Auctions with Toeholds: an Experimental Study

Slides:



Advertisements
Similar presentations
Behavioural Industrial Organization Sotiris Georganas.
Advertisements

EC941 - Game Theory Prof. Francesco Squintani Lecture 4 1.
CPS Bayesian games and their use in auctions Vincent Conitzer
Ultimatum Game Two players bargain (anonymously) to divide a fixed amount between them. P1 (proposer) offers a division of the “pie” P2 (responder) decides.
Non myopic strategy Truth or Lie?. Scoring Rules One important feature of market scoring rules is that they are myopic strategy proof. That means that.
Private-value auctions: theory and experimental evidence (Part I) Nikos Nikiforakis The University of Melbourne.
By: Bloomfield and Luft Presenter: Sara Aliabadi October,9,
Liz DiMascio Paige Warren- Shriner Mitch Justus DUTCH AND ENGLISH AUCTIONS IN RELATION TO THE TULIP MARKET.
Congestion Games with Player- Specific Payoff Functions Igal Milchtaich, Department of Mathematics, The Hebrew University of Jerusalem, 1993 Presentation.
Game Theory 1. Game Theory and Mechanism Design Game theory to analyze strategic behavior: Given a strategic environment (a “game”), and an assumption.
4 Why Should we Believe Politicians? Lupia and McCubbins – The Democratic Dilemma GV917.
Biased Price Signals and Reaction of Bidders in Vickrey Auction IPOs Aytekin Ertan December 2009.
An Introduction to Game Theory Part IV: Games with Imperfect Information Bernhard Nebel.
B OUNDED R ATIONALITY in L ABORATORY B ARGAINING with A SSYMETRIC I NFORMATION Timothy N. Cason and Stanley S. Reynolds Economic Theory, 25, (2005)
BEE3049 Behaviour, Decisions and Markets Miguel A. Fonseca.
Week 10 1 COS 444 Internet Auctions: Theory and Practice Spring 2008 Ken Steiglitz
Outline  In-Class Experiment on First-Price Sealed-Bid Auctions  Professor John Morgan: Internet Auctions  The Winner’s Curse Hypothesis: Kagel and.
Week 11 1 COS 444 Internet Auctions: Theory and Practice Spring 2008 Ken Steiglitz
January 29, 2004 Experimental Economics 1 Outline  In-class experiment on IPV First-Price Auctions  Data from Cox, Robertson, and Smith (1982)  Glenn.
Project 2-Guidelines. Recall- Class Project-Goals  Determine what would be expected to happen if each company bid the same amount as its signal.  Determine.
The Agencies Method for Coalition Formation in Experimental Games John Nash (University of Princeton) Rosemarie Nagel (Universitat Pompeu Fabra, ICREA,
1 A unified approach to comparative statics puzzles in experiments Armin Schmutzler University of Zurich, CEPR, ENCORE.
Chapter 12 Choices Involving Strategy Copyright © 2014 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written.
Econ 805 Advanced Micro Theory 1 Dan Quint Fall 2009 Lecture 4.
Chapters 29, 30 Game Theory A good time to talk about game theory since we have actually seen some types of equilibria last time. Game theory is concerned.
Experimental Economics NSF short course David Laibson August 11, 2005.
Tests of Revenue Equivalence in Internet Magic Auctions David Lucking-Reiley “I have never considered the lab to be a substitute for field empirical work.”
Data Analysis Econ 176, Fall Populations When we run an experiment, we are always measuring an outcome, x. We say that an outcome belongs to some.
Steffen Staab 1WeST Web Science & Technologies University of Koblenz ▪ Landau, Germany Network Theory and Dynamic Systems Auctions.
Auctions serve the dual purpose of eliciting preferences and allocating resources between competing uses. A less fundamental but more practical reason.
Auctions serve the dual purpose of eliciting preferences and allocating resources between competing uses. A less fundamental but more practical reason.
Coordination and Learning in Dynamic Global Games: Experimental Evidence Olga Shurchkov MIT The Economic Science Association World Meeting 2007.
Auctions serve the dual purpose of eliciting preferences and allocating resources between competing uses. A less fundamental but more practical reason.
Oligopoly Theory1 Oligopoly Theory (6) Endogenous Timing in Oligopoly The aim of the lecture (1) To understand the basic idea of endogenous (2) To understand.
First Price common value auctions: Bidder behavior and the winner’s curse John Kagel, Dan Levin, Raymon Battaglio & Donal Meyer.
Incomplete Information and Bayes-Nash Equilibrium.
1 Types of Auctions English auction –ascending-price, open-outcry Dutch auction –descending-price, open-outcry 1 st price sealed bid auction –known as.
Chapter 12 Game Theory Presented by Nahakpam PhD Student 1Game Theory.
Game theory basics A Game describes situations of strategic interaction, where the payoff for one agent depends on its own actions as well as on the actions.
Chapter 28 Game Theory.
Correlated equilibria, good and bad: an experimental study
Intermediate Microeconomics
By Audrey Hu and Liang Zou University of Amsterdam
Bayesian games and their use in auctions
CPS Mechanism design Michael Albert and Vincent Conitzer
Tools for Decision Analysis: Analysis of Risky Decisions
Rational Emotions: Orders of Reasoning and Managing Complexity Timo Ehrig and Shyam Sunder Executive Summary How to model expectations, when people anticipate.
“Swooping” in for the win: Bidding strategies in pay-per-bid auctions
Bidding on an Antique.
Lecture 8 Asymmetric Information: Adverse Selection
Econ 805 Advanced Micro Theory 1
Game Theory.
Unit 4 SOCIAL INTERACTIONS.
Game Theory in Wireless and Communication Networks: Theory, Models, and Applications Lecture 2 Bayesian Games Zhu Han, Dusit Niyato, Walid Saad, Tamer.
Learning 6.2 Game Theory.
Asymmetric auctions with resale: an experimental study
Chapter 29 Game Theory Key Concept: Nash equilibrium and Subgame Perfect Nash equilibrium (SPNE)
Vincent Conitzer Mechanism design Vincent Conitzer
Vincent Conitzer CPS 173 Mechanism design Vincent Conitzer
Chapter 14 & 15 Repeated Games.
Chapter 14 & 15 Repeated Games.
Statistical significance using p-value
EASTERN MEDITERRANEAN UNIVERSITY DEPARTMENT OF INDUSTRIAL ENGINEERING IENG314 OPERATIONS RESEARCH II SAMIR SAMEER ABUYOUSSEF
UNIT II: The Basic Theory
Chapter 26 Comparing Counts.
Lecture 10 Coordination and Reputation
Information, Incentives, and Mechanism Design
Auction Theory תכנון מכרזים ומכירות פומביות
CPS Bayesian games and their use in auctions
Presentation transcript:

Auctions with Toeholds: an Experimental Study Sotiris Georganas, Ohio State and U Bonn Rosemarie Nagel, U Pompeu Fabra (Barcelona) Econometric Society NASM 2008

Motivation Takeover battles can be viewed as CV auctions Often one of the parties already owns a part of the company sold (a toehold) Owning a toehold significantly alters the strategic setup Toeholds or other asymmetries lead in theory to explosive equilibria, toehold owners get the company for very low price (Bulow et al 1999, Klemperer 1998) However empirical studies (Betton, Eckbo 2000) find that only 50% of bidders acquire a toehold Why? When company taken over, often there can be a counteroffer, and a counteroffer to that -> takeover battle More on explosive equilibria later We do an experiment in a controlled environment to see what is going on Bulow klemperer huang JPE Klemperer 1998 almost common values Sotirios Georganas - Auctions with Toeholds

Model 2 risk neutral bidders i and j compete in an English auction for one unit of an indivisible good (e.g. a company) Own share θi of the company, common knowledge Signals ti ~U[0,100] Company value: sum of signals ti+ tj Unique Symmetric Equilibrium (Bulow et al. 1999): Next slide showing the bids Sotirios Georganas - Auctions with Toeholds

Equilibrium Bids No toeholds, diagonal Sotirios Georganas - Auctions with Toeholds

Experimental Design Experiments at LeeX in Barcelona, computerised (zTree), 16 subjects at each session 3 treatments: toehold pairs 1%-5% (2 ind. obs.), 1%-20% (4 obs.) and 1%-50% (6 obs.) Toehold common knowledge, players alternate roles 50 periods, random matching No bankruptcies Sotirios Georganas - Auctions with Toeholds

Results Win frequencies 1-5 1-20 1-50 Actual 0.51 0.451 0.3375 Two types very close, no explosive effect, weak types win far too often Win frequencies 1-5 1-20 1-50 Actual 0.51 0.451 0.3375 Predicted 0.167 0.038 0.0158 Sotirios Georganas - Auctions with Toeholds

Tobit regressions Why tobit? Known to people who work with english auctions: unobservable bids (especially with two persons) Sotirios Georganas - Auctions with Toeholds

Results contd. Deviation from equilibrium bids very high Leads to deviation in prices Treatment Mean actual price Mean predicted price Mean deviation 1-5 88,7 61 27.7 1-20 73.8 60.3 13.5 1-50 76.9 69.2 7.7 Actual and predicted have same ranking, but differ a lot Explanation? Sotirios Georganas - Auctions with Toeholds

Payoff flatness High type in 1-20 Emphasize when signal 100 you bid 200! Sotirios Georganas - Auctions with Toeholds

Equilibrium has very flat payoffs we find that when the ratio of the two players toeholds is larger than 10 (e.g. 1% and 10%), the strong bidder can deviate almost 50% from his optimal bid with a negligible loss in expected payoff. Payoffs extremely flat, convergence to equilibrium almost impossible Alternatively: LOR Sotirios Georganas - Auctions with Toeholds

Generality of flatness Payoff flatness is a problem in all explosive equilibria In our case, weak type bids less than 100 for almost all values Thus, strong type gains almost nothing by bidding more than 100 Hence, flat payoff above 100 Payoffs always flat in equilibria where weak type bids very passively Sotirios Georganas - Auctions with Toeholds

Consequences of flatness Suppose subjects’ initial bidding is not close to equilibrium Flat payoffs mean their incentives to adjust towards equilibrium are weak According to most learning models convergence to equilibrium will be very slow If agents have some small unmodeled reasons for disequilibrium behavior, incentives towards equilibrium don’t suffice for convergence In practice : equilibria that suffer from flatness are not good predictions for real life situations Corollary: we will never see explosive equilibria in real life! Sotirios Georganas - Auctions with Toeholds

If equilibrium fails… Bounded rationality models can describe behavior better In this case, non equilibrium models like a levels of reasoning model (Nagel 95, Stahl/Wilson 95, Crawford/Iriberri 2008) LOR models assume people best respond But consistency of beliefs assumption is relaxed Players are characterized by their beliefs L1 players believe others randomize and best respond to this belief, L2 players assume everyone else is L1 and best respond, L3 best respond to L2 and so on… I use the level k version used in auctions by Crawford Sotirios Georganas - Auctions with Toeholds

Levels of reasoning L1 starting from random! LOR is not an equilibrium concept and thus it could describe behavior better Sotirios Georganas - Auctions with Toeholds

Model fit Low likelihood good Only 1-50 does not work well Sotirios Georganas - Auctions with Toeholds

Profitability of toeholds: sunk costs This means, when you have already bought the toehold Sotirios Georganas - Auctions with Toeholds

Profitability of toeholds: ex ante Including the cost Sotirios Georganas - Auctions with Toeholds

Comparison with previous results Avery and Kagel (1997), Kagel and Rose (forthcoming) analyse somewhat similar games (almost common values, sealed bid and English resp.) Do not vary toehold size Similarly to us do not find explosive equilibria Attribute deviation to naïve players (just add advantage to signal, similar to L1) No analysis of payoff flatness No levels of reasoning In kagel rose we have Cv but one person has CV+k, experienced subjects, English auction with 4 bidders, no winners curse (because experienced) Avery kagel: two person sealed bid wallet game (ti+tj) Sotirios Georganas - Auctions with Toeholds

Extensions Other equilibrium concepts (QRE?) Sealed bids Vary number of bidders Different toehold concept (insider gets informational advantage) Sotirios Georganas - Auctions with Toeholds

Conclusions No signs of explosive equilibria LOR describes data better Equilibrium payoffs extremely flat No convergence in reasonable time LOR describes data better Toeholds do raise probability of winning But sometimes not enough to cover the cost to acquire them This may be the explanation why bidders acquire no toeholds Sotirios Georganas - Auctions with Toeholds

Appendix: deviations in payoff space Sotirios Georganas - Auctions with Toeholds

Flatness with actual data… So with actual behaviour, better incentives to play a best response and it seems this is more or less what subjects are doing. Sotirios Georganas - Auctions with Toeholds