A lab-in-the-field experiment on Auction Design
How to cite
Dillon, Andrew; Tomaselli, Nicolo. 2024. "A lab-in-the-field experiment on Auction Design." AEA RCT Registry. AEARCTR-0013641. https://doi.org/10.1257/rct.13641-1.2
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
We study auction design in a low-income country where licenses to participate in a novel marketing system are auctioned among agricultural input dealers. We use auctions as a price discovery mechanism, and we measure bidders’ willingness to pay (WTP) to purchase licenses to take advantage of new business opportunities.
Intervention
We have designed a field lab experiment with agricultural input dealers in Burkina Faso. In this country, adoption of agricultural technologies and innovations is low due to both supply and demand constraints. We use auction design to study the case of private-sector agricultural input dealers and how they price a market-based innovation, the Village Input Fair, that can increase their market share and, consequently, the uptake of agricultural inputs.
Auctions are used to determine prices (Berry et al., 2020), and our experiment consists of 5 rounds of auctions in which 60 input dealers bid to obtain licenses and sell agricultural inputs in rural villages during a Village Input Fair, or VIF — an event that allows input dealers to meet potential customers (Dillon and Tomaselli, 2023). We are particularly interested in understanding bidder responses to variations in reserve price rules and license grouping during the auction.
Experimental design
75 ag-input dealers will be invited to participate in the experiment and will be split into 3 groups. On each day, 25 ag-input dealers will participate in the experiment. Each ag-input dealer will participate in five rounds of bidding. Of the five rounds, one round is the control arm for the experiment, and the other four rounds are treatments. The arms as as follows:
In the control arm, no reserve price is set. In Treatment 1, the reserve price for each of the two licenses is set to half the marginal cost of organizing the fair. In Treatment 2, the reserve price is set to λ times the expected license revenues, where λ is the cost/revenue ratio of the whole license portfolio. In Treatment 3, the reserve price for each of the two licenses is set to half the marginal cost of organizing the fairs, and three fairs are auctioned together based on their geographical position. In Treatment 4, no reserve price is set, and three fairs are auctioned together based on their geographical position.
The lab in the field will generate data with the four treatment conditions described above plus the control. Since bidders will bid for licenses that, by construction, are statistically similar across treatment conditions, we expect to be able to neatly compare the moments of the bids with the different reserve prices and the with/without grouping.
Design details
Cited by
No citation data available for this registration yet.