This paper studies pay-per-bid auctions, a type of auction where bidders incur a cost each time they place a bid. There is strong empirical evidence that these auctions can generate higher expected revenues than second price auctions. In particular, the data shows that pay-per-bid auctions generate revenue that is on average 50 percent higher than what can be achieved in second price auctions. In order to explain this excess revenue puzzle, I develop a general theoretical framework for pay-per-bid auctions where bidders are strategic and forward-looking. In this model, bidders have private valuations and can enter and exit throughout the duration of the auction. Bidders have uncertainty about how many other bidders are active in each round and I allow the bidders to have non-equilibrium beliefs about this number. I then propose a simulated method of moments estimator that can be used to recover the structural parameters even when bidders have biased beliefs. In the empirical application, I use data from an online auction website called Swoopo and I show that the excess revenues we observe cannot be generated by a model where bidders have rational expectations but is consistent with a model where bidders have biased beliefs. Finally, I quantify the effect of biased beliefs using counterfactual analysis. I find that the excess revenues we observe are not sustainable and without biased beliefs, the bidders may choose not to participate in these auctions.
This paper studies the identification of dynamic models of market entry and exit where firms can choose to buy or lease fixed inputs. We are particularly interested in the identification of counterfactual experiments that change the stochastic process of the state variables that relate to the selling price and/or leasing price of the fixed input. First, we show that without additional restrictions, the structural functions that represent entry cost, fixed cost and scrap value are not point identified. Second, for applications where the researcher has data on both leasing price and selling price of the fixed input, we propose plausible exclusion restrictions that provide point identification of our counterfactual experiments, despite the structural functions being nonidentified. Third, we propose a simple simulation based method to construct a confidence region for counterfactual choice probabilities in applications where counterfactual experiments are not point identified. We illustrate our results and methods using numerical experiments.
Accepted, Journal of Development Economics
We examine how a government-initiated e-commerce platform (GEP) influences the sales of a local specialty in China’s Pu’er tea market. Using a unique dataset from field experiments and surveys of 983 farmers, which cover over 95% of local tea output over five years, we analyze changes in online and offline sales over time. We employ two-way fixed effects (TWFE) models to identify the causal impact of GEP access. The results reveal significant substitution effects: for tea of a given quality, access to the GEP increases online sales by 16.649% and decreases offline sales by 15.549%, indicating an overall shift from offline to online sales. On the extensive margin, households that previously only sold offline become more likely to sell online. On the intensive margin, adopters expand their online channels and list a wider range of tea qualities. We observe a pattern in how farmers enter the online market: they typically start selling through social media before moving to e-commerce platforms, including GEPs. These findings suggest that the substitution effect arises not only through the GEP platform itself but also through its bundled public services, such as cooperative packaging and regional branding. The mediation analysis suggests that the increase in online sales channels and product variety explains the effect of GEP access on the shift of online transactions.
The influence of social networks on exercise habits and health capital accumulation is an area that is gaining significant attention. Using a nationally representative sample, this study investigates the role of peer groups in shaping the exercise behavior of junior high school students in China. By exploiting the random assignment of students to classrooms, we evaluate the effect of peers’ exercise habits on the likelihood of forming persistent exercise habits. Our empirical study affirms that having peers who exercise regularly has a robust and positive impact on the student’s own exercise habits. Further analysis shows that this effect is more pronounced for male students, while family income and geographic factors appear to have minimal influence. To explain this result, we identify challenges that female students face in forming intra-class friendships in junior high school, which subsequently reduces their potential to fully benefit from their peers. Finally, we highlight an important policy dilemma: while peer influence does promote exercise, it also has the unintended consequence of negatively affecting academic performance. The dual-faceted nature of the effect of peers complicates the design of effective policies, as initiatives aimed at increasing physical activity may come at the expense of the student’s academic achievement. Our study offers policymakers evidence-based guidelines to help them navigate the complexities of encouraging physical activity while also mitigating the potential consequences on academics.
Evaluating the Efficiency of Canadian Spectrum License Auctions
Explaining Markup Heterogeneity in Pay-per-bid Auctions
Testing Informational Assumptions in Empirical Entry Games