Stereotypes, heuristics and inequality


Dlaczego często zdarza się tak, że kobiety są niżej wynagradzane niż mężczyźni, nawet wtedy, gdy zajmują równoważne stanowiska i posiadają takie samo wykształcenie? W grę wchodzą mniej lub bardziej uświadamiane mechanizmy uproszczeń i stereotypów oraz  uogólnień. Decydując o zatrudnieniu Anny i Jana innego zachowania się po nich spodziewamy przy negocjacji stawek płacowych, a potem inaczej wynagradzamy ich pracę. Dlaczego, bo zakładamy, że kobiety są mniej skłonne negocjować, podejmować ryzykowne decyzje a za to z większym prawdopodobieństwem zostaną w domu przynajmnniej na jakiś czas wychowywać dzieci. I nie ma znaczenia, że Anna wcale dzieci nie planuje i lubi dreszczyk emocji z twardych negocjacji. Jakie skutki te i inne uproszczenia mają dla całego rynku pracy?

Why so often women receive lower salary then men, even when they are equally well educated and work in the same positions? Social science research provides striking results in this context. It seems that the major role is played by more or less conscious mechanisms of simplifications and stereotypes. Our project is going to study specific aspect of stereotypes and averages in the context of labor market inequality: to what extent information on workers’ productivity affect fairness of his or her salary?

Incomplete information of the employer about the potential employee yields incentives to mitigate the risk of employing less productive worker by following averaged or simplified information about social group with which the worker is identified. One may average positively – e.g. assuming people with certain education have certain qualifications. One can also average negatively, e.g. assuming that some characteristic unrelated to productivity reveals some information about the candidate. This is the case of ethnic, racial or gender discrimination, but also bias against older workers, workers with disabilities, etc. If this averaging is in line with objective data, economic literature identifies them as statistical discrimination (Phelps, 1972, Arrow, 1973). If the averaging is not in line with the data, they stem from misperceptions (wrong information about the objective data) or taste-based discrimination (Becker, 1957). Distinguishing between the two is a challenge for a number of reasons and our project is an attempt to provide a wide battery of novel indirect and direct tests to separate statistical discrimination from taste-based discrimination in the context of wages.

We posit that subjective differencing between candidates and/or workers is instrumental in delineating statistical from taste based discrimination and thus explaining the wage dispersion. The goal of our project is to propose methodological approaches to separate pure statistical discrimination from tasted-based discrimination. In order to reach this objective, we operationalize three working hypotheses:

H1: statistical discrimination is smaller in thin labor markets

H2: statistical discrimination is smaller in jobs requiring higher extent of creativity

H3: information on productivity reduces statistical discrimination


Źródło finansowania | Financing: Narodowe Centrum Nauki, OPUS

Projekt realizowany | Timeline: 02/2020 -- 02/2023

Kierownik | Principal Investigator: Magdalena Smyk

Budżet łączny | Total budget: 388 615 zł

  • wynagrodzenia dla podstawowych wykonawców | compensation to researchers: 144 000 zł
  • stypendia dla młodych badaczy | scholarships for young scholars: 90 000 zł
  • komputery i oprogramowanie | hardware and software: 0 zł
  • konferencje i inne wyjazdy | conference travels: 32 400 zł
  • materiały, dane i usługi obce | data, usables and outsourced services: 20 600 zł
  • koszty pośrednie dla FAME | overheads for FAME: 91 962 zł

This project combines applied labor economics with carefully designed experiments . The empirical studies will exploit novel approaches in labor economics: the notion of market thickness (i.e. how big is a market for specific skills in a specific industry in terms of size and turnover) and the notion of task content of jobs (i.e. how non-codifiable is the actual sequence of actions that the worker has to do to achieve the outcomes required on a given job). These two approaches have had wide prior applications, but not in the context of labor market inequality. We also propose novel identification strategies, to address the potential endogeneity bias. The experiments will provide an innovative and carefully designed approach to testing H3. We will infer willingness to acquire information about individual productivity and responsiveness to this information.

We will develop three research papers (two separate microeconometric studies and one with an experiment), with the aim to further the frontier of research on wage inequality. Seeking sources of wage inequality that cannot be explained by individual characteristics is currently one of the most challenging fields of applied economics. The methodological novelties that we propose will enable improving the understanding of the wage inequality mechanisms and thus permit identification of effective policies. In the spirit of scientific integrity and transparency, all codes and full documentation of our research will be freely available online for replication and extension.

Wage inequality was considered to sparkle many social movements and political unrests. In most surveys, individuals declare aversion to earnings inequality and unfair treatment. Yet, in vast majority of the labor markets inequality persists. We will provide non-technical summaries of our experiments and empirical studies, explaining to general audience why this discrepancy between values and action persists and what could potentially be done about it. We will provide an overview of policy implications and share those insights with general audience via presence in traditional and social media, engaging with stakeholders and participating in public debate on this topic.