When correspondence studies fail to detect hiring discrimination
When correspondence studies fail to detect hiring discrimination: Evidence from the French public and private sectors
Pierre Cahuc, Stéphane Carcillo, Andreea Minea, Marie-Anne Valfort 03 January 2020
Correspondence studies are often used to detect discrimination on the part of recruiters, but they do not inform us about decisions at the interview and job-offer stages. This column describes how a correspondence study suggests that individuals of North African origin are strongly discriminated against in the French private sector, while they are treated equally in the public sector. However, survey results indicate that both public and private sector employers express discriminatory preferences and beliefs against North African candidates, and wages of young unskilled North African males are lower than those of their French compatriots in both sectors. The findings suggest that correspondence studies should be complemented with other investigation methods.
Correspondence studies are widely used in economics, sociology, and personnel psychology to detect hiring discrimination in the labour market (Baert 2018, Bertrand and Duflo 2017, Neumark 2018). In these studies, researchers send fictitious job applications, differing only in a minority trait, to real job offers. A lower callback rate from employers to the applicants with the minority trait is interpreted as proof of unequal treatment. An important limitation of this methodology is that it provides limited information about hiring decisions, as it does not inform us about what is going on at the interview stage and the job-offer stage.
Our contribution (Cahuc et al. 2019) exhibits a situation in which discrimination at the invitation stage is a poor predictor of discrimination at the hiring stage, which suggests that many correspondence studies may fail to detect discrimination and should be complemented by additional investigation methods.
To show this, we focus on the chances of a callback for low-educated candidates of French versus North African origin who apply to the private or the public sector. These populations have been selected because second-generation immigrants of North African origin face strong labour market discrimination. The choice of public and private sectors is motivated by potentially large differences in recruitment behaviour. In contrast to private sector recruiters, public sector recruiters are generally weakly constrained by profitability requirements.
Our correspondence study shows that individuals of North African origin are strongly discriminated against in the private sector, while they are treated equally in the public sector. The average callback rate of our applicants is low – for 100 applications sent, they receive an average of 4.5 callbacks in the public sector and 3.3 callbacks in the private sector. Being of North African origin leads to a significant drop in the callback rate when individuals apply to private sector vacancies (by about 2 points) but no penalty is associated with ethnicity in the public sector, and this finding is robust to the introduction of a variety of controls related to the type of position applied for, employers’ size, and economic activity. In particular, individuals of French and North African origin are equally treated in both central and local government vacancies, despite the fact that recruitment procedures may differ across these occupations.
To understand what drives this discrimination gap between the two sectors, we run a survey on a sample of public and private sector recruiters, representative of those to which our fictitious applications were sent. The survey results displayed in Figure 1 indicate that both public and private sector employers express marked discriminatory preferences and beliefs against North African candidates, with only small, non-statistically significant differences between the two sectors. Previous studies have shown that North Africans are as under-represented among public sector hires as they are among private sector ones, especially when it comes to low-educated youth (Fougère and Pouget 2004, Berson 2016). Additionally, we find that the wage of young unskilled North African males is lower than that of their French compatriots in the private and public sectors. These findings are at odds with the absence of differences in callback rates observed at the invitation stage in the public sector between French and North African candidates.
Figure 1 Discriminatory beliefs of recruiters in the private and public sectors
Note: This figure reports the average value of an index of discriminatory beliefs by sector.
To reconcile these empirical findings, we set out a model illustrating a situation in which the absence of invitation discrimination does not automatically result in the absence of hiring discrimination. In this model, recruiters interview workers to get information about their productivity. Interviews are costly. For each hire, recruiters first choose the number of interviewed applicants and then hire the best one. Labour costs are lower in the public sector insofar as job creation is generally subsidised or determined by a budget allocated to a specific job, which is not paid by the recruiter and which cannot be used for any other purpose if the job is not created. In this framework, public sector employers value job creation more than private employers for whom the value of marginal jobs is driven downwards by competition. Accordingly, public employers are inclined to interview more applicants – notably more minority candidates – because the probability of filling jobs increases with the number of interviews. However, following the interviews, only the best applicant is hired. Insofar as interviews do not allow employers to extract all the necessary information on applicants’ productivity, the final selection can be influenced by statistical discrimination.
We show that a sector in which the expected profits from filled jobs are higher displays lower differences in callback rates between applicants of different origin, but can also display stronger hiring discrimination among applicants invited for interviews. This case is precisely the one uncovered by our survey of recruiters, which indicates that employers in the public sector value job creation more than private sector employers. Therefore, the absence of invitation discrimination in the public sector, which likely flows from the high valuation of job creation by recruiters in that sector, is compatible with significant hiring discrimination.
To gauge the empirical relevance of this mechanism, we estimate the model relying on data from the French Labor Force Survey. We find that hiring discrimination against young North African high school dropouts, conditional on being invited for interview, is stronger in the public than in the private sector for relevant empirical parameter values. More generally, we show that depending on the labour market tightness and on the bargaining power of workers, discrimination at the hiring stage can be either stronger or weaker in the public sector than in the private sector.
Overall, these results cast doubt on the ability of correspondence studies to detect hiring discrimination in any circumstance. They suggest that correspondence studies should be complemented with other investigation methods, including when these studies focus only on private sector firms, to the extent that empirical studies in the US and Europe typically find large total factor productivity differences across firms, which translate into differences in expected profits from job creation.
Baert, S (2018), “Hiring Discrimination: An Overview of (Almost) All Correspondence Experiments Since 2005”, in S M Gaddis (ed.), Audit Studies Behind the Scenes with Theory, Method, and Nuance, Springer, 63-77.
Berson, C (2016), “Private vs Public Sector Wage Gap: Does Origin Matter?”, Metroeconomica 67(4): 717-741.
Bertrand, M, and E Duflo (2017), “Field experiments on discrimination”, in A Banerjee and E Duflo (eds), Handbook of Field Experiments, North Holland.
Cahuc, P, S Carcillo, A Minea, and M-A Valfore (2019), “When Correspondence Studies Fail to Detect Hiring Discrimination”, CEPR Discussion Paper 14028.
Fougère, D, and J Pouget (2004), “L’emploi public s’est-il diversifié ? Sexe, niveau d’études, origine sociale et origine nationale des salariés de la fonction publique et des collectivités territoriales”, Complément au Rapport sur la diversité de la function publique remis par Madame Dominique Versini au Ministre de la Fonction Publique et de la réforme de l’Etat, Paris. La Documentation Française.
Neumark, D, (2018), “Experimental Research on Labor Market Discrimination”, Journal of Economic Literature 56(3): 799-866.