1. Outline  In-Class Experiment on First-Price Sealed-Bid Auctions  Professor John Morgan: Internet Auctions  The Winner’s Curse Hypothesis: Kagel and Levin (1986)  Revenue equivalence and strategic equivalence hypotheses: Auction Formats

2. Information Structure

3. Public Information Auctions  Two separate auction markets simultaneously:  Bidding in the first auction market continued as before under private information conditions  After the bids were submitted (but before they were posted), a public information signal was introduced and subjects were ask to bid again.  Public information signal: The lowest of the private information signals distributed, x L was posted.

4. Risk Neutral Nash Equilibrium (without Public Information)

5. The Winner Curse Hypothesis (without Public Information)

6. Risk Neutral Nash Equilibrium (with Public Information X L )

7. The Winner Curse Hypothesis (with Public Information X L )

8. A Summary RNNE Winner’s Curse (Strategic Discounting) No Yes Public Information R2 R3 R2 R1 R1 > R2 > R3

9. Small and Large Groups  Small Groups (3-4 bidders)  Large Groups (5-7 bidders)  All subjects were experienced bidders

10. Research Hypotheses  Hypothesis 1: Under private information conditions, market outcomes are consistent with RNNE.  Hypothesis 2: Announcing X L, the lowest private information signal, raises average seller’s revenues by the average amount predicted under the RNNE model.  Hypothesis 3: Under private information conditions, experienced bidders avoid winner’s curse (i.e., average profits are closer to the RNNE level than the zero/negative profits predicted under the winner’s curse) (a weakened form of Hypothesis 1)

11. Research Hypotheses  Hypothesis 4: Public information raises average seller’s revenue (a weakened version of Hypothesis 2).  Hypothesis 5: Hypotheses 3 and 4 apply uniformly to experiments with small and large numbers of bidders.  Hypothesis 6: Bidders are sensitive to the strategic implications of the auctions (  and public information condition) so that when the winner curse hypothesis and the RNNE model coincide, the RNNE model provides a reasonable characterization of the data.

12. Experimental Design

13. Results: Private Information Conditions

14. Small Group

16. Large and Small Groups

17. Large Group

19. Effects of signal quality 

20. Risk Neutral Nash Equilibrium (without Public Information)

21. The Winner Curse Hypothesis (without Public Information)

22. Empirical Bidding Functions

23. Effects of Public Information on Profits

24. A Summary RNNE Winner’s Curse (Strategic Discounting) No Yes Public Information R2 R3 R2 R1 R1 > R2 > R3

25. Effects of Public Information on Seller’s Revenue R2-R3R2-R1

26. Effects of Public Information | Winner’s Curse

27. Risk Neutral Nash Equilibrium (with Public Information X L )

28. The Winner Curse Hypothesis (with Public Information X L )

29. Empirical Bidding Functions

30. Research Hypotheses  Hypothesis 1: Under private information conditions, market outcomes are consistent with RNNE.  Hypothesis 2: Announcing X L, the lowest private information signal, raises average seller’s revenues by the average amount predicted under the RNNE model.  Hypothesis 3: Under private information conditions, experienced bidders avoid winner’s curse (i.e., average profits are closer to the RNNE level than the zero/negative profits predicted under the winner’s curse) (a weakened form of Hypothesis 1)

31. Research Hypotheses  Hypothesis 4: Public information raises average seller’s revenue (a weakened version of Hypothesis 2).  Hypothesis 5: Hypotheses 3 and 4 apply uniformly to experiments with small and large numbers of bidders.  Hypothesis 6: Bidders are sensitive to the strategic implications of the auctions (  and public information condition) so that when the winner curse hypothesis and the RNNE model coincide, the RNNE model provides a reasonable characterization of the data.

32. Summary  H1 and H2  Rejected  H3 and H4  Not rejected  H5  Rejected  H6  Not rejected