VoxEU Column Labour Markets

Unexplored dimensions of discrimination in Italy: Sexual orientation and physical appearance

Discrimination on the basis of sexual orientation is perceived as widespread in the EU. This column provides the evidence. Fake CVs, some explicitly gay, were sent to job ads in Rome and Milan. Gay men were found to have a 30% lower chance of being called back. And while beauty increased callback rates for women, being overtly lesbian had no effect.

Discrimination in the labour market is maybe one of the most studied topics in economics but certainly the least understood (Tavares et al. 2007, Booth and Leigh 2006). Indeed, when explaining the adverse labour-market outcomes of certain categories of workers (for example, ethnic minorities or women), it is very difficult to disentangle between discrimination and other often-unobserved aspects such as low ability or social norms and peer effects.

Recently, researchers have tackled this issue using new methods such as field studies and natural experiments. Many of these field experiments have focused on females and immigrants. The results convincingly show that there is indeed discrimination against minority workers and women.

There exist, however, other types of discrimination that have rarely been studied by economists. Using data from the Eurobarometer 2008, Figure 1 gives the perception of discrimination in EU27 countries. The graph reveals, for example, that discrimination on the basis of sexual orientation is perceived as widespread by more than 50% of the people, whereas this percentage is only slightly higher than 35% for gender. Yet field studies exploring labour-market discrimination based on sexual orientation are rare, and non-existent in Europe.

Another kind of discrimination that also has been neglected by economists is the one based on physical appearance. The empirical literature has mostly analysed the correlation between beauty and labour-market outcomes (for a review of this literature see Hamermesh 2011) finding a strong association between good looks and labour-market outcomes. However, evidence demonstrating a causal relationship is scarce.

Figure 1. Perception of discrimination in the EU

Note: The figure gives the percentage of people answering that discrimination on the basis of the listed source is 'widespread'. The percentages of people declaring that it is “Rare”, “Do not know” and with missing values are not displayed in the figure. Source: Eurobarometer 2008, EU27.

We propose a field experiment aiming at shedding light on the causal relationship between sexual orientation and physical appearance and labour-market outcomes.

Sending fictitious résumés

For that, we study the effect of perceived homosexuality and of physical appearance in the labour market by sending fictitious résumés to real help-wanted ads in Rome and Milan. These are the two Italian cities with the largest labour markets. Also, most of the help-wanted ads in websites commonly used to advertise job vacancies are for jobs in Milan and in Rome.

We experimentally manipulate 'perceived homosexuality' by randomly modifying résumés by adding items that reveal sexual preferences. At the same time, we randomly attach to the résumés a picture of the candidate where these pictures have been previously ranked in terms of beauty. We restrict the field experiment to seven occupations, the most frequent ones that do not require specific skills, i.e. administrative clerk, bookkeeper, call-centre operator, receptionist, sales clerk, secretary and shop assistant.

The résumé items revealing homosexual preferences were periods of internship in pro-gay advocacy groups that are real, well-known by the public at large, and city-specific. The names of these groups were very explicit about the nature of the activity, for example Arcilesbica Roma, Centro di Iniziativa Gay-Arcigay and Di Gay Project. Applicants in the control group have instead worked as interns for a period of similar length in a non-gay cultural association or in a company. In order to better match the occupation, the tasks performed during the internship were different across applicants. For each city and for each occupation, one of the three types of internship was associated with two identities, one male and one female.

For physical appearance, we randomly assign to each résumé a picture chosen among 89 previously collected photos of individuals aged between 20 and 35 years old. The photo was chosen in such a way that it minimised differences between the age declared in the résumé and the real age of the person depicted.

The experiment started on 17 January 2012 and ended on 21 February 2012. During this period, for each city and occupation, we selected the most recent employment ads published in two websites, Job Rapido and Monster. They are the most popular websites among actual jobseekers. We answered to 531 ads, 336 in Milan and 195 in Rome. We typically sent four résumés in response to each ad, two from the treatment group and two from the control group. We sent 2,320 résumés in total.

The overall response rate was about 11%, with a minor difference between males and females (10.83% and 11.24%, respectively). Looking at the percentages by city, the response rate was higher in Rome (about 16%), where men were more likely to be called back than women (17.48% versus 14.96%). On the contrary, in Milan, the overall response was roughly divided by two (about 8%) and men were less likely to be called back than women (7.19% versus 9.10%).

The 'call-centre operator' type of job was the occupation which received the highest rate of callbacks, followed by 'receptionist' and 'sales clerk'. Interestingly, people who sent their CVs to a secretary or a shop assistant job got a very low callback rate, i.e. 4.66% and 2.95% respectively.

We find that there is a statistically significant penalty (in terms of callback rates) associated to gay men of about 3% whereas gay women do not seem to show a significant difference in callback rate with respect to straight women. The 3% penalty for gay men is quite high since the callback rate for males is 10%, which means that compared to gay men, they have 30% less chance to be called back.

We also investigate whether the penalty associated with gay people is mitigated for high-skilled individuals. Interestingly, we find the opposite result. The penalty is actually higher for high-skilled gay people, with an associated magnitude of more than 8% for gay men. No penalty or premium is instead associated to high-skilled lesbians, confirming that only men are penalised in the labour market for their sexual orientation.

When we instead look at differences in response rates by picture beauty, our analysis indicates a significant premium for attractive women of about 2% and no significance difference between handsome and ugly men. We then investigate whether the beauty premium for women varies by skills. We find that high-skilled attractive women are called back less often than low-skilled attractive women. This may indicate that beauty might not be an advantage for high-skilled women.

Lesbian stereotypes and beauty premiums

Different economic arguments can be considered to explain our results. Standard theories of discrimination (taste-based and statistical discrimination) can explain our results. The statistical discrimination model is typically used to make predictions about lesbians. Stereotypes about lesbians, for example that they are more focused on their career, that they are less likely to have children or that they are more masculine, are considered to be an important source of bias. In the statistical discrimination framework, lesbians therefore are predicted to do better than their straight counterparts while gay men should do worse. We find that this is true for gay men but not for lesbians.

We can also apply the standard economic theories of discrimination to explain the negative impact on the labour-market outcomes of less attractive individuals. Our finding may suggest the existence of discrimination against the less attractive workers, especially women. The fact that high-skilled pretty women obtain less beauty premium than low-skilled pretty women may indicate the fear of competition with these women for certain types of jobs. Importantly, we show that the beauty results are particularly relevant for occupations requiring the interaction with customers as secretaries, receptionists and general customer service.


Booth, Alison and Andrew Leigh (2010), “Do employers discriminate by gender in female-dominated occupations? Results from a field experiment”, VoxEU.org, 2 February

Hamermesh, Daniel S. (2011), Beauty Pays. Why Attractive People are more Successful Princeton: Princeton University Press.

Patacchini, Eleonora , Ragusa, Giuseppe and Yves Zenou (2012), “Unexplored Dimensions of Discrimination in Europe: Homosexuality and Physical Appearance”, CEPR Discussion Paper 9179.

Tavares, José A. Tiago V. de V. Cavalcanti (2007), "Gender discrimination lowers output per capita (a lot)", VoxEU.org, 16 October

2,520 Reads