Contests are commonly used by governments, educational institutions, companies, and NGOs to inspire technology innovations, foster learning, and promote productivity. This paper investigates how to design a contest that can maximise participant’s engagement (effort) in the contest. We manipulate the information disclosure policy, i.e., whether the actual number of contestants is announced to the participants or not and find both theoretically and experimentally that the optimal disclosure policy depends on the convexity of the cost of effort function. This study provides profound policy implications that the best design depends on the types of contests. In areas such as digital innovation, where the margin cost of effort is increasing (pushing the limit), contestants exert higher total effort when not knowing the number of competitors, whereas, for production contest in the factory, where the marginal cost is decreasing (learning effect), it is better to announce the number of contestants.

Empirical evidence across R&D in the high-tech industry, finance, and sports tournaments demonstrates that competitions for rank-based rewards often lead to excessive risk-taking, causing ethics issues, market volatility, and athlete safety concerns. This paper examines how reward structure influences contestants’ risk strategy. We provide both the theoretical prediction and the empirical evidence through lab experiments to show that although unequalised prize allocation and increased contestant numbers both raise risk-taking, they affect the risk strategy differently. The result underscores the need to consider higher moments like skewness, in addition to variance, for accurate risk assessment. Our study has important implications for the design of rank-based contests such as annual evaluations within a company, R&D races, political campaigns, and sports tournaments. Principals should consider the risk-taking behaviour in the institutional design and seek comprehensive measurements when evaluating the risk exposure.

Due diligence costs for bidders are substantial in high-stakes auctions such as natural resources, company takeover, or digitalised assets auctions empowered by blockchain technology. The costs are often shared with auctioneers who risk low participation and lower bid prices. A selection process that identifies bidders with the highest expected value for the asset improves seller revenue and total welfare. We experimentally compare three two-stage auction mechanisms (auction with an endogens entry stage) and find the indicative bidding mechanism, which admitted a limited number of bidders to enter the auction based on bidders’ non-binding bid in the entry stage, generates weakly higher revenue and social welfare than the other two alternatives. This advantage arises from increased bidder willingness to participate, which the model does not predict. This study presents compelling empirical evidence strongly advocating for the widespread adoption of indicative bidding and unravels the underlying channels that drive this mechanism.

The development of technology and AI has dramatically increased not only the efficiency but also the complexity, variability and uncertainty of today’s economic eco-system. Outcomes in financial markets, sports competitions, or even career and education performance may fluctuate in ways that obscure a person’s actual ability level. When luck(randomness) is a significant factor in outcomes, on the one hand, seeking feedback is crucial in helping the individual to accurately access their true abilities to ensure personal development. On the other hand, individuals might avoid feedback to keep the “happy blindness” after a favourable outcome or to keep “self-preserving denial” after a bad result. This study uses experimental methods to examine individuals’ demand for feedback when luck plays a role, illuminating the design of feedback mechanisms. In addition, we examine the gender gap on this topic. Previous literature found that females are more likely to avoid feedback than males in competitive environments, which might contribute to the gender gap observed in the labour market. Hence, this study also carries profound implications for workforce gender equality in highly competitive industries such as finance and high-tech.

Most contemporary contests are team contests where rival teams compete head-to-head in distinct component battles. For example, in the development of autonomous driving, companies compete on disjoint fronts such as machine learning, robotics, sensor technology, regulation, economics and market analysis. More often than not, contest designers aim not only to generate the most innovative ideas but also to motivate all participants to promote a culture of innovation and maintain competitiveness in the market. This paper examines the prize design that maximises the total effort of all contestants in the team contests, theoretically and experimentally. Theoretically, we prove that when the two competing teams are sufficiently asymmetric, the optimal prize design is a majority rule with a headstart, a biased prize allocation rule favouring the weaker team by awarding it an advantage. We conduct lab experiments to compare the optimal prize allocation rules with three other commonly used ones in the real world. Our theoretical and experimental results provide critical insight into the design of team contests and justify the use of biased prize allocation rules, especially in areas such as technology development and R&D.

Expert advice is important when individuals make decisions that involve specialised knowledge or experience. However, the incentives are often not fully aligned between the decision-makers and the experts. For example, financial experts might recommend the wrong products because of higher commissions and a tech company might push subscription plans beyond your needs. Conventional wisdom holds that competition supports more information transmission. In this project, we examine experimentally how having two experts giving advice affects experts’ lying behaviour and decision makers’ welfare under an extended model based on the classic cheap talk model where there is no competition for the expert.