LEEM Seminar - Paul Pezanis-Christou (U. of Adelaïde, Australia)

"A rationale for the bidder's curse in the Takeover game"
Quand ? Le 19-12-2018,
de 12:00 à 13:00
Où ? Salle SCI - ISEM
S'adresser à
Ajouter un événement au calendrier vCal
iCal

LEEM Seminar

Paul Pezanis-Christou  (U. of Adelaïde, Australia)

Title: A rationale for the bidder's curse in the Takeover game

Authors: Paul Pezanis-Christou and Hang Wu.

Abstract: The paper reports on the Takeover game (a.k.a. the Acquiring a Company game) where the buyer of an asset of unknown value is asked to make an offer to the asset’s owner. The owner knows the asset’s value and accepts the offer only if it is greater than the asset’s value. Whatever the asset’s value, the buyer knows that upon owning it, its value will be multiplied by . This game has been extensively studied in laboratory experiments as well as with executives in MBA classes and a well-established outcome is that participants tend to bid too high for the asset’s ownership and typically suffer the bidder’s curse. We propose a rationale for this pattern based on a naive Impulse Balance Equilibrium argument (Ockenfels and Selten, 2005, Pezanis-Christou and Wu, 2018). This approach postulates that buyers do not attempt to maximize their expected profit from acquiring the asset but to balance their anticipated regrets (impulses) from acquiring and from not acquiring it. We consider two variants and show that their parameter-free versions lead to a bidder’s curse. Assuming an impulse weighting parameter to characterize an uneven weighting of impulses, the models also rationalize optimal (zero) bidding as the result of buyer’s infinite aversion to the anticipated regret of overbidding. We assess the models’ goodness-of-fit with the experimental data and find that they explain behavior well and better than cursed equilibrium (Eyster and Rabin, 2005).  We also draw in- and out-of-context predictions and find that experiencing first a simplified (binary) version of the game generates better out-of-context predictions for behavior in its standard version than the opposite scenario. (This is still ‘work in progress’ but a draft will be available soon.)