On Regularity and Normalization in Sequential Screening

TL;DR

This paper critically examines the regularity assumptions in a multi-agent sequential screening model, revealing their limitations and implications for valuation distributions and agent signals.

Contribution

It highlights the non-invariance of regularity assumptions under relabeling and their restrictions on valuation distributions, providing new insights into model assumptions.

Findings

Regularity assumptions are not relabeling invariant.

They exclude valuation distributions with bounded support.

Assumptions imply valuations are signals plus independent noise.

Abstract

We comment on the regularity assumptions in the multi-agent sequential screening model of Eso and Szentes (2007). First, we observe that the regularity assumptions are not invariant to relabeling each agent's signal realizations. Second, we show that the regularity assumptions rule out valuation distributions with common bounded support. Third, we show that if each signal realization is labeled to equal the expected valuation, then the regularity assumptions imply that each agent's valuation is equal to his signal realization plus independent mean-zero noise.