Testing affiliation in private-values models of first-price auctions using grid distributions

TL;DR

This paper develops a tractable empirical model for first-price auctions with potentially dependent, non-affiliated valuations, and tests for affiliation using real auction data, finding no evidence against affiliation.

Contribution

It introduces a new test for affiliation in private-values models and applies it to real-world auction data, expanding empirical tools in auction theory.

Findings

No rejection of affiliation hypothesis in the Michigan road-resurfacing auctions

Provides a practical framework for testing valuation dependence in auctions

Enhances empirical analysis of auction models with dependent private values

Abstract

Within the private-values paradigm, we construct a tractable empirical model of equilibrium behavior at first-price auctions when bidders' valuations are potentially dependent, but not necessarily affiliated. We develop a test of affiliation and apply our framework to data from low-price, sealed-bid auctions held by the Department of Transportation in the State of Michigan to procure road-resurfacing services: we do not reject the hypothesis of affiliation in cost signals.