Unleashing the Power of Teamwork : Evidence from Machine Learning Contests

The increasing prevalence of gig economy platforms has made asynchronous collaboration in virtual teams a popular mode of work in today’s globalized environment. However, the exact value of teamwork remains elusive. In this study, we empirically analyze machine learning contests from Kaggle to investigate the impact of teamwork on performance. We find that participants achieve significantly better performance when working in teams than working independently. Furthermore, experienced participants benefit more from teamwork than their less-experienced counterparts. Our results also demonstrate that rival and nonrival incentives have opposite effects on team performance. The presence of symbolic rewards (nonrival) improves the performance of teams, whereas team performance decreases with the size of the monetary rewards (rival). Finally, our mediation analysis provides suggestive evidence that teamwork enhances performance by stimulating the team to work harder and more creatively. Our study provides critical insights into the value of teamwork and has practical implications for team formation and reward design