Hidden Discrimination: The Importance of Identifying and Addressing all Forms of Discrimination

By Kai Barron, Ruth Ditlmann, Stefan Gehrig, and Sebastian Schweighofer-Kodritsch

Photo Credits: David Aussenhofer and WZB, Berlin

Labor market discrimination is a critically important policy issue around the world. When one individual receives preferential treatment over another on the basis of gender or ethnicity, this often violates basic ethical principles. Moreover, such discrimination predominantly harms socio-economically weaker groups, thereby reinforcing inequality. For this reason, discrimination has received substantial attention from academics in several disciplines, including economics, psychology and sociology. However, a new wave of research is asking whether the standard taxonomy used for classifying and understanding discrimination within the economics literature is too narrow and, therefore, neglects to identify some forms of discrimination.

Traditionally, the economics literature distinguishes between discrimination based on taste and discrimination resulting from beliefs that are accurate in a statistical sense. For example, if a male employer preferentially hires men because he intrinsically prefers interacting with men rather than women, this is classified as taste-based discrimination. If the employer instead preferentially hires men because on average men tend to actually be more productive in the job in question than women, then this constitutes statistical discrimination.

Recent work suggests that this taxonomy may miss several important aspects of discrimination. In particular, it omits discrimination emanating from statistically inaccurate beliefs due, for example, to widely held inaccurate stereotypes.

In a new contribution to this discussion, BCCP Fellows Kai Barron and Sebastian Schweighofer-Kodritsch and their co-authors Ruth Ditlmann and Stefan Gehrig study gender discrimination across a range of experimentally controlled hiring settings that vary in the degree to which employers’ decisions reveal discrimination. This is done by presenting employers with choices between job candidates, where the information that employers observe about these job candidates is carefully controlled: employers observe information about the gender and qualifications of the candidates. The authors consider various scenarios, including those in which the two candidates are (i) clearly ranked by qualification, (ii) equally qualified, and (ii) differently qualified (holding different qualifications). The data reveals evidence of both explicit (i.e., “obvious”) and implicit (i.e., “hidden”) discrimination against women. Neither of these forms of discrimination is justified by true performance differences between male and female job candidates, but they are consistent with prevailing gender stereotypes. The two forms of discrimination differ in how clearly they reveal the employer’s bias, with some individuals only discriminating when their discrimination is obscured by the choice setting. In the study, some employers are willing to discriminate even when job candidates are equally qualified (explicit discrimination), while others only discriminate when the candidates are differently qualified and, therefore, are not easily ranked (implicit discrimination).  The analysis highlights the central role played by the contextual features of the hiring setting in conjunction with prevailing stereotypes in determining whether and how discrimination will manifest. The authors also provide several suggestions of how these findings may inform policy to effectively combat discrimination. One example is for decision makers (e.g. employers) to commit ex ante to a clear definition of how candidates’ relative merits on the relevant criteria will be traded off in the overall assessment before receiving information about the candidates.

The full paper “Explicit and Implicit Belief-Based Gender Discrimination: A Hiring Experiment” is available as WZB Discussion Paper No. SP II 2020-306.

This text is jointly published by BCCP News and BSE Insights.