So far in my short series of columns on racial discrimination in the labor market, we have discussed that studies show that discrimination exists and is significant, and that discrimination based on any non-market characteristic, regardless of the reason for that discrimination, leads to inefficiency.
Today, I want to focus more on the phrase “regardless of the reason for that discrimination.” This is the phrase that has spurred the most critical feedback from readers. A couple of readers have emailed me to explain why their particular reason for discrimination in their hiring is not grounded in a prejudicial attitude but in data and therefore makes good business sense, relieving inefficiencies rather than causing them.
What you guys described in your emails is a special type of discrimination called statistical discrimination. Labor economist George Borjas defines statistical discrimination as “treating an individual on the basis of membership in a group and knowledge of that group’s history.”
For example, on average, Black individuals have less formal education than White individuals. Thus, statistically speaking, if I cannot verify applicants’ level of education, hiring a White worker over a Black worker makes sense, as it is more likely I am getting an educated worker that way. It’s not that I prefer White workers over Black workers; it’s that I prefer educated workers over uneducated workers. If the education statistics were different, my hiring decision would be different.
It is easy to convince ourselves through this type of reasoning that statistical discrimination is ethically acceptable and that it makes “good business sense.”
Why is that not true?
First, a little more statistics. Knowing the average of a group of values tells one very little about any of the individual values. The average tells us nothing about the distribution of values — is everyone clumped somewhere near the average, or are they clumped on two ends far from the average, or are they spread somewhat evenly across the spectrum? The average of 5 and 5 is the same as the average of 0 and 10 or of 4 and 6.
Basing a hiring decision on average performance of a group means you’ll be wrong about half the time. You will falsely assume a highly productive applicant is less productive or vice versa.
This is ethically problematic for a couple of reasons:
1. You are judging an individual based on an average, not on his own capabilities. Imagine yourself being judged based on the average person who looks like you. For some of you, that may be to your benefit. For most of you, I imagine you would prefer to be evaluated on your own skills, not those of others like you.
2. Statistical discrimination often occurs based on the hirer’s perception of an average, rather than on a statistical average. This is the situation described in emails I have received. The discrimination described was not based on something easily measurable, like education, but instead was based on experience or feelings of one race’s likelihood of showing up, working hard, etc. This is especially troubling, as these feelings often are stereotypes, not averages, and they are even worse indicators of an individual’s productivity than an actual, statistical average might be.
No matter how you slice it or how you justify it, racial discrimination, even based on real or perceived statistics, does not make good business sense.
P.S. Yes, I, personally, have been involved in many hiring decisions both as a manager in a fast food establishment and as a member of several hiring committees in my current position. The temptation to resort to statistical discrimination is real, but it is right and far more rewarding to do the hard work of learning about applicants individually, not simply as a member of a group, before making a hiring decision.