,随着日趋接近于众筹的目标金额,新增的支持金额(人数)存在显著的加速过程。这种产品众筹中体现出来的动态性意味着,众筹的起始阶段和结束阶段由实时的社会信息所决定的筹资状况以及尽快实现众筹目标金额过半的筹资是决定产品众筹成败的关键。

Real-time Social Information and Dynamics of Reward-based Internet Crowdfunding: Based on Big Data

In China, reward-based crowdfunding on the Internet is developing fast and was valued at 18.8 billion RMB in 2017. However, academic research on reward-based crowdfunding focuses mainly on static cross-sectional data, including the factors that affect the success of crowdfunding, the role of geographical distance, social capital and social networks in crowdfunding, etc. A few scholars undertake dynamic research, such as Burtch et al. (2012) and Kuppuswamy and Bayus (2013, 2015), who study the changes and reasons for support in reward-based crowdfunding over time. Overall, however, dynamic research is still in its infancy. This paper focuses on reward-based crowdfunding in China, learns from Kuppuswamy and Bayus (2013, 2015), and draws on the theories of social responsibility diffusion and the target gradient effect to explore the dynamic impact of decisions by other supporters (real-time social information) on the decision-making of potential supporters. We systematically collect real-time big data in the process of crowdfunding and explore the impact of real-time social information.We use Web crawler software for big data, Octopus collection software, to collect real-time information at a fixed time every day on the three reward-based crowdfunding platforms in China—Jingdong, Taobao, and Zhongchou —for around 130 days from October 2015 to February 2016. Then we use data mining and collation to build up a database to record such data as the amount of support and number of people in each crowdfunding project, including both the new additions every day and the total, the number of people focusing on the project, whether the project is promoted on the website homepage, whether the project has reached its crowdfunding target, the total number of crowdfunding projects in process on the website, and the support for the most popular project on the website. Based on a total sample of 3,949 reward-based crowdfunding projects, we find that real-time social information, measured as the total amount of support and people, has a dynamic impact on the decisions of potential supporters. First, social responsibility diffusion plays an important role in reward-based crowdfunding. At the beginning, when potential supporters find that others have provided enough help and they believe the remaining amount required will be met by others, the social responsibility or compassion of potential supporters will either diminish or spread. Therefore, the regression coefficient of the amount of support added daily and the cumulative support reflecting the real-time social information is significantly positive but shows a significant downward trend over time. At the end, under the “all or nothing” rule of crowdfunding, the social responsibility and compassion of potential supporters recovers and the regression coefficient returns to a significant positive and rising trend toward the cut-off date. However, in China we find that the diffusion effect of social responsibility is stronger in the initial stage, and, although the waning effect is relatively weak at the end, it is much more obvious in projects that have not yet reached the target level near the deadline.Nevertheless, the target gradient effect cannot be ignored in reward-based crowdfunding. The supportive decisions of others positively affect peoples judgments about the value of the shared goal, and there is thus significant acceleration in support as the crowdfunding project approaches its financing objective. This paper supports the results of Kuppuswamy and Bayus (2013, 2015) on Kickstarter. However, they use dummy variables or the absolute amount of support to examine the absolute time effect. Based on dynamic panel data, this paper uses the percentage of support and the number added every day as the dependent variable, and examines the effect under real-time information (the cumulative amount of support). We also find that the weakening effect is much more obvious in projects that have not yet reached the target level near the deadline. Our result is more persuasive. Our findings provide a reference for initiators of reward-based crowdfunding. They should pay close attention to the diffusion of social responsibility in the initial stage of the competition. During the middle stage, initiators should secure more than half of the required financing as fast as possible. In the end stage of crowdfunding, the sponsor should be concerned to reawaken the sense of responsibility or compassion among potential supporters to achieve the financing target.