管理评论 ›› 2020, Vol. 32 ›› Issue (11): 102-111,259.

• 技术与创新管理 • 上一篇    下一篇

参与者个数对竞赛绩效的影响——基于设计类与编程类创新竞赛数据的分位数回归分析

胡锋1,2, 赵红3, 刘超1,2   

  1. 1. 北京工业大学经济与管理学院, 北京 100124;
    2. 北京现代制造业发展研究基地, 北京 100124;
    3. 中国科学院大学经济与管理学院, 北京 100190
  • 收稿日期:2018-07-19 出版日期:2020-11-28 发布日期:2020-12-05
  • 通讯作者: 胡锋(通信作者),北京工业大学经济与管理学院副教授,博士
  • 作者简介:赵红,中国科学院大学经济与管理学院教授,博士;刘超,北京工业大学经济与管理学院教授,博士。
  • 基金资助:
    国家自然科学基金青年项目(71802014);北京市社会科学基金基地项目(18JDGLB041)。

The Influence of the Number of Solvers on the Contest Performance: Quantile Regression Analysis Based on the Data from Designing and Coding Innovation Contests

Hu Feng1,2, Zhao Hong3, Liu Chao1,2   

  1. 1. College of Economics and Management, Beijing University of Technology, Beijing 100124;
    2. Research Base of Beijing Modern Manufacturing Development, Beijing 100124;
    3. School of Economics and Management, University of Chinese Academy of Sciences, Beijing 100190
  • Received:2018-07-19 Online:2020-11-28 Published:2020-12-05

摘要: 参与者个数对竞赛绩效的影响是创新竞赛研究中学者们长期关注的议题之一。而现有研究对于上述影响的认识并不一致,存在或为正向影响,或为负向影响的分歧。本文通过归纳、提炼现有研究,区分了通过平均值和最优值测度竞赛绩效两类方法,提出了参与者个数影响竞赛绩效的两种效应:“动机效应”和“路径效应”,用以调和以往研究的不一致。基于分别在两个知名的创新竞赛在线平台上收集的共计6771个设计类与编程类创新竞赛数据,本文建立基于马尔科夫链蒙泰卡罗方法估计的分位数回归模型,实证了上述两个效应的存在:参与者个数负向影响平均值测度的竞赛绩效,正向影响最优值测度的竞赛绩效。此外,与最小二乘回归相比,分位数回归提供了关于上述两种效应更为详细的洞察力。最后,本文给出了相应的理论贡献、管理意义和未来研究方向。

关键词: 创新竞赛, 参与者个数, 竞赛绩效, 分位数回归

Abstract: The influence of the number of solvers on the contest performance is a long-lasting question in innovation contests research. However, the findings about this effect in the existing literature are not consistent. Some are positive, while others are negative. In order to reconcile those contradictory insights, this research sorts out and sums up the existing researches, defines the measurement of innovation performance in terms of the average quality of solutions or the quality of the best solution, and proposes the incentive effect and the parallel path effect that underline the influence of the number of solvers on the contest performance. Based on these reasoning, this research collects behavioral data of 6,771 innovation contests for designing and coding problems from two well-known online platforms, builds a quantile regression model, and empirically proves the existence of the two effects, namely, the number of solvers negatively influences the average quality of solutions, and positively influences the quality of the best solution. Besides this, compared with the ordinary least square regression, results from quantile regression provide more subtle and comprehensive insights about these effects. At last, academic contribution, managerial implication, and future research direction are presented.

Key words: innovation contests, the number of solvers, contest performance, quantile regression