Focal Points in Experimental All-pay and Winner-pay Pooled Auctions with Similar Items

How to cite

Hsu, Dexter; Wang, Joseph Tao-yi. 2024. "Focal Points in Experimental All-pay and Winner-pay Pooled Auctions with Similar Items." AEA RCT Registry. AEARCTR-0013079. https://doi.org/10.1257/rct.13079-1.0

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

Dexter Hsu* , Joseph Tao-yi Wang

* Primary investigator

Registry ID
AEARCTR-0013079
DOI
https://doi.org/10.1257/rct.13079-1.0
First registered
Last updated
Status
In development
Trial dates
2024-03-01 – 2025-03-01
Countries
Taiwan, Province of China (Taipei)
JEL codes
C92, D44, I22, I28
Registry record
http://www.socialscienceregistry.org/trials/13079

Abstract

We plan to compare first-price and all-pay auctions with two units sold, and observe the effect of focal points on bidding, revenue and efficiency. Specifically, we plan to conduct laboratory experiments, in which each session has 6 bidders drawing private values independently (for each period) and competing to win one of two prizes, with the second prize valued at a fraction (80%, 40%, 20% etc.) of the first prize. The risk neutral Bayesian Nash equilibrium (RNBNE) predicts revenue equivalence and both formats being efficient. However, experimental results would likely be affected by the fraction (80%, 40%, 20% etc.), becoming bidding focal points. If this occurs, the first-price auction would yield higher revenue compared to the all-pay auction, and have efficiency levels closer to 100%. In contrast, the single-unit, first-price auction does not have such focal points, and would likely result in lower (first-prize) revenue and efficiency than all-pay, since bidders tend to overbid in all-pay auctions.

Experimental design

We adopt a focal point set Λ = {0, 0.2, 0.4, 0.8, 1}, creating a total of 5 experimental treatments.
For each treatment, we recruit 36 subjects, randomly distributed across 6 experimental groups.
Among the six groups, three groups experience all-pay auctions first (“All-Pay × Winner-Pay”),
and the other three experience winner-pay auctions first (“Winner-Pay × All-Pay”). This design
allows us to investigate how the magnitude of λ affects participants’ bidding behavior in winner-
pay and all-pay auctions. Additionally, it facilitates an exploration of the potential order effect.

Design details

Randomization method
Subjects are recruited by ORSEE. Group assignment randomized one the fly using experimental software zTree. Experimental parameters drawn randomly prior to the experiment using zTree.
Randomization unit
individual
Clusters
180 Subjects
Observations
10,800 bids, 1800 auctions
Arms
36 subjects in each of the five treatments
Primary outcomes
Subjects' bids, seller revenue, and efficiency

Cited by

No citation data available for this registration yet.

Keywords