Amelie Wulff*, Tom Gargani**

Séminaires internes
phd seminar

Amelie Wulff*, Tom Gargani**

University of Konstanz*, AMSE**
Local Newspaper Closures and their Effect on Lending Discrimination*
How Much Is There to Distribute? Ordinal vs. Cardinal Measurement of Attributes**
Co-écrit avec
Agnes Bäker, Jan Riepe*
Nicolas Gravel**
Lieu

IBD Salle 21

Îlot Bernard du Bois - Salle 21

AMU - AMSE
5-9 boulevard Maurice Bourdet
13001 Marseille

Date(s)
Mardi 9 septembre 2025| 11:00 - 12:30
Contact(s)

Alexandre Arnout : alexandre.arnout[at]univ-amu.fr
Philippine Escudié : philippine.escudie[at]univ-amu.fr
Armand Rigotti : armand.rigotti[at]univ-amu.fr

Résumé

*Discrimination represents an important moral problem in the field of business ethics. While most of the extant literature studies discrimination against employees, we examine how changes in the local information environment influence organizational decision-making, focusing on the discriminatory behavior of frontline employees. Leveraging a staggered difference-in-differences design, we exploit the closures of local newspapers across U.S. counties between 2009 and 2022 to assess how these closures on discriminatory behavior in mortgage lending. Analyzing over 187 million loan applications, we find that the loss of local journalism is associated with significantly higher denial rates for applicants who are ethnic minorities or women, even after controlling for loan and applicant characteristics, bank fixed effects, and local economic trends. We find stronger effects in more heterogenous communities, pointing to local journalism as a unifying force in otherwise fragmented counties. Our findings contribute to the business ethics literature by documenting how local newspapers play a critical role in indirectly reducing discriminatory practices. These results highlight how persistent, community-level information exposure can shape ethical behavior in organizations.

**This paper presents an integrated axiomatic framework for measuring how much of a single attribute is distributed among a collection of agents, depending on whether the attribute is measured on a cardinal or ordinal scale. We require that comparisons of distributions be consistent with how the attribute studied is expressed - positively (e.g.utility) or negatively (e.g. disutility) - a property we term reversal consistency. We show that the only anonymous, Paretian and separable ordering satisfying the cardinal reversal consistency is the ordering defined by the mean. In the ordinal case, the only anonymous and Paretian ordering that satisfies the ordinal reversal consistency - when restricting our comparisons to distributions with a unique median - is the ordering defined by the median. We establish an impossibility theorem showing that no anonymous and Paretian ordering satisfies ordinal reversal consistency when the median can be non-unique. Finally, we explore the joint characterization of the lower and upper medians as a possible way to circumvent this impossibility.