B. Tsirelson

Single unit auctions; still enigmatic

Recent works

Michael Landsberger, Boris Tsirelson,
"Single unit auctions; still enigmatic."
SSRN 195128.
Available online (free of charge) from Social Science Research Network:

A research preprint, 27 pages, 8 figures, bibl. 15 refs.

We consider the well-known model proposed by Milgrom and Weber in 1982, where players are risk neutral, their valuations V1,...,Vn are neither private nor common, they depend on privately observed signals X1,...,Xn and a common, unobservable, random variable S (possibly, multidimensional). We prove that if

the existing results for single unit auctions are not valid anymore, in many cases. In fact, it is not known how many equilibria (if any) exist in such auctions, and how to define (construct, select) the relevant equilibrium. The reason is that in many cases if an equilibrium exists, it can not be supported by monotone (threshold) participation strategies. The driving force of this result is the dual meaning of interdependent signals. When signals are independent, a higher signal is good news. Under dependence, in a monotone equilibrium, a higher signal is both, good and bad news.

  1. Introduction.
  2. Some properties of participation in auctions.
  3. Treshold participation strategies.
  4. Large number of correlated signals.
  5. Heavy tails and necessary conditions.
  6. Good news bad news
  7. Appendix: Exchangeability.
back to Preprints of 1999 and before back to my homepage