Mechanism Choice and Strategic Bidding in Divisible Good Auctions: An Empirical Analysis of the Turkish Treasury Auction Market

Abstract

An important question in the theory of divisible good auctions is to determine whether a discriminatory auction yields higher revenues to the auctioneer than a uniform price or Vickrey auction. This question can be answered empirically using estimates of the bidders' true marginal valuations for the good being auctioned. Based on the share auction model of Wilson (1979), I propose an estimation method to reconstruct the distribution of bidders' marginal valuations using data on individual bids.I apply the method to data from the Turkish Treasury auction market. Using the estimated marginal valuations of the bidders to calculate what the revenue would have been under a uniform price or a Vickrey auction. I find that, ex-post, the discriminatory auction yielded more revenue to the Turkish Treasury than the alternatives. However, in ex-ante expected revenue comparisons, I fail to reject revenue equivalence between the discriminatory auction and the "best-case" outcome of the uniform price auction, in which bidders reveal their valuations truthfully. I investigate the robustness of my results to the presence of a common-value component in bidders' utilities when comparing the revenue performance of the discriminatory and Vickrey mechanisms. I also show that the estimation framework can be generalized to allow for deterministic vs. uncertain supply, asymmetries, and stochastic participation by bidders.

Centralizing Disconnected Markets? An Irrelevance Result

“…With imperfect divisibility of goods or buyers who can submit only a maximal amount of bids the analysis becomes more complex due to discontinuities and rationing. This has been demonstrated recently by Horaçsu and McAdams (2010) and Kastl (2011Kastl ( , 2012.…”

Structural Econometric Methods in Auctions: A Guide to the Literature

“…This can often lead to various multi-object models being non-identified. Coping techniques for partial identification based on the first-order conditions (e.g., Cantillon and Pesendorfer (2007)), or more basic optimality conditions (e.g., Hortaçsu and McAdams (2010)) are discussed below.…”

“…Because of this difficulty, researchers have often chosen to employ a modified version of the GPV method, where costs are inferred from bids using some form of a necessary optimality condition, after which counterfactual exercises are based directly on the sample of estimated costs, rather than on their underlying distribution (e.g., see Hortaçsu and McAdams (2010) and Cantillon and Pesendorfer (2007)). Coping techniques for computing counterfactuals have also been developed, but as these are highly context-specific.…”

CDS Credit Event Auctions

“…We must emphasize here that in reality bidders probably do not submit a strictly downward sloping demand function and rather submit a step function. In such a case what we identify and estimate here based on the first order conditions are like the bound of the distribution of signals (Hortacsu and Mcadams, 2010). We shall abstract away from these considerations in this paper.…”

“…Resampling procedure Hortacsu and Mcadams (2010) describe an approach for consistently estimating the residual supply curve for a bidder. We describe their resampling procedure here.…”