In the last few weeks, this blog has covered a variety of topics regarding the gender pay gap. I have shown evidence that women are working more than in the past. And, that they’re more educated too. In fact, women are now more educated than men (my wife is giving me a look of knowing superiority). However, despite these facts, women still make only 80 percent of what men make (now she’s giving me the finger!). Last week, we noted that one likely culprit is caregiving. But another culprit also plays a role: gender discrimination.
What is Discrimination?
When economists think of discrimination, we actually have two different definitions in mind: prejudicial discrimination and statistical discrimination.
The first kind of discrimination is called “prejudicial discrimination.” Prejudicial discrimination happens when an employer pays someone less or doesn’t hire them out of disdain, not due to performance. An employer who thinks all women are less productive than men is being prejudiced…and a jerk. These employers will typically hire fewer women or only hire them at low wages.
The only good news about prejudicial discrimination is that it should vanish over time. And not because people naturally become more progressive. Prejudicial discrimination should vanish because it’s just not profitable. After all, if women are getting paid less than men because some employers discriminate, then they are a bargain. Women would be underpaid but overperforming. When a non-discriminating employer realizes this fact, then they will hire women and become more profitable than competing, discriminating employers. Eventually, the discriminating employers will go bye-bye.
Another kind of discrimination is called “statistical discrimination.” It occurs when an employer applies something it thinks about a group of people to an individual. Where someone attended college is a common example. Imagine two interviewees with similar resumes and who interview equally well. If one went to Harvard, then the employer may hire that one over the one from a less prestigious school. It’s not that the employer doesn’t like the non-Harvard interviewee, but they are still statistically discriminating. The employer may even be basing their belief on fact. In their experience, employees hired from Harvard may actually perform better on average. But, the employer could be very wrong about the two people sitting in front of her. The better worker may come from the less prestigious school.
When we say it with schools, statistical discrimination almost sounds normal. Isn’t this how the world works? Maybe. But applying the idea to gender will get uncomfortable fast. What if an employer wants to hire someone she is sure will work continuously for the next 20 years? We saw last week that women are in fact more likely than men to take time off when young children are present. So, a 25-year old man is more likely to be around continuously for 20 years than a 25-year old woman. An employer with two equal candidates sitting in front of her may discriminate against the woman, even if this particular woman has no plans to take off. It’s not that the employer doesn’t like women — but a characteristic of the group drives their decision. That’s discrimination.
Here’s the problem — statistical discrimination can persist even in a competitive market. If the employer is acting on true characteristics of a group, then on average the employer will find the information useful. The information is also discriminatory. If women are statistically discriminated against, then they will make less and get hired less in certain jobs than men. Statistical discrimination sounds less mean than prejudice, but’s it is every bit as damaging. Worse, it’s more persistent.
Does Gender Discrimination Still Happen?
Identifying gender discrimination in the real world is difficult. After all, both prejudicial and statistical discrimination occur when one person is treated differently than another otherwise identical person. But, proving two people are identical is very difficult. Even if two people appear identical in data, they may not be. For example, two people with similar education and experience may have different interview styles. Since most data do not contain “soft” info like interview styles, researchers have to take creative approaches.
I want to highlight one such study to show you that, yes, women likely still face discrimination in labor markets. (I imagine the women reading are like…duh!). The study was conducted by a group of researchers led by Shelly Correll in 2007, and used a combination of two approaches.
The first approach involved a laboratory experiment with students. The students were shown fictional resumes of individuals with stereotypically male (e.g, Matthew) or female (e.g., Sarah) names. Additionally, some of the resumes indicated that the individual was a parent by listing “Parent-Teacher Association Coordinator” under the “Other Relevant Activities” header. The resumes were otherwise designed to be of identical quality.
The students were then told that the resumes were real and that the employer wanted some outside help deciding an annual salary for the candidates. The results suggested that mothers paid a significant penalty in suggested salary. The suggested salary of $137,000 salary for mothers was about 10 percent lower than fathers at $150,000. Non-mothers seemed to do just fine.
Figure 1. Suggested Salaries for Resumes of Women and Men, by Parenthood Status
Of course, a laboratory with students is very different than the real world. So, the second approach in the study sent the resume to real employers. The researchers examined whether or not the candidate got contacted for an interview. Since the candidates were fictitious, the researchers created e-mails and phone numbers for them and then recorded which “candidates” were contacted by an employer. Sneaky. Once again, mothers got the short end of the stick. They got fewer callbacks than non-mothers and fathers.
Figure 2. Share of Women and Men Getting Callbacks, by Parenthood Status
Both these results are consistent with a story of statistical discrimination. Why statistical? Because it’s not like the students or employers targeted all women for lower pay or fewer interviews…just mothers. Perhaps the perception is that mothers are more likely to take time off. Or, that they will need flexible hours. In any case, the end result is the same. Regardless of their actual performance or likelihood of taking time off, a large group of women will end up getting paid less and having a harder time finding a job.
The last two week’s posts tell a pretty consistent story. Basically, caregiving seems to be a triple-whammy as far as wage equality for women. First, caregiving directly costs women money by requiring them to take flexible jobs. Second, it costs them money in the long run through lost experience. Finally, people seem to practice gender discrimination against women because of it. Both students in a lab and employers in the field seemed to be practicing gender discrimination against mothers specifically. Next week, I want to close out our discussion of the gender wage gap and discuss some solutions. But, until then, know that caregiving responsibilities — and the discrimination that seems to come with them — are a major impediment to wage equality.