Bidding frictions in Ascending Auctions with Robert A. Miller (supersedes "Ascending Auctions with Costly monitoring", SUBMITTED)
This paper develops a strategy for identifying and estimating the valuation distribution in ascending auctions where bidders have an unknown number of bidding opportunities. Our model leads to equilibrium jump bidding and potentially inefficient allocations. The identification strategy does not rely on equilibrium uniqueness or that it be separating. We apply the model to data drawn from a monthly financial market, where the treasurers of state governments purchase from local banks savings vehicles for unallocated state funds via a procurement auction mechanism based on the offered interest rate. The data exhibit features suggesting the presence of frictions limiting the ability to re-bid. We prove that the distribution of valuations are identified and estimation is feasible even in the presence of multiple and mixed strategy equilibria. We show that it is also possible to allow for unobserved heterogeneity with additional structure on how reserve prices are set. The model estimates provide insight into a local banking market can shed light on the effects the recession in 2008.
In this paper I empirically investigate prediction markets for binary options (Arrow-Debreu securities) for political events. Advocates of prediction markets have suggested that asset prices are consistent estimators of the ``true" probability of a state of the world being realized. This claim is impossible to test outside of a laboratory setting. I am therefore constrained to testing whether the market reaches a ``consensus". In other words, is there evidence that traders update beliefs based on the path of play and eventually reach agreement on the underlying true state of the world. For example, I use information on traders' portfolios over time and measure whether their asset positions shift closer to the same state of the world. I find little evidence for convergence in beliefs. Given that the platform has succeeded in attracting a large number of traders whose actions might reveal their private belief, I then determine whether an econometrician using data beyond execution prices can leverage this data to estimate the consensus belief. I use an incomplete specification of equilibrium outcomes, which echoes the approach of Sutton and Haile and Tamer (2003), to derive bounds on beliefs from order submission decisions. Ordersprovide little extra information. Interval estimates of mean beliefs cannot exclude aggregate beliefs equal to 0.5. These negative results are most likely due to the presence of a large number of loss-making ``noise traders". Their suboptimal choices allow informed traders to transact at prices that conceal their private information. The results suggest that betting-like markets might be worth investigating as alternative aggregation mechanisms.