Real-World Negotiations

by Charles Hofacker


Two new large datasets on real-world negotiations, freely available for research


Author: Brad Larsen

Two new large datasets on real-world negotiations, freely available for research

During this time when many experimental studies and other data collection activities have had to be put on hold, we would like to make you aware of two new, large data resources that we have made publicly available. Both datasets contain microdata from price negotiations in the field.

The first dataset comes from Larsen (2020), “The Efficiency of Real-World Bargaining: Evidence from Wholesale Used-Auto Auctions.” It contains over 250,000 observations of a secret reserve price auction followed by sequential bargaining from business-to-business transactions in the wholesale used-car industry. The dataset spans sales from 2007 through part of 2010 for several large auction houses. Each observation contains details on car characteristics, the auction, the secret reserve price, and all actions taken during the price negotiations (including all offers). The dataset is ideal for studying car markets broadly or for studying auction or negotiations. Data and details are available at http://www.nber.org/data/used-car-bargaining.

The second dataset comes from Backus, Blake, Larsen, and Tadelis (2020), “Sequential Bargaining in the Field: Evidence from Millions of Online Bargaining Interactions.” It contains over 25 million price negotiation sequences from eBay’s Best Offer platform. Each observation contains information on all negotiation actions (including all offers); anonymized identifiers for the buyer, seller, and product; and some information about buyers’ and sellers’ characteristics. The dataset spans sales from mid 2012 to mid 2013, and is ideal for studying e-commerce broadly or for studying negotiations. Data and details are available at http://www.nber.org/data/bargaining.

Both datasets contain a large number of variables describing individual differences across bargaining interactions. The datasets are rich enough to support machine learning analysis or more traditional statistical analysis.

Please distribute to any researchers or students who might be interested. Do not hesitate to contact us with questions.

Matt Backus, Columbia, matthew.backus@columbia.edu
Tom Blake, eBay, tomblake@gmail.com
Brad Larsen, Stanford, bjlarsen@stanford.edu
Steve Tadelis, UC Berkeley, stadelis@berkeley.edu