PhD Defence: Timm Behler "Telling Talent: Essays on Discrimination and Promotion Contests"

Image

Opponent:
Professor Markus Dertwinkel-Kalt, Department of Economics, University of Miinster, Germany

Grading committee:
Professor Johan Stennek, Department of Economics, University of Gothenburg, Sweden
Dr. Anja Prummer, Department of Economics, Johannes Kepler University Linz, Austria
Docent Erik Mohlin, Department of Economics, Lund University, Sweden

Chair:
Docent Dawei Fang, Department of Economics, University of Gothenburg, Sweden

Abstract

Salience-Based Stereotyping

I propose a model in which an agent from a socially dominant group (e.g., an ethnic majority) learns about the average productivities of a socially dominated group (e.g., an ethnic minority) and his own group. Productivity is determined by a random intrinsic productivity parameter and education. The agent updates his beliefs over- or underweighting the importance of education based on how salient it appears to him among a given group. The model can account for several stylized facts: i) The agent will hold persistently negative beliefs about the dominated group; ii) he will be subject to in-group bias, exhibiting overly positive beliefs about his own group; and iii) adding a new dominated group can improve the agent's view of the old dominated group. Additionally, it predicts that iv) stereotyping is particularly extreme if the agent learns mostly from "tokens"---in which case, perversely, the agent's beliefs become more negative as his sample mean increases; v) affirmative action can either hurt or benefit the dominated group; and that vi), when there are two dominated groups and affirmative action is only aimed at one of them, the other is always hurt by it. Lastly, the model provides a novel connection between taste-based, accurate statistical, and inaccurate statistical discrimination, whereby taste-based or accurate statistical discrimination in education lead to inaccurate beliefs, which then lead to inaccurate statistical discrimination.
 

Meritocracy in Hierarchical Organizations

Competitive promotion mechanisms are often regarded as meritocratic because they select talented players with a high probability. We show that when incorporating promotion-contests into hierarchical organizations, this is not necessarily the case. If the organization’s middle tier can influence the career advancement of players below them, they may block the promotion of talented players out of fear that it could negatively affect their own career prospects. Uncompetitive seniority-based promotion schemes can alleviate this problem. Pessimistically, however, we identify a novel trade-off: the promotion scheme that maximizes expected ability at the middle manager level may not maximize the expected ability at the top tiers of the hierarchy. Whether this trade-off occurs in equilibrium depends crucially on (1) the average ability at the bottom of the hierarchy and (2) how well middle managers can infer the ability of their subordinates.
 

Promotions Under Excessive Workplace Surveillance

When designing promotion contests, managers might not only be interested in providing effort incentives, but also in selecting the employee with the highest ability. In this paper, I show that noisy contests can perform better regarding this objective than noiseless contests. While noise generally distorts the correlation between output and ability, noiseless contests induce mixed strategies which can distort this correlation even more. I further show that the uniform noise level that is optimal in selecting the employee with the highest ability coincides with the noise level that maximizes (expected) aggregate effort. My main interpretation is that excessive workplace surveillance can backfire (i) by hampering meritocratic selection, and (ii) by reducing productivity.
 

The full thesis: Telling Talent: Essays on Discrimination and Promotion Contests