Going... going... wrong: a test of the level-k (and cognitive hierarchy) models of bidding behaviour

How to cite

Rasooly, Itzhak. 2021. "Going... going... wrong: a test of the level-k (and cognitive hierarchy) models of bidding behaviour." AEA RCT Registry. AEARCTR-0008011. https://doi.org/10.1257/rct.8011-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Itzhak Rasooly*

* Primary investigator

Registry ID
AEARCTR-0008011
DOI
https://doi.org/10.1257/rct.8011-1.1
First registered
Last updated
Status
Completed
Trial dates
2021-08-06 – 2021-08-09
Countries
United Kingdom of Great Britain and Northern Ireland ()
Registry record
http://www.socialscienceregistry.org/trials/8011

Abstract

In this paper, I design and implement a series of tests of the level-k model of auctions. I begin by studying which (simple) environments can best disentangle the level-k model from its leading rival, Bayes-Nash equilibrium. I then implement these environments in the (virtual) laboratory to see which theory, if either, can explain the observed bidding behaviour.

Experimental design

The experiment proceeds as follows:

1. Subjects provide some basic demographic information in an introductory survey.
2. Subjects take part in a first-price auction with "cancelled" bids. Before the auction, subjects must successfully complete a quiz that verifies that they understand the auction rules.
3. Subjects take part in an all-pay auction (again, they are first quizzed on their understanding of the rules).
4. Subjects play Alaoui and Penta's version of the 11-20 game.
5. Finally, subjects complete Crosetto and Filippin's "bomb risk elicitation task".

There are 2 treatments. In one treatment (to which 2/3 of subjects are randomly allocated), subject bids are integers. In the other treatment, bids are multiples of 5.

Design details

Randomization method
By computer
Randomization unit
At the individual level
Clusters
I anticipate that there will be around 90 subjects (the exact number depends on the number of recruited participants who show up to the experiment).
Observations
See previous
Arms
See previous
Primary outcomes
The bids submitted by participants

Cited by

No citation data available for this registration yet.

Keywords